The Test Bench

Every claim we could get our hands on — famous portfolios, gurus, viral systems, timing signals, thesis ETFs, option strategies — run the honest way: real trading costs, failed companies left in, checked out-of-sample. The verdict is the colored label; the number is the evidence. Filter, sort, judge.

How to read this. The bench splits in two. What to own — assets and strategies — gets a five-grade report card: Return · Risk · Sharpe · Friction · Tax (A = great · C = middling · F = ugly), with the whole market on top as the benchmark. Signals — the overlays you’d time with — get one grade: the edge they add over buy-and-hold. Click any row for the full review, plus a second grade for how honestly we tested it.

🛡️ If your fund had used the Risk Reducer

Here's the part that should make a fund manager uncomfortable. We took the average stock fund and the famous "all-weather" portfolios, and overlaid one rule: when the Risk Reducer flags caution, step the book aside to cash; otherwise stay invested. Same funds, same holdings — the only change is sitting out the danger zones. Acting on the prior day's signal (no peeking), 2011–2026 including 2022:

fund	worst drawdown			return /yr
	as-is	+RR	cut	as-is	+RR	gain
Average stock fund (S&P 500)	-34%	-12%	cut 66%	+12.7	+10.8	-1.9
Classic 60/40 fund	-22%	-6%	cut 71%	+7.8	+6.6	-1.1
All Weather fund (Dalio)	-24%	-11%	cut 54%	+4.2	+3.9	-0.3
Permanent Portfolio fund	-18%	-8%	cut 53%	+5.7	+4.1	-1.6
Golden Butterfly fund	-20%	-6%	cut 71%	+6.7	+5.1	-1.6
HFEA leveraged book (3x)	-66%	-29%	cut 57%	+19.1	+16.1	-3.0

The Risk Reducer cut a third to two-thirds of the worst drawdown off every one — the most on the stock-heavy books that needed it. On the average stock fund it did that return-neutral; on the already-defensive portfolios it costs a touch of return, because they rarely hit the danger zone and the rule mostly benches them during recoveries. Far less pain for roughly the same money, from a single step-aside rule — what the funds charging you 1% a year didn't do. Members get the live read. (Simplified overlay: whole book to T-bills on the signal; real funds have mandates and taxes. The point is the danger zones, not the exact trade.)

What to own — the report card

Five grades, one footing. Return, worst drawdown and Sharpe come straight from the test; friction is what it costs to actually run, tax is how kindly it treats a taxable account. The whole market sits on top as the benchmark every row is graded against. Hover a column header; click any row for the full review.

Asset / strategy · sold by	Return	Risk	Sharpe	Friction	Tax
Buy the whole market — cap-weight S&P 500the benchmark every row is graded against	B+11%	D-55%	C0.62	A	A
Famous Systems, Audited · 15 tested
▸The Permanent Portfolio (Browne) Harry Browne	C+7%	B-17%	B0.96	B	B
What it's meant to do 25% each stocks / long bonds / cash / gold = stability in any economic climate — Harry Browne What we found The most genuinely crisis-proof of the lot, now proven across BOTH a deflationary crash (FLAT in 2008, while stocks fell -37%) AND an inflationary one (-12% in 2022). Best Sharpe here (0.96), a third of the market's drawdown. The price is the same as ever: ~half the long-run return of stocks. Real preservation -- just know that's what it is. The numbers +7.1%/yr, Sharpe 0.96, worst drawdown -17% (vs SPY -55%), 2008 +0%, 2022 -12% · real test, 2006-2026 (incl 2008) Performance A a real, durable edge Rigor B 4/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked portfolio_audit.py; adjusted total-return (ohlcv_long), cash earns the 3mo T-bill.
▸Golden Butterfly (Portfolio Charts) 'Tyler' / Portfolio Charts	C+8%	B-21%	B0.85	B	B
What it's meant to do 20% each US stocks / small-cap value / short bonds / long bonds / gold = better all-weather — 'Tyler' / Portfolio Charts What we found The best-built of the all-weather lot across BOTH crises -- gold and short bonds cushioned 2008 and 2022 better than the bond-heavy rivals, for the highest return of the preservation set (+7.8%/yr). Solid, dependable; still well below stocks on long-run return. The numbers +7.8%/yr, Sharpe 0.85, worst drawdown -21%, 2008 -7%, 2022 -14% · real test, 2006-2026 (incl 2008) Performance A a real, durable edge Rigor B 4/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked portfolio_audit.py; small-cap value proxied by IWM, short bonds by T-bills (ohlcv_long).
▸10-Month Moving-Average Timing (dshort / Faber) Mebane Faber / dshort	C+7%	C-26%	C0.72	B	B
What it's meant to do Hold the index above its 10-month moving average; go to cash below it -- catch the trend, dodge the bears — Mebane Faber / dshort What we found The famous trend-timing rule dshort tracks monthly -- now tested ourselves over 71 years, through 1973-74, 2000-02 and 2008. Verdict: it gives up just ~0.5%/yr for HALF the worst drawdown (-26% vs -53%) and a better Sharpe (0.72 vs 0.58) -- the bear-market insurance is real. The catch is regime: in the recent 2010-2026 decade (no slow bears, one fast 2020 crash a monthly signal can't dodge) it LAGGED buy-hold by ~4.5%/yr. Real protection, but it costs you in the long bulls. The numbers 71-yr S&P test: +7.2%/yr vs buy-hold +7.7%, worst drawdown -26% vs -53% (Sharpe 0.72 vs 0.58) · our test, S&P 500 1955-2026 Performance A a real, durable edge Rigor B 4/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked sma_timing.py; S&P 500 price index 1955-2026 (ohlcv_long) + the 2010-26 sub-window.
▸The 'In & Out' Composite Gate Popular QuantConnect strategy	—	A31%	F0.12	B	B
What it's meant to do An 8-signal composite tells you when to be in or out of the market — Popular QuantConnect strategy What we found It works — a little. Eight stitched-together signals get beaten roughly three-to-one by one humble signal (our Risk Reducer). Complexity isn't edge. The numbers Cuts drawdown (−31% vs −55%) for edge +0.12 — but beaten ~3:1 by our Risk Reducer alone · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked audit_gem_inout.py; approximate params, original lookbacks unverified.
▸The Classic 60/40 Conventional 'balanced' wisdom	C+9%	C-31%	C0.79	B	C
What it's meant to do 60% stocks / 40% bonds = the safe, balanced default — Conventional 'balanced' wisdom What we found The 'balanced/safe' default bled in BOTH crises -- -17% in 2008 and -16% in 2022 -- with a -31% max drawdown. Fine in calm decades, but it delivers far less protection than its reputation exactly when you need it, in every kind of crisis. The numbers +8.6%/yr, Sharpe 0.79, worst drawdown -31%, 2008 -17%, 2022 -16% · real test, 2006-2026 (incl 2008) Performance C marginal — barely better than a coin flip Rigor B 4/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked portfolio_audit.py; bonds = intermediate Treasuries (IEF), adjusted total-return (ohlcv_long).
▸Point & Figure Charting (X's and O's) Charting tradition / Dorsey Wright	D+7%	C-38%	D0.59	B	B
What it's meant to do The double-top buy / double-bottom sell signals time the market better than holding — Charting tradition / Dorsey Wright What we found Point & Figure is trend-following in costume. Going long its double-top buy signals and flat on the double-bottom sells, it gives up ~1%/yr of return for a meaningfully smaller drawdown (-38% vs -57%) and a hair better Sharpe -- real risk reduction. But a plain 10-month moving average does the SAME job better (it halves the drawdown; P&F only trims a third). The X's and O's add ceremony, not edge -- and bigger boxes (3%) beat small ones by whipsawing less. The numbers Long 1960-2026: +6.8%/yr vs buy-hold +7.5%, worst drawdown -38% vs -57% (3% box, Sharpe 0.59 vs 0.53) · our test, S&P 500 1960-2026 Performance C marginal — barely better than a coin flip Rigor B 4/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked pnf.py; % box + 3-box reversal on the S&P 500 price index, long/flat, cash earns 3%/yr.
▸The '91% Beat the Market' Screen A famous investing club	—	—	—	B	B
What it's meant to do Our stock screen beat the S&P 500 about 91% of the time — A famous investing club What we found Real but oversold. Strip the survivorship bias and pay real trading costs and the edge is about a third of the brochure. Still positive — just nowhere near the marketing. The numbers Claimed ~91% beat-rate; honest (net of cost, survivorship-free) ~32% · claimed vs. honest hit-rate Performance D oversold — famous, underdelivers Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked Reproduced the screen survivorship-free + net of cost; AAII-style 91% figure is gross/cost-free.
▸Global / Dual Momentum (GEM) Gary Antonacci	—	—	—	B	C
What it's meant to do 17.43% a year at 0.87 Sharpe — momentum across stocks/bonds beats buy & hold — Gary Antonacci What we found The headline number is a 40-year in-sample fit. Trade it in the real ETF era, net of cost, and it trails plain SPY. The 12-month lookback isn't even the best one — a sign of overfit. The numbers Book: 17.4%/0.87 (1973–2013 in-sample). ETF-era net 2009–26: 9.7%/0.78, TRAILS SPY 15.3%/1.07 · ETF-era net return / Sharpe Performance D oversold — famous, underdelivers Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked audit_gem_inout.py; ETF replication starts 2009 so misses 2008 (favorable to GEM, still trails).
▸The '50% Profit Management' Rule OptionAlpha-style options sellers	—	—	—	B	B
What it's meant to do Close winning option trades at 50% profit and your returns improve — OptionAlpha-style options sellers What we found The 'higher win rate' is real and totally beside the point. Closing early pays a four-legged exit spread every time, caps your winners, and keeps your losers. The win-rate lift is a mirage. The numbers Same ~93% win rate as holding, but expectancy ~$12/trade WORSE on real bid/ask · $ per trade (real fills) Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked iron_condor_audit.py on real SPY bid/ask; n=15 cycles (small — scaling).
▸The Flight to Long Bonds (TLT) Conventional 'risk-off' wisdom	—	C-34%	—	B	B
What it's meant to do When stocks fall, move to long Treasury bonds for safety — Conventional 'risk-off' wisdom What we found The reflex everyone preaches -- flee to long bonds when stocks drop -- FAILED the test. In 2022 bonds fell WITH stocks (rate shock), so switching to TLT on a stress warning DEEPENED the drawdown instead of cushioning it. Cash (-14%) and gold (-23%) protected; long bonds didn't. The numbers Switching to long bonds on the warning made drawdown WORSE: -34% -> -43% (2011-26) · drawdown when switching SPY->TLT on the stress warning Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked retail_switch.py; SPY->TLT full switch on the Risk Reducer warning, 2011-2026 incl the 2022 bond crash.
▸All Weather (Ray Dalio) Ray Dalio / Bridgewater	D+6%	B-23%	C0.78	B	B
What it's meant to do An allocation that makes money in any economy (30% stocks, 55% bonds, 15% gold/commodities) — Ray Dalio / Bridgewater What we found The deep history splits the verdict cleanly: All Weather HELD in the 2008 deflationary crash (-2%, its long bonds rallied as rates fell) but FAILED in the 2022 inflation shock (-18%, as much as stocks -- bonds fell WITH them). It's all-weather except the one weather, inflation, that the 1970s and 2022 are made of. The 'any economy' claim is half-true at best. The numbers +6.5%/yr, Sharpe 0.78, worst drawdown -23%, 2008 -2% (held) but 2022 -18% (failed) · real test, 2006-2026 (incl 2008) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked portfolio_audit.py; commodities proxied by oil, adjusted total-return (ohlcv_long).
▸HFEA ('Excellent Adventure', 3x leveraged) Bogleheads / 'Hedgefundie'	A+24%	F-65%	—	D	C
What it's meant to do 55% 3x-stocks (UPRO) / 45% 3x-bonds (TMF) = leveraged risk parity, huge returns — Bogleheads / 'Hedgefundie' What we found Now stress-tested across BOTH crises, and the verdict only hardens: highest return on the board (+23.5%/yr) and an UN-SURVIVABLE -65% drawdown -- it lost -32% in 2008 and -59% in 2022, two separate wipeouts a decade apart. A leveraged strategy that hands back a generation's gains in a year. Return you cannot actually hold onto. The numbers +23.5%/yr BUT worst drawdown -65%, 2008 -32%, 2022 -59% · real test (3x), 2006-2026 (incl 2008) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked portfolio_audit.py; UPRO/TMF simulated as real 3x DAILY returns (captures the leverage decay).
▸Hedge Funds, Average (HFRI composite, net of fees) The hedge fund industry	F+-8%	—	—	B	B
What it's meant to do Smart money, absolute returns, worth 2-and-20 — The hedge fund industry What we found The expenses are the story: 2-and-20 (a 2% management fee PLUS 20% of profits) turns ~12% gross into ~7.6% net before the gap even opens; fund-of-funds stack another ~1-and-10 on top. The average hedge fund has trailed a cheap index fund for 15+ years and only 'wins' by losing less in bad years. You pay hedge-fund prices for less-than-index returns. The numbers Avg hedge fund ~6-8%/yr since 2009 vs S&P ~13%; beat the index only in down years (2015, 2018, 2022) · HFRI composite net of fees, since 2009 Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked HFR/HFRI composite (net-of-fee, industry standard); see also fam-buffett-bet.
▸The Buffett Bet (index fund vs hedge funds-of-funds) Protege Partners (vs Buffett)	C+7%	—	—	B	B
What it's meant to do Five hand-picked hedge funds-of-funds will beat a plain S&P 500 index fund over 10 years — Protege Partners (vs Buffett) What we found Buffett's $1M wager, settled. A no-effort index fund roughly TRIPLED the average of five expert-selected hedge funds-of-funds over a decade -- net of their double layer of fees (2-and-20 at the funds, plus another fee layer at the fund-of-funds). The cleanest single indictment of paying up for 'smart money.' The numbers 2008-2017: Vanguard S&P 500 index +125.8% (~7.1%/yr) vs the 5 funds-of-funds avg ~+36% (~3.2%/yr) · 2008-2017, net of all fees Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked Berkshire 2017 shareholder letter; Buffett vs Protege Partners 2008-2017.
▸The 4-ETF Momentum Rotation Rotation-strategy crowd	—	—	B0.83	B	C
What it's meant to do Rotate into the strongest of {SPY, QQQ, GLD, IBB} and you beat buy & hold — Rotation-strategy crowd What we found The clever rotation loses to just holding all four equally. The basket already self-hedges through gold — the timing layer is subtracting value, not adding it. The numbers Equal-weighting the 4 ETFs (Sharpe 0.83) BEATS every top-1/top-2/gated rotation; cash gate HURT · Sharpe — equal-weight vs. every rotation Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked momentum_bucket.py / momentum_gated.py.
Build Your Own Index — One Cold Rule · 20 tested
▸Cap-Weight (the index everyone buys) S&P / Vanguard et al.	B+12%	C-34%	C0.74	A	A
What it's meant to do Weight every holding by its market value -- the default 'total market' index — S&P / Vanguard et al. What we found The honest baseline: buy & hold, +12%/yr, Sharpe 0.74, -34% drawdown. It genuinely WORKS and is hard to beat -- but NOT because cap-weight is optimal. It won this era by concentrating ever-harder into a handful of mega-caps. Beatable by a cold rule (see momentum); not by 'opinion'. The numbers +12.5%/yr CAGR, Sharpe 0.74, worst drawdown -34% (2017-21 +16/yr, 2022-now +9/yr) · %/yr return (vs. just buying the index) Performance A a real, durable edge Rigor C 3/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cold_index.py; Sharadar DAILY+SEP, delisted incl; no transaction cost yet (annual rebalance = low turnover).
▸iShares MSCI USA Momentum Factor (MTUM) Buyable momentum ETF	A+18%	C-34%	B0.89	B	B
What it's meant to do Just buy the top momentum stocks -- no work, off the shelf — Buyable momentum ETF What we found The famous $15B+ momentum ETF -- real edge (+6%/yr vs SPY) but the overrated default: SPMO and even Fidelity's FDMO beat it. Its risk-adjusting and twice-a-year rebalance give up ground to simpler momentum. Good, not best. The numbers +18.5%/yr net of fee, Sharpe 0.89, worst drawdown -34%, expense 0.15% (risk-adjusted 6/12mo return, cap-weighted, ~125 stk, semiannual) · %/yr return (vs. just buying the index) Performance A a real, durable edge Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked momentum_etf.py; Sharadar SFP closeadj (fee baked in), vs our synthetic equal-weight momentum (+27.2%/yr) and SPY.
▸Fidelity Momentum Factor (FDMO) Buyable momentum ETF	A+16%	C-34%	B0.87	B	B
What it's meant to do Just buy the top momentum stocks -- no work, off the shelf — Buyable momentum ETF What we found Fidelity's momentum factor: a quiet over-performer (+4%/yr vs SPY, Sharpe 0.87) that beat the famous MTUM at the same 0.15% fee. The unglamorous pick that works. The numbers +16.4%/yr net of fee, Sharpe 0.87, worst drawdown -34%, expense 0.15% (multi-signal momentum, cap-weighted) · %/yr return (vs. just buying the index) Performance A a real, durable edge Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked momentum_etf.py; Sharadar SFP closeadj (fee baked in), vs our synthetic equal-weight momentum (+27.2%/yr) and SPY.
▸Vanguard US Momentum Factor (VFMO) Buyable momentum ETF	A+16%	C-37%	C0.74	B	B
What it's meant to do Just buy the top momentum stocks -- no work, off the shelf — Buyable momentum ETF What we found Vanguard's momentum factor, 0.13% fee: solid (+3%/yr vs SPY). A broad (~550-stock) build -- more diversified, a touch less punch than the concentrated S&P version. The numbers +15.8%/yr net of fee, Sharpe 0.74, worst drawdown -37%, expense 0.13% (rules-based momentum, ~550 stk, monthly) · %/yr return (vs. just buying the index) Performance A a real, durable edge Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked momentum_etf.py; Sharadar SFP closeadj (fee baked in), vs our synthetic equal-weight momentum (+27.2%/yr) and SPY.
▸Total US market (VTI) What people actually buy	B+11%	D-55%	C0.64	A	A
What it's meant to do Total US market is the index to own — What people actually buy What we found 'Own the entire US market' turns out to be ~the same as owning the S&P 500 (10.9%/yr, Sharpe 0.64) -- large caps dominate the total market, so the two are nearly identical. A fine, honest default; just don't expect it to differ from the S&P. The numbers +10.9%/yr, Sharpe 0.64, worst drawdown -55% (2008 -37%, 2022 -20%) · %/yr buy & hold (2004-2026) Performance A a real, durable edge Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked index_choice.py; adjusted total-return (ohlcv_long), common window incl 2008 + 2022.
▸Dow Jones 30 (DIA) What people actually buy	C+10%	D-52%	C0.61	A	A
What it's meant to do Dow Jones 30 is the index to own — What people actually buy What we found The old-economy Dow: lower long-run return (+9.8%/yr) but the best crisis behavior in 2022 (-7%) thanks to its value tilt. A steadier large-cap ride for less upside. The numbers +9.8%/yr, Sharpe 0.61, worst drawdown -52% (2008 -32%, 2022 -7%) · %/yr buy & hold (2004-2026) Performance A a real, durable edge Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked index_choice.py; adjusted total-return (ohlcv_long), common window incl 2008 + 2022.
▸Price Momentum (buy what's already rising) Trading Truth (our test)	A+27%	D-45%	B0.90	B	B
What it's meant to do Rank the universe by 12-month return; hold the top 100, equal-weight — Trading Truth (our test) What we found THE cold winner. The disciplined version of 'fastest-growing cap': it beat cap-weight by +15%/yr, with a BETTER Sharpe (0.9), and it held up in BOTH the 2017-21 mega-cap bull (+25/yr) AND through the 2022 reset and after (+30/yr). Caveat: a momentum-friendly decade, a still-brutal -45% drawdown, costs not yet modeled. But the cold answer to your question is momentum. The numbers +27.2%/yr CAGR, Sharpe 0.9, worst drawdown -45% (2017-21 +25/yr, 2022-now +30/yr) · %/yr return (vs. just buying the index) Performance B it works Rigor C 3/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cold_index.py; Sharadar DAILY+SEP, delisted incl; no transaction cost yet (annual rebalance = low turnover).
▸Invesco S&P 500 Momentum (SPMO) Buyable momentum ETF	A+22%	C-31%	A1.06	B	B
What it's meant to do Just buy the top momentum stocks -- no work, off the shelf — Buyable momentum ETF What we found The best buyable momentum ETF we tested -- and it's NOT the famous one. Captured +10%/yr over SPY with the BEST Sharpe of anything here (1.06) and the smallest drawdown (-31%), for a 0.13% fee. Cap-weighting the S&P 500's momentum leaders rode the same wave our synthetic test found, with a far smoother ride. If you want momentum off the shelf, this is it. The numbers +22.1%/yr net of fee, Sharpe 1.06, worst drawdown -31%, expense 0.13% (S&P 500 momentum score, cap-weighted, ~100 stk, semiannual) · %/yr return (vs. just buying the index) Performance B it works Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked momentum_etf.py; Sharadar SFP closeadj (fee baked in), vs our synthetic equal-weight momentum (+27.2%/yr) and SPY.
▸Nasdaq-100 (QQQ) What people actually buy	A+15%	D-53%	C0.76	A	A
What it's meant to do Nasdaq-100 is the index to own — What people actually buy What we found Highest return AND best risk-adjusted return on the board (+15.0%/yr, Sharpe 0.76) -- but the tech concentration that drove it also drove the worst crashes (-42% in 2008, -33% in 2022). The most reward, and the scariest ride to get it. The numbers +15.0%/yr, Sharpe 0.76, worst drawdown -53% (2008 -42%, 2022 -33%) · %/yr buy & hold (2004-2026) Performance B it works Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked index_choice.py; adjusted total-return (ohlcv_long), common window incl 2008 + 2022.
▸Fastest-Growing Market Cap (top 100) Trading Truth (our test)	A+22%	F-63%	C0.76	B	B
What it's meant to do Rank by 12-month market-cap growth; hold the top 100, equal-weight — Trading Truth (our test) What we found Your instinct is right -- it beat cap-weight by +9%/yr. But cap growth is mostly price growth, so this is momentum's wilder cousin: same idea, a far worse -63% drawdown. Momentum (ranking on price directly) does the same job with less wreckage. The numbers +21.9%/yr CAGR, Sharpe 0.76, worst drawdown -63% (2017-21 +24/yr, 2022-now +20/yr) · %/yr return (vs. just buying the index) Performance C marginal — barely better than a coin flip Rigor C 3/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cold_index.py; Sharadar DAILY+SEP, delisted incl; no transaction cost yet (annual rebalance = low turnover).
▸Lowest-Volatility 100 Low-vol 'anomaly' crowd	D+7%	D-43%	D0.49	B	B
What it's meant to do Hold the 100 calmest stocks (lowest trailing volatility), equal-weight — Low-vol 'anomaly' crowd What we found Defensive as advertised, but it lagged badly in a bull (+7%/yr vs cap-weight's +12). A smoother ride to a smaller number. Earns its keep only if you weight risk-adjusted return over absolute -- most buy-and-holders don't. The numbers +6.9%/yr CAGR, Sharpe 0.49, worst drawdown -43% (2017-21 +11/yr, 2022-now +3/yr) · %/yr return (vs. just buying the index) Performance C marginal — barely better than a coin flip Rigor C 3/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cold_index.py; Sharadar DAILY+SEP, delisted incl; no transaction cost yet (annual rebalance = low turnover).
▸Invesco Dorsey Wright Momentum (PDP) Buyable momentum ETF	A+15%	C-35%	C0.73	B	B
What it's meant to do Just buy the top momentum stocks -- no work, off the shelf — Buyable momentum ETF What we found Real momentum, but the 0.62% fee is a tax: +2%/yr vs SPY, roughly HALF the edge of SPMO at 5x the cost. Same idea, worse wrapper. Watch the expense ratio. The numbers +14.7%/yr net of fee, Sharpe 0.73, worst drawdown -35%, expense 0.62% (relative-strength rank, ~100 stk, quarterly) · %/yr return (vs. just buying the index) Performance C marginal — barely better than a coin flip Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked momentum_etf.py; Sharadar SFP closeadj (fee baked in), vs our synthetic equal-weight momentum (+27.2%/yr) and SPY.
▸Alpha Architect US Quant Momentum (QMOM) Buyable momentum ETF	A+14%	C-39%	C0.61	B	B
What it's meant to do Just buy the top momentum stocks -- no work, off the shelf — Buyable momentum ETF What we found The concentrated bet (~50 stocks): it promised the most punch and delivered the LEAST (+1%/yr vs SPY) with the worst drawdown (-39%). Concentration cut both ways and lost. More conviction, less payoff. The numbers +13.7%/yr net of fee, Sharpe 0.61, worst drawdown -39%, expense 0.29% (concentrated high-momentum ~50 stk, quarterly) · %/yr return (vs. just buying the index) Performance C marginal — barely better than a coin flip Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked momentum_etf.py; Sharadar SFP closeadj (fee baked in), vs our synthetic equal-weight momentum (+27.2%/yr) and SPY.
▸S&P 500, equal-weighted (RSP) What people actually buy	B+10%	F-60%	D0.59	A	A
What it's meant to do S&P 500, equal-weighted is the index to own — What people actually buy What we found The popular 'equal-weight beats cap-weight' claim did NOT hold here: Sharpe 0.59 vs the cap-weighted S&P's 0.64, and the WORST drawdown of the bunch (-60%). Its small/value tilt was a drag through the mega-cap era. Sensible idea, didn't pay this cycle. The numbers +10.2%/yr, Sharpe 0.59, worst drawdown -60% (2008 -40%, 2022 -12%) · %/yr buy & hold (2004-2026) Performance C marginal — barely better than a coin flip Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked index_choice.py; adjusted total-return (ohlcv_long), common window incl 2008 + 2022.
▸S&P MidCap 400 (MDY) What people actually buy	B+10%	D-55%	D0.55	A	A
What it's meant to do S&P MidCap 400 is the index to own — What people actually buy What we found Mid-caps were supposed to be the 'sweet spot.' They weren't -- Sharpe 0.55, below the plain S&P. The size premium didn't show up. The numbers +10.1%/yr, Sharpe 0.55, worst drawdown -55% (2008 -36%, 2022 -13%) · %/yr buy & hold (2004-2026) Performance C marginal — barely better than a coin flip Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked index_choice.py; adjusted total-return (ohlcv_long), common window incl 2008 + 2022.
▸The 100 Biggest, Equal-Weight 'Own the blue chips' folk wisdom	C+9%	C-36%	D0.56	A	A
What it's meant to do Hold the 100 largest companies, equal-weight — 'Own the blue chips' folk wisdom What we found Owning the giants equally LOSES to cap-weighting them (-4%/yr). The lesson: cap-weight's whole edge this decade was over-weighting the very biggest, not owning big names per se. 'Buy the leaders' equally is worse than the index. The numbers +8.9%/yr CAGR, Sharpe 0.56, worst drawdown -36% (2017-21 +13/yr, 2022-now +5/yr) · %/yr return (vs. just buying the index) Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cold_index.py; Sharadar DAILY+SEP, delisted incl; no transaction cost yet (annual rebalance = low turnover).
▸Cheapest 100 by P/E (value) Value investors everywhere	B+10%	F-61%	D0.50	B	B
What it's meant to do Hold the 100 lowest-P/E stocks, equal-weight — Value investors everywhere What we found Value's lost decade, in one line: +10%/yr (below cap-weight) with a -61% drawdown. Buying cheap was a cold, honest rule that simply did not pay in 2016-2026. The numbers +10.2%/yr CAGR, Sharpe 0.5, worst drawdown -61% (2017-21 +11/yr, 2022-now +10/yr) · %/yr return (vs. just buying the index) Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cold_index.py; Sharadar DAILY+SEP, delisted incl; no transaction cost yet (annual rebalance = low turnover).
▸Cheapest 100 by Price/Book (deep value) Graham / deep-value crowd	C+8%	F-71%	F0.39	B	B
What it's meant to do Hold the 100 lowest price-to-book stocks, equal-weight — Graham / deep-value crowd What we found Deeper value, deeper hole: +8%/yr and the WORST drawdown of every cold rule we tried (-71%). Price-to-book as an organizing principle has been a wealth hazard this era. The numbers +7.5%/yr CAGR, Sharpe 0.39, worst drawdown -71% (2017-21 +7/yr, 2022-now +8/yr) · %/yr return (vs. just buying the index) Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cold_index.py; Sharadar DAILY+SEP, delisted incl; no transaction cost yet (annual rebalance = low turnover).
▸Most Expensive 100 by P/E (hot growth) Story-stock / 'disruption' chasers	B+10%	D-41%	D0.54	B	B
What it's meant to do Hold the 100 highest-P/E stocks, equal-weight — Story-stock / 'disruption' chasers What we found The trap that looks like momentum but isn't. It rode the 2017-21 hype (+18/yr) then COLLAPSED after 2022 (+2/yr). Paying any price for 'growth' died in the reset -- while real momentum kept working. Expensive != winning. The numbers +10.3%/yr CAGR, Sharpe 0.54, worst drawdown -41% (2017-21 +18/yr, 2022-now +2/yr) · %/yr return (vs. just buying the index) Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cold_index.py; Sharadar DAILY+SEP, delisted incl; no transaction cost yet (annual rebalance = low turnover).
▸Russell 2000 small-cap (IWM) What people actually buy	C+9%	F-59%	D0.49	A	A
What it's meant to do Russell 2000 small-cap is the index to own — What people actually buy What we found The famous 'small-cap premium': absent. Worst risk-adjusted return of every index here (Sharpe 0.49), no return edge to compensate. Small caps were a worse deal, not a better one. The numbers +9.1%/yr, Sharpe 0.49, worst drawdown -59% (2008 -34%, 2022 -20%) · %/yr buy & hold (2004-2026) Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked index_choice.py; adjusted total-return (ohlcv_long), common window incl 2008 + 2022.
Option Strategies — Net of Real Fills · 10 tested
▸Short Put TastyTrade / premium sellers	—	—	—	C	D
What it's meant to do Selling cash-secured puts on the index is high-probability income — TastyTrade / premium sellers What we found Real edge that survives real fills: +48/trade crossing the spread (keeps 94% of the +51 mid-price number their studies quote), 87% win. One leg = one tight, liquid spread to cross — the most friction-proof trade of the lot. But the win rate hides the tail — worst trade -4930 on a 1-lot. The numbers mid +51 → cross +48/trade, keeps 94% of edge, 87% win, worst -4930 · $ per trade (real fills) The options detail Wins 87% of trades · $+51/trade at the mid → $+48 once you cross the real bid/ask · 94% of the edge survives · worst single trade $-4930. And this is the version that only works on SPY — on a less-liquid name the wider spread turns most of these negative. Performance A a real, durable edge Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸Jade Lizard TastyTrade / premium sellers	—	—	—	C	D
What it's meant to do The jade lizard collects a credit with no upside risk — TastyTrade / premium sellers What we found Real edge that survives real fills: +27/trade crossing the spread (keeps 66% of the +41 mid-price number their studies quote), 66% win. Half the tail of a strangle — the capped call side removes upside risk as designed. But the win rate hides the tail — worst trade -2530 on a 1-lot. The numbers mid +41 → cross +27/trade, keeps 66% of edge, 66% win, worst -2530 · $ per trade (real fills) The options detail Wins 66% of trades · $+41/trade at the mid → $+27 once you cross the real bid/ask · 66% of the edge survives · worst single trade $-2530. And this is the version that only works on SPY — on a less-liquid name the wider spread turns most of these negative. Performance A a real, durable edge Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸Big Lizard + Risk Reducer Trading Truth (our test)	—	—	—	C	D
What it's meant to do Gate the Big Lizard with the Risk Reducer to cut the naked-ATM-put tail — Trading Truth (our test) What we found The flagship Risk Reducer applied to an option trade: it cuts the worst trade 75% (-6929→-1712) and max drawdown 70% by standing aside in risk-off regimes. The price: about half the expectancy ($107→$18). A real tail-reducer, not a free lunch. The numbers worst -6929→-1712, worst drawdown -14659→-4382 · $ per trade (real fills) Performance A a real, durable edge Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸Risk Reducer book-exit (put-write) Trading Truth (our test)	—	—	—	C	D
What it's meant to do Apply the Risk Reducer as a book-level risk-off EXIT on a real put-write to cut the tail — Trading Truth (our test) What we found The flagship Risk Reducer, applied as a book-level risk-off EXIT (flatten mid-trade, not just gate entries), cut a real put-write's max drawdown 74% and turned 2022 positive -- the best risk overlay we've tested. Costs ~58% of annual return; the prior-day-signal causality check is still pending (lookahead not yet cleared). The numbers worst drawdown -$61,674 -> -$15,960 (cut 74%), 2022 -$41,800 -> +$1,919, return-vs-drawdown 0.31 -> 0.50 · return per $ of worst drop Performance A a real, durable edge Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✗ look-ahead checked es_putwrite_book.py with the correctly-scaled Risk Reducer; ONE regime (2022-25, no 2008/2020); 29 risk-off flatten events.
▸Short Strangle TastyTrade / premium sellers	—	—	—	C	D
What it's meant to do The 16-delta strangle is TastyTrade's bread-and-butter income trade — TastyTrade / premium sellers What we found Real edge that survives real fills: +31/trade crossing the spread (keeps 79% of the +40 mid-price number their studies quote), 75% win. Undefined-risk: keeps its edge through the spread, but you carry the naked tail. But the win rate hides the tail — worst trade -4859 on a 1-lot. The numbers mid +40 → cross +31/trade, keeps 79% of edge, 75% win, worst -4859 · $ per trade (real fills) The options detail Wins 75% of trades · $+40/trade at the mid → $+31 once you cross the real bid/ask · 79% of the edge survives · worst single trade $-4859. And this is the version that only works on SPY — on a less-liquid name the wider spread turns most of these negative. Performance B it works Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸Short Straddle TastyTrade / premium sellers	—	—	—	C	D
What it's meant to do Selling the ATM straddle harvests the most premium — TastyTrade / premium sellers What we found Real edge that survives real fills: +62/trade crossing the spread (keeps 60% of the +103 mid-price number their studies quote), 68% win. Undefined-risk: keeps its edge through the spread, but you carry the naked tail. But the win rate hides the tail — worst trade -6550 on a 1-lot. The numbers mid +103 → cross +62/trade, keeps 60% of edge, 68% win, worst -6550 · $ per trade (real fills) The options detail Wins 68% of trades · $+103/trade at the mid → $+62 once you cross the real bid/ask · 60% of the edge survives · worst single trade $-6550. And this is the version that only works on SPY — on a less-liquid name the wider spread turns most of these negative. Performance B it works Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸Big Lizard TastyTrade / premium sellers	—	—	—	C	D
What it's meant to do The big lizard sells fat ATM premium with no upside risk — TastyTrade / premium sellers What we found Real edge that survives real fills: +107/trade crossing the spread (keeps 73% of the +146 mid-price number their studies quote), 74% win. Undefined-risk: keeps its edge through the spread, but you carry the naked tail. But the win rate hides the tail — worst trade -6929 on a 1-lot. The numbers mid +146 → cross +107/trade, keeps 73% of edge, 74% win, worst -6929 · $ per trade (real fills) The options detail Wins 74% of trades · $+146/trade at the mid → $+107 once you cross the real bid/ask · 73% of the edge survives · worst single trade $-6929. And this is the version that only works on SPY — on a less-liquid name the wider spread turns most of these negative. Performance B it works Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸Poor Man's Covered Call (PMCC) Diagonal / PMCC crowd	—	—	—	C	D
What it's meant to do A long LEAPS call + rolling short calls = a cheap covered-call income machine — Diagonal / PMCC crowd What we found It made money ($434/campaign, 67% win) — but ONLY as leveraged long SPY. The call-selling 'income' it's sold for was a -596/campaign DRAG: in a bull market your short calls keep getting run over. Capped-upside beta, not income. And real downside: worst -3713, worst drawdown -8051. The numbers 67% win, $434/campaign, worst -3713, worst drawdown -8051 · $ per trade (real fills) Performance C marginal — barely better than a coin flip Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸Iron Condor TastyTrade / premium sellers	—	—	—	C	D
What it's meant to do The iron condor is a safe, defined-risk income strategy — TastyTrade / premium sellers What we found Sold as 'safe, defined-risk income' — but it loses money even at patient mid-price fills (-13/trade), and worse (-23) once you cross the spread. 4 legs means paying the bid/ask 4 times in and 4 out. The 60% win rate is a mirage. (Independently confirms our iron-condor audit.) The numbers mid -13 → cross -23/trade, 60% win, worst -677 · $ per trade (real fills) The options detail Wins 60% of trades · $-13/trade at the mid → $-23 once you cross the real bid/ask · worst single trade $-677. And this is the version that only works on SPY — on a less-liquid name the wider spread turns most of these negative. Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸Iron Fly TastyTrade / premium sellers	—	—	—	C	D
What it's meant to do The iron fly collects rich premium with defined risk — TastyTrade / premium sellers What we found Sold as 'safe, defined-risk income' — but it loses money even at patient mid-price fills (-6/trade), and worse (-64) once you cross the spread. 4 legs means paying the bid/ask 4 times in and 4 out. The 50% win rate is a mirage. (Independently confirms our iron-condor audit.) The numbers mid -6 → cross -64/trade, 50% win, worst -1091 · $ per trade (real fills) The options detail Wins 50% of trades · $-6/trade at the mid → $-64 once you cross the real bid/ask · worst single trade $-1091. And this is the version that only works on SPY — on a less-liquid name the wider spread turns most of these negative. Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
Thesis ETFs — Did the Theme Pay? · 25 tested
▸Space & Defense (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Space & Defense thesis is worth a dedicated fund — Thematic ETF industry What we found Only 63% of these funds beat just buying SPY over their own life; the median trailed by 2.3%/yr. 7% are already dead. The theme actually paid — one of the few that did. The numbers 28 funds, 7% dead, 63% beat SPY · %/yr vs. just buying the index (net of fee) Performance B it works Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸5G & Connectivity (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The 5G & Connectivity thesis is worth a dedicated fund — Thematic ETF industry What we found Only 100% of these funds beat just buying SPY over their own life; the median trailed by 5.0%/yr. 0% are already dead. The theme actually paid — one of the few that did. The numbers 1 funds, 0% dead, 100% beat SPY · %/yr vs. just buying the index (net of fee) Performance B it works Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸AI & Big Data (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The AI & Big Data thesis is worth a dedicated fund — Thematic ETF industry What we found Only 68% of these funds beat just buying SPY over their own life; the median trailed by 5.5%/yr. 13% are already dead. The theme actually paid — one of the few that did. The numbers 38 funds, 13% dead, 68% beat SPY · %/yr vs. just buying the index (net of fee) Performance B it works Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Semiconductors (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Semiconductors thesis is worth a dedicated fund — Thematic ETF industry What we found Only 81% of these funds beat just buying SPY over their own life; the median trailed by 19.4%/yr. 6% are already dead. The theme actually paid — one of the few that did. The numbers 18 funds, 6% dead, 81% beat SPY · %/yr vs. just buying the index (net of fee) Performance B it works Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Robotics & Automation (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Robotics & Automation thesis is worth a dedicated fund — Thematic ETF industry What we found Only 50% of these funds beat just buying SPY over their own life; the median trailed by 0.9%/yr. 0% are already dead. A coin-flip vs. just holding the index, before you even count the dead ones. The numbers 8 funds, 0% dead, 50% beat SPY · %/yr vs. just buying the index (net of fee) Performance C marginal — barely better than a coin flip Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Cannabis (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Cannabis thesis is worth a dedicated fund — Thematic ETF industry What we found Only 0% of these funds beat just buying SPY over their own life; the median trailed by 41.4%/yr. 60% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 10 funds, 60% dead, 0% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸EV & Battery (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The EV & Battery thesis is worth a dedicated fund — Thematic ETF industry What we found Only 8% of these funds beat just buying SPY over their own life; the median trailed by 18.9%/yr. 15% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 13 funds, 15% dead, 8% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Robotics-Health/MedTech (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Robotics-Health/MedTech thesis is worth a dedicated fund — Thematic ETF industry What we found Only 0% of these funds beat just buying SPY over their own life; the median trailed by 17.2%/yr. 25% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 4 funds, 25% dead, 0% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Clean Water/Agri/Food (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Clean Water/Agri/Food thesis is worth a dedicated fund — Thematic ETF industry What we found Only 0% of these funds beat just buying SPY over their own life; the median trailed by 13.8%/yr. 55% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 11 funds, 55% dead, 0% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Clean / Renewable Energy (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Clean / Renewable Energy thesis is worth a dedicated fund — Thematic ETF industry What we found Only 0% of these funds beat just buying SPY over their own life; the median trailed by 12.0%/yr. 41% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 17 funds, 41% dead, 0% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Fintech (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Fintech thesis is worth a dedicated fund — Thematic ETF industry What we found Only 12% of these funds beat just buying SPY over their own life; the median trailed by 11.3%/yr. 56% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 9 funds, 56% dead, 12% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Blockchain & Web3 Equity (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Blockchain & Web3 Equity thesis is worth a dedicated fund — Thematic ETF industry What we found Only 36% of these funds beat just buying SPY over their own life; the median trailed by 10.5%/yr. 19% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 26 funds, 19% dead, 36% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Longevity & Aging (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Longevity & Aging thesis is worth a dedicated fund — Thematic ETF industry What we found Only 8% of these funds beat just buying SPY over their own life; the median trailed by 10.3%/yr. 15% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 13 funds, 15% dead, 8% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Genomics & Biotech (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Genomics & Biotech thesis is worth a dedicated fund — Thematic ETF industry What we found Only 6% of these funds beat just buying SPY over their own life; the median trailed by 9.9%/yr. 21% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 19 funds, 21% dead, 6% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Gaming & Esports (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Gaming & Esports thesis is worth a dedicated fund — Thematic ETF industry What we found Only 12% of these funds beat just buying SPY over their own life; the median trailed by 8.7%/yr. 29% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 17 funds, 29% dead, 12% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Disruptive / Innovation (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Disruptive / Innovation thesis is worth a dedicated fund — Thematic ETF industry What we found Only 9% of these funds beat just buying SPY over their own life; the median trailed by 7.8%/yr. 25% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 92 funds, 25% dead, 9% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Rare Earth & Materials (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Rare Earth & Materials thesis is worth a dedicated fund — Thematic ETF industry What we found Only 0% of these funds beat just buying SPY over their own life; the median trailed by 7.4%/yr. 25% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 4 funds, 25% dead, 0% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Internet & Social (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Internet & Social thesis is worth a dedicated fund — Thematic ETF industry What we found Only 7% of these funds beat just buying SPY over their own life; the median trailed by 7.3%/yr. 37% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 19 funds, 37% dead, 7% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Water (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Water thesis is worth a dedicated fund — Thematic ETF industry What we found Only 0% of these funds beat just buying SPY over their own life; the median trailed by 5.5%/yr. 17% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 6 funds, 17% dead, 0% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Infrastructure (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Infrastructure thesis is worth a dedicated fund — Thematic ETF industry What we found Only 24% of these funds beat just buying SPY over their own life; the median trailed by 5.4%/yr. 23% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 43 funds, 23% dead, 24% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Cloud & Software (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Cloud & Software thesis is worth a dedicated fund — Thematic ETF industry What we found Only 22% of these funds beat just buying SPY over their own life; the median trailed by 4.4%/yr. 27% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 11 funds, 27% dead, 22% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Clean Transport/Smart Mobility (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Clean Transport/Smart Mobility thesis is worth a dedicated fund — Thematic ETF industry What we found Only 50% of these funds beat just buying SPY over their own life; the median trailed by 3.9%/yr. 0% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 2 funds, 0% dead, 50% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Cybersecurity (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Cybersecurity thesis is worth a dedicated fund — Thematic ETF industry What we found Only 25% of these funds beat just buying SPY over their own life; the median trailed by 3.4%/yr. 20% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 15 funds, 20% dead, 25% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Nuclear & Uranium (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Nuclear & Uranium thesis is worth a dedicated fund — Thematic ETF industry What we found Only 44% of these funds beat just buying SPY over their own life; the median trailed by 1.9%/yr. 17% are already dead. The theme was sold hard; the funds trailed the index and many were buried. The numbers 12 funds, 17% dead, 44% beat SPY · %/yr vs. just buying the index (net of fee) Performance F doesn't do what it claims Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
▸Crypto Price (BTC/ETH) (thesis ETFs) Thematic ETF industry	—	—	—	C	B
What it's meant to do The Crypto Price (BTC/ETH) thesis is worth a dedicated fund — Thematic ETF industry What we found 76 funds tracked, 13% already dead. This is price exposure, not an equity thesis — we don't grade it against the stock index. Tracked for the graveyard. The numbers 76 funds, 13% dead · %/yr vs. just buying the index (net of fee) Performance — control / benchmark Rigor B 4/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked thesis_etf_tracker.py; SFP closeadj (net of fee), dead funds tracked to death (no survivorship cheat).
Stock-Picking Scanners · 9 tested
▸CAN SLIM (full) William O'Neil / IBD	D+6%	—	—	B	C
What it's meant to do High relative-strength leaders near highs WITH real earnings growth outperform — William O'Neil / IBD What we found The one scanner that cleared every bar: skill over peers, survives out-of-sample, AND beats the index net of cost. The earnings leg is what flips it from hype to real. The numbers skill over similar stocks +5.0, early-years +2.9 → the later years +6.6, net return vs the index +6.1%/yr · %/yr beating the index (after costs) Performance A a real, durable edge Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked scanner_tracker.py — skill over similar stocks = 63d alpha over equal-weight survivorship-free universe = skill; net return vs the index net of Corwin-Schultz cost by cap bucket; IS<2021, the later years≥2021.
▸Value (Earnings Yield E/P) Academic / value tradition	—	—	—	B	C
What it's meant to do Cheap stocks (high earnings yield) beat the market — Academic / value tradition What we found Weak but real, and the in-sample/out-of-sample split tells the famous value story: dead in the 2010s, breathing again lately. The numbers skill over similar stocks +1.0 (early-years −0.6 → the later years +2.1, the value 'lost decade' then revival) · %/yr beating the index (after costs) Performance C marginal — barely better than a coin flip Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked scanner_tracker.py — skill over similar stocks = 63d alpha over equal-weight survivorship-free universe = skill; net return vs the index net of Corwin-Schultz cost by cap bucket; IS<2021, the later years≥2021.
▸Earnings-Surprise (SUE / PEAD) Academic (Bernard-Thomas)	D+4%	—	—	B	C
What it's meant to do Stocks that beat earnings drift higher for weeks (post-earnings drift) — Academic (Bernard-Thomas) What we found Genuine stock-picking skill over peers — but the drift gets eaten by trading costs. Real effect, not a tradeable edge net of friction. The numbers skill over similar stocks +2.9 (real skill) but net return vs the index −3.7%/yr after cost · %/yr beating the index (after costs) Performance D oversold — famous, underdelivers Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked scanner_tracker.py — skill over similar stocks = 63d alpha over equal-weight survivorship-free universe = skill; net return vs the index net of Corwin-Schultz cost by cap bucket; IS<2021, the later years≥2021.
▸IBD Relative-Strength Leaders IBD	D+6%	—	—	B	C
What it's meant to do Buy the highest relative-strength names near 52-week highs — IBD What we found Price leadership alone picks winners over peers but loses to the index net of cost. Momentum without an earnings filter is half a strategy. The numbers skill over similar stocks +2.2 but net return vs the index −6.1%/yr · %/yr beating the index (after costs) Performance D oversold — famous, underdelivers Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked scanner_tracker.py — skill over similar stocks = 63d alpha over equal-weight survivorship-free universe = skill; net return vs the index net of Corwin-Schultz cost by cap bucket; IS<2021, the later years≥2021.
▸CANSLIM + Sector Overlay O'Neil ('leaders in leading groups')	D+6%	—	—	B	C
What it's meant to do Restricting CANSLIM picks to the strongest sectors should improve it — O'Neil ('leaders in leading groups') What we found A sensible-sounding overlay that HURTS. Filtering CANSLIM to leading sectors cut the edge (+3.6 vs +5.0 over peers) and gutted the net return — and the control is the clincher: picks in LEADING sectors (+3.6) ≈ picks in LAGGING sectors (+3.8), so the sector signal adds nothing. The sector strength is already inside each stock's relative strength; the overlay is redundant. The numbers skill over similar stocks +3.6 vs base +5.0; net return vs the index +0.7 vs +6.1%/yr; leading ≈ lagging · %/yr beating the index (after costs) Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked scanner_tracker.py — skill over similar stocks = 63d alpha over equal-weight survivorship-free universe = skill; net return vs the index net of Corwin-Schultz cost by cap bucket; IS<2021, the later years≥2021.
▸CANSLIM + Earnings Acceleration Earnings-momentum crowd	D+6%	—	—	B	C
What it's meant to do Requiring earnings growth to be ACCELERATING on top of price leadership sharpens the screen — Earnings-momentum crowd What we found Same lesson as the sector overlay: bolting an earnings-acceleration filter onto a price screen shrank it without adding edge (+3.6 vs +5.0). Earnings acceleration ALONE is worse than random (+−0.6 over peers). By the time earnings confirm, the price already moved — the story is redundant with relative strength. The numbers skill over similar stocks +3.6 vs base +5.0; net return vs the index −0.1 vs +6.1%/yr · %/yr beating the index (after costs) Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked scanner_tracker.py — skill over similar stocks = 63d alpha over equal-weight survivorship-free universe = skill; net return vs the index net of Corwin-Schultz cost by cap bucket; IS<2021, the later years≥2021.
▸Buy the Fastest Earnings Growers Growth-chasing folk wisdom	A+22%	—	—	B	C
What it's meant to do Owning the companies with the highest earnings growth makes money — Growth-chasing folk wisdom What we found One of the most expensive 'obvious' ideas in markets. Buying the highest-earnings-growth names LOSES — worse than picking 50 stocks at random — because the growth is already priced in and you overpay for it. Growth is a great quality; chasing it is a terrible strategy. The numbers skill over similar stocks −2.0 (WORSE than random), net return vs the index −22%/yr, hit 28% · %/yr beating the index (after costs) Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked scanner_tracker.py — skill over similar stocks = 63d alpha over equal-weight survivorship-free universe = skill; net return vs the index net of Corwin-Schultz cost by cap bucket; IS<2021, the later years≥2021.
▸6-Month Price Momentum (top 50) Momentum tradition	—	—	—	B	C
What it's meant to do Recent winners keep winning — Momentum tradition What we found Worked in-sample, decayed out-of-sample. Naked price momentum has been arbitraged thin. The numbers skill over similar stocks −0.6 (early-years +1.2 → the later years −2.1, regime-flipped) · %/yr beating the index (after costs) Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked scanner_tracker.py — skill over similar stocks = 63d alpha over equal-weight survivorship-free universe = skill; net return vs the index net of Corwin-Schultz cost by cap bucket; IS<2021, the later years≥2021.
▸Random-50 (null control) Us (the control)	—	—	—	B	C
What it's meant to do Nothing — 50 stocks picked at random. Proves the harness isn't lying. — Us (the control) What we found Our honesty check. It SHOULD show no skill — and it does. When it briefly showed +25% we knew the engine had a bug, found it, and fixed it. The control caught the lie. The numbers skill over similar stocks −0.2 (lands at zero, as it must) · %/yr beating the index (after costs) Performance — control / benchmark Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked scanner_tracker.py — skill over similar stocks = 63d alpha over equal-weight survivorship-free universe = skill; net return vs the index net of Corwin-Schultz cost by cap bucket; IS<2021, the later years≥2021.

Signals we tested — do they add an edge?

These aren’t things to own — they’re overlays you’d time with. So they earn one grade: the edge they add over simply buying and holding. Most add none.

Signal · tradition	The Claim	The Verdict	Edge
Market-Timing Signals (Gates) · 60 tested
▸The Risk Reducer Various / folklore	Tells you when to be in the market vs. step aside	FLAGSHIP Tested against every index series we could download; beat all of them. Roughly halved drawdowns. Our flagship — recipe stays in the kitchen.	A+ +0.48 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Tested against every index series we could download; beat all of them. Roughly halved drawdowns. Our flagship — recipe stays in the kitchen. The numbers Sharpe 1.42, worst drawdown -13%, in-market 77% of days · edge vs. buy & hold Performance A+ the one that survived Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked The method stays in the kitchen with the recipe — only the live read and the track record are public.
▸MA200 Various / folklore	Tells you when to be in the market vs. step aside	MARGINAL Barely moves the needle once tested fairly. Famous, not powerful.	C +0.08 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Barely moves the needle once tested fairly. Famous, not powerful. The numbers Sharpe 1.01, worst drawdown -20%, in-market 86% of days · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸MA10mo(210d) Various / folklore	Tells you when to be in the market vs. step aside	MARGINAL Barely moves the needle once tested fairly. Famous, not powerful.	C +0.07 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Barely moves the needle once tested fairly. Famous, not powerful. The numbers Sharpe 1.01, worst drawdown -19%, in-market 86% of days · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸VIX9D<VIX Various / folklore	Tells you when to be in the market vs. step aside	MARGINAL Barely moves the needle once tested fairly. Famous, not powerful.	C +0.06 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Barely moves the needle once tested fairly. Famous, not powerful. The numbers Sharpe 1.00, worst drawdown -17%, in-market 70% of days · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸HY_OAS falling20d Various / folklore	Tells you when to be in the market vs. step aside	MARGINAL Barely moves the needle once tested fairly. Famous, not powerful.	C +0.05 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Barely moves the needle once tested fairly. Famous, not powerful. The numbers Sharpe 0.99, worst drawdown -22%, in-market 57% of days · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸VIX<25 Various / folklore	Tells you when to be in the market vs. step aside	MARGINAL Barely moves the needle once tested fairly. Famous, not powerful.	C +0.05 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Barely moves the needle once tested fairly. Famous, not powerful. The numbers Sharpe 0.99, worst drawdown -23%, in-market 89% of days · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸IG_OAS<60dMA Various / folklore	Tells you when to be in the market vs. step aside	MARGINAL Barely moves the needle once tested fairly. Famous, not powerful.	C +0.04 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Barely moves the needle once tested fairly. Famous, not powerful. The numbers Sharpe 0.98, worst drawdown -19%, in-market 63% of days · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸StLouisFSI<0.5 Various / folklore	Tells you when to be in the market vs. step aside	MARGINAL Barely moves the needle once tested fairly. Famous, not powerful.	C +0.03 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Barely moves the needle once tested fairly. Famous, not powerful. The numbers Sharpe 0.97, worst drawdown -25%, in-market 95% of days · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸VIX<VIX3M(contango) Various / folklore	Tells you when to be in the market vs. step aside	MARGINAL Barely moves the needle once tested fairly. Famous, not powerful.	C +0.01 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Barely moves the needle once tested fairly. Famous, not powerful. The numbers Sharpe 0.94, worst drawdown -24%, in-market 92% of days · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸The Hedgeye Quads (GIP macro model) Hedgeye (GIP / 'quad' macro model)	Read the growth + inflation 'quad' and tilt your asset allocation to the regime	MARGINAL A genuinely useful lens for THINKING about regimes -- different assets lead when the economy reflates versus slows, and that's real. As a tradeable allocation system, though, …	C Sharpe (perfect-foresight rotation vs 60/40)
What it's meant to do Read the growth + inflation 'quad' and tilt your asset allocation to the regime — Hedgeye (GIP / 'quad' macro model) What we found A genuinely useful lens for THINKING about regimes -- different assets lead when the economy reflates versus slows, and that's real. As a tradeable allocation system, though, the prescribed rotation (growth/tech in Quad 1, energy/materials/financials in Quad 2, gold/commodities in Quad 3, long bonds/utilities/gold in Quad 4) trailed a plain 60/40 over the years we can measure -- and even granting it a PERFECT forecast of the next quad, removing the hard part, it reached only Sharpe 0.78 versus 0.98 for 60/40 and 1.06 for buy-and-hold (dividends counted on every asset, both sides). The fair caveat, in its favor: 2011-2026 was one long equity bull -- the toughest possible test for any strategy that holds bonds, gold and commodities part-time. A regime model is built for the stagflation or deflation decade our macro data can't reach, so call this unproven on its home turf, not debunked. The one durable nugget is real: stagflation -- growth down, inflation up -- is the single regime that genuinely bites equities. The catch is you'd have to predict it. The numbers As an allocation model it trailed a plain 60/40 -- and even a PERFECT quad forecast only reached Sharpe 0.78 -- though the only window we can test is an equity bull, its hardest case · Sharpe (perfect-foresight rotation vs 60/40) Performance C marginal — barely better than a coin flip Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✗ point-in-time · ✓ look-ahead checked quad_macro.py; INDPRO & CPI year-over-year rate-of-change -> 4 quads; prescribed per-quad asset baskets (an approximation of the published leanings), equal-weighted, vs 60/40 and SPY. Perfect-foresight version uses the contemporaneous quad. Revised (not point-in-time) macro data and a 100%-equity alt-benchmark both flatter the comparison; the window is a single equity-bull regime.
▸The 5-10-20 EMA Timer Chartist momentum-timer tradition	A chart-based signal that times the market	MARGINAL One of the original momentum timers: bias flips on full crossovers of the 5-, 10- and 20-day EMAs (long when 5>10>20, cash when 5<10<20). It does what trend-following does -- …	C +0.04 edge vs. buy & hold
What it's meant to do A chart-based signal that times the market — Chartist momentum-timer tradition What we found One of the original momentum timers: bias flips on full crossovers of the 5-, 10- and 20-day EMAs (long when 5>10>20, cash when 5<10<20). It does what trend-following does -- cuts the worst drawdown from -55% to -30% -- but at a steep return cost (+6.5% vs buy-hold +9.8%/yr), and its risk-adjusted edge is a rounding error in-sample (+0.04) that goes negative out-of-sample (-0.04). A real tool, beaten by a slower trend filter (and by our own signal). The numbers cuts drawdown -55%->-30% but gives up 3.3%/yr; edge +0.04 in-sample, -0.04 out-of-sample · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked chartist_timers.py; total-return SPY gates vs 3mo T-bill, edge vs buy-hold, the later years=2014+. The slow trend filter beats both as overlays too.
▸Up/Down Volume (Nasdaq A-D volume) StockCharts / breadth-volume chartists	A chart-based signal that times the market	MARGINAL The Nasdaq up/down-volume breadth panel (EMA21 & EMA50 of net up-down volume) as a long/cash gate. 'Both EMAs above zero' has no full-sample edge (edge -0.01) -- it's simpl…	C -0.01 edge vs. buy & hold
What it's meant to do A chart-based signal that times the market — StockCharts / breadth-volume chartists What we found The Nasdaq up/down-volume breadth panel (EMA21 & EMA50 of net up-down volume) as a long/cash gate. 'Both EMAs above zero' has no full-sample edge (edge -0.01) -- it's simply long 61% of the time, tracking the index. The flattering the later years +0.29 isn't supported by the full sample (in-sample edge is nil), and adding the 'climbing' filter actively hurts (-0.27). A volume-weighted cousin of the advance/decline line, same verdict: breadth confirms the move, it doesn't time it. The numbers EMA21 & EMA50 above zero: edge -0.01 (no full-sample edge); 'climbing' filter -0.27; the later years +0.29 a period artifact · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_recession_timers.py; total-return SPY gate vs 3mo T-bill, edge vs buy-hold, the later years=2018+.
▸Donchian Channel Breakout (20/55-day) Richard Donchian / trend-following tradition	A classic breakout/trend system that beats the market	MARGINAL Buy the N-day high, sell the N-day low — the original trend-following channel. As an S&P long/cash timer (20- or 55-day), it cuts the worst drawdown (-55% -> -37%) but hands b…	C -0.05 edge vs. buy & hold
What it's meant to do A classic breakout/trend system that beats the market — Richard Donchian / trend-following tradition What we found Buy the N-day high, sell the N-day low — the original trend-following channel. As an S&P long/cash timer (20- or 55-day), it cuts the worst drawdown (-55% -> -37%) but hands back most of the return (+6-7%/yr vs +10.8% buy-and-hold) and shows no full-sample risk-adjusted edge. The flattering out-of-sample number sits on a negative in-sample — the data-mining tell. It was built for a DIVERSIFIED basket of FUTURES (leverage, shorting, many uncorrelated markets), where trend-following has a real record; a single mean-reverting equity index is the wrong instrument, and it shows. The numbers +6-7%/yr vs buy-hold +10.8%; worst drawdown -55%->-37%; edge -0.02 to -0.07 (the the later years bump sits on a negative in-sample) · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked trend_systems.py / swing_structure.py; total-return SPY long/cash, edge vs buy-hold, the later years=2014+, multiple sensitivities tested.
▸The Turtle System (Dennis & Eckhardt) Richard Dennis & William Eckhardt (1983)	A classic breakout/trend system that beats the market	MARGINAL The famous 1983 experiment that turned novices into traders — 20- and 55-day breakout entries, 10- and 20-day exits, ATR-sized with pyramiding and a 2N stop. Stripped to a lon…	C -0.07 edge vs. buy & hold
What it's meant to do A classic breakout/trend system that beats the market — Richard Dennis & William Eckhardt (1983) What we found The famous 1983 experiment that turned novices into traders — 20- and 55-day breakout entries, 10- and 20-day exits, ATR-sized with pyramiding and a 2N stop. Stripped to a long/cash S&P timer, both systems badly trail buy-and-hold (+4-5%/yr vs +10.8%) for only a modest drawdown cut, and the risk-adjusted edge is negative on the full sample. The Turtles traded a DIVERSIFIED basket of FUTURES with leverage and shorts — that, not single-index equity timing, is where the method earned its legend. We tested the part that maps to our bench; the futures version is a different, fairer test we flag but don’t claim here. The numbers System 1 (20/10) +5.0%/yr, System 2 (55/20) +4.0%/yr vs +10.8% buy-hold; edge -0.07 / -0.15 · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked trend_systems.py / swing_structure.py; total-return SPY long/cash, edge vs buy-hold, the later years=2014+, multiple sensitivities tested.
▸Higher Highs & Higher Lows (Dow Theory) Dow Theory / classic chart-reading	A classic breakout/trend system that beats the market	MARGINAL The textbook definition of a trend — hold while each swing high and swing low tops the last, step aside when it flips to lower-high + lower-low. We coded it with causal pivots…	C -0.12 edge vs. buy & hold
What it's meant to do A classic breakout/trend system that beats the market — Dow Theory / classic chart-reading What we found The textbook definition of a trend — hold while each swing high and swing low tops the last, step aside when it flips to lower-high + lower-low. We coded it with causal pivots (a swing is only confirmed a few bars after the peak, no peeking) at three sensitivities (5/10/20-bar swings). It cuts the worst drawdown (-55% -> -30%) but trails buy-and-hold badly on return (+5-7%/yr vs +10.8%) and has no risk-adjusted edge anywhere — negative edge in-sample AND out-of-sample, robust across every pivot setting. Waiting for structure to break costs you the rebounds; on a market that mostly goes up, that’s a losing trade. The numbers Long/cash on SPY: +5-7%/yr vs buy-hold +10.8%; drawdown -55%->-30%; edge -0.07 to -0.16, negative out-of-sample too · edge vs. buy & hold Performance C marginal — barely better than a coin flip Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked trend_systems.py / swing_structure.py; total-return SPY long/cash, edge vs buy-hold, the later years=2014+, multiple sensitivities tested.
▸NYSE Advance/Decline Line Classic market-breadth canon	NYSE Advance/Decline Line times market turns	OVERSOLD The single most-watched breadth gauge. It looks good in the recent test years (+0.71) -- but in the earlier years the very same rule HURT (its score was negative). A signal th…	D +0.71 edge vs. buy & hold
What it's meant to do NYSE Advance/Decline Line times market turns — Classic market-breadth canon What we found The single most-watched breadth gauge. It looks good in the recent test years (+0.71) -- but in the earlier years the very same rule HURT (its score was negative). A signal that helps in one stretch and hurts in another can't be trusted to help in the next; the recent win is most likely luck. Famous, not dependable. The numbers the later years edge +0.71 vs buy-hold; luck threshold ~+0.40 (clears it) · edge vs. buy & hold Performance D oversold — famous, underdelivers Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸VVIX<110 Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.02 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.92, worst drawdown -22%, in-market 81% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸MA100 Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.02 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.91, worst drawdown -18%, in-market 81% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸MOVE<100 Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.04 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.89, worst drawdown -19%, in-market 80% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸FedHike<50%(long) Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.04 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.89, worst drawdown -34%, in-market 79% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸NFCI_risk<0 Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.07 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.86, worst drawdown -32%, in-market 99% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸Curve 10Y2Y>0 Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.12 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.81, worst drawdown -34%, in-market 85% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸SKEW<145 Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.16 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.77, worst drawdown -34%, in-market 84% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸GoldenCross50>200 Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.17 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.77, worst drawdown -34%, in-market 85% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸MA50 Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.17 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.77, worst drawdown -20%, in-market 74% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸VRP+ (VIX>rvol20) Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.20 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.74, worst drawdown -35%, in-market 87% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸VIX<20 Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.21 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.72, worst drawdown -18%, in-market 74% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸HY_OAS<60dMA Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.22 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.71, worst drawdown -30%, in-market 63% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸HY_OAS<5% Various / folklore	Tells you when to be in the market vs. step aside	DOESN'T WORK Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal.	F -0.23 edge vs. buy & hold
What it's meant to do Tells you when to be in the market vs. step aside — Various / folklore What we found Costs you risk-adjusted return vs. just holding. The legend is bigger than the signal. The numbers Sharpe 0.70, worst drawdown -33%, in-market 77% of days · edge vs. buy & hold Performance F doesn't do what it claims Rigor D 2/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✗ out-of-sample · ✓ point-in-time · ✓ look-ahead checked —
▸Put/Call Ratio (All) CBOE / contrarian sentiment	A CBOE sentiment/vol reading that times the market	DOESN'T WORK The famous total put/call ratio is NOISE as a market-timing gate -- both directions tested, neither clears the noise floor. A beloved sentiment gauge that isn't an edge on its…	F -0.02 edge vs. buy & hold
What it's meant to do A CBOE sentiment/vol reading that times the market — CBOE / contrarian sentiment What we found The famous total put/call ratio is NOISE as a market-timing gate -- both directions tested, neither clears the noise floor. A beloved sentiment gauge that isn't an edge on its own. The numbers edge -0.02 (early-years -0.36 -> the later years +0.19); below the luck threshold (+0.23) · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cboe_gates.py; both directions x decile thresholds; shuffled-null floor +0.23 (95th pct) -- none clear it honestly.
▸Put/Call Ratio (Equity) CBOE / contrarian sentiment	A CBOE sentiment/vol reading that times the market	DOESN'T WORK Equity put/call ratio: same verdict as the total -- noise as a standalone timing gate.	F -0.04 edge vs. buy & hold
What it's meant to do A CBOE sentiment/vol reading that times the market — CBOE / contrarian sentiment What we found Equity put/call ratio: same verdict as the total -- noise as a standalone timing gate. The numbers edge -0.04 (early-years -0.36 -> the later years +0.16); below the floor · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cboe_gates.py; both directions x decile thresholds; shuffled-null floor +0.23 (95th pct) -- none clear it honestly.
▸Put/Call Ratio (Index) CBOE / contrarian sentiment	A CBOE sentiment/vol reading that times the market	DOESN'T WORK Index put/call ratio: noise. Index option flow is hedging-driven and doesn't time the market here.	F -0.08 edge vs. buy & hold
What it's meant to do A CBOE sentiment/vol reading that times the market — CBOE / contrarian sentiment What we found Index put/call ratio: noise. Index option flow is hedging-driven and doesn't time the market here. The numbers edge -0.08 (early-years -0.26 -> the later years +0.02); below the floor · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cboe_gates.py; both directions x decile thresholds; shuffled-null floor +0.23 (95th pct) -- none clear it honestly.
▸VXN (Nasdaq Implied Vol) CBOE	A CBOE sentiment/vol reading that times the market	DOESN'T WORK The ONLY series that nominally cleared the floor -- but on an out-of-sample-flattering threshold while its full-window and in-sample scores are negative. An the later years mirage, exactl…	F -0.17 edge vs. buy & hold
What it's meant to do A CBOE sentiment/vol reading that times the market — CBOE What we found The ONLY series that nominally cleared the floor -- but on an out-of-sample-flattering threshold while its full-window and in-sample scores are negative. An the later years mirage, exactly what the null floor exists to flag. Noise. The numbers edge -0.17; the the later years +0.33 is a cherry-picked threshold while full & in-sample are negative · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cboe_gates.py; both directions x decile thresholds; shuffled-null floor +0.23 (95th pct) -- none clear it honestly.
▸VXTH (Tail-Hedge Index) CBOE	A CBOE sentiment/vol reading that times the market	DOESN'T WORK The 'black-swan' tail-hedge index. As a gate it actively hurts -- a hedging-cost index isn't a timer.	F -0.23 edge vs. buy & hold
What it's meant to do A CBOE sentiment/vol reading that times the market — CBOE What we found The 'black-swan' tail-hedge index. As a gate it actively hurts -- a hedging-cost index isn't a timer. The numbers edge -0.23 (early-years -0.65 -> the later years +0.00); worst of the set · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cboe_gates.py; both directions x decile thresholds; shuffled-null floor +0.23 (95th pct) -- none clear it honestly.
▸RVX (Russell Implied Vol) CBOE	A CBOE sentiment/vol reading that times the market	DOESN'T WORK Small-cap implied vol level: noise, like every other vol-level gate. Breadth is the outlier that works.	F -0.04 edge vs. buy & hold
What it's meant to do A CBOE sentiment/vol reading that times the market — CBOE What we found Small-cap implied vol level: noise, like every other vol-level gate. Breadth is the outlier that works. The numbers edge -0.04 (early-years -0.03 -> the later years -0.04); flat-to-negative · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked cboe_gates.py; both directions x decile thresholds; shuffled-null floor +0.23 (95th pct) -- none clear it honestly.
▸Net Liquidity (Fed B/S − RRP) Macro / FinTwit thesis	A macro / intermarket reading that times the market	DOESN'T WORK The famous 'net liquidity drives the S&P' thesis is a DUD as a daily timing gate -- noise both directions. (Proxy = WALCL minus reverse-repo, missing the TGA leg we don't have…	F -0.12 edge vs. buy & hold
What it's meant to do A macro / intermarket reading that times the market — Macro / FinTwit thesis What we found The famous 'net liquidity drives the S&P' thesis is a DUD as a daily timing gate -- noise both directions. (Proxy = WALCL minus reverse-repo, missing the TGA leg we don't have; the classic form is weekly. But the popular daily version people trade on doesn't hold up.) The numbers edge -0.12 as a daily SPY gate; below the luck threshold (+0.33) · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked unused_data_sweep.py; 36 series, both directions x deciles, shuffled-null floor +0.33 (95th) -- none cleared honestly.
▸US Dollar (DTWEXBGS) Macro / intermarket	A macro / intermarket reading that times the market	DOESN'T WORK A strong/weak dollar as an equity-timing gate is noise -- neither the level nor its momentum beats the floor. The dollar matters for the economy; it doesn't time the index her…	F -0.08 edge vs. buy & hold
What it's meant to do A macro / intermarket reading that times the market — Macro / intermarket What we found A strong/weak dollar as an equity-timing gate is noise -- neither the level nor its momentum beats the floor. The dollar matters for the economy; it doesn't time the index here. The numbers edge -0.08 (level) / -0.23 (13-week change); noise · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked unused_data_sweep.py; 36 series, both directions x deciles, shuffled-null floor +0.33 (95th) -- none cleared honestly.
▸Recession Models (Sahm / claims / leading) Macro / recession trackers	A macro / intermarket reading that times the market	DOESN'T WORK Recession indicators (Sahm rule, jobless-claims trend, OECD leading) are slow by design and noise as equity-timing gates. The Sahm rule nominally cleared the floor but on a ne…	F +0.07 edge vs. buy & hold
What it's meant to do A macro / intermarket reading that times the market — Macro / recession trackers What we found Recession indicators (Sahm rule, jobless-claims trend, OECD leading) are slow by design and noise as equity-timing gates. The Sahm rule nominally cleared the floor but on a negative in-sample = a data-mining artifact the discipline flags, not an edge. The numbers Sahm rule edge +0.07 (early-years -0.35 -> the later years +0.35 = an the later years mirage); claims/leading all noise · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked unused_data_sweep.py; 36 series, both directions x deciles, shuffled-null floor +0.33 (95th) -- none cleared honestly.
▸Intermarket Ratios (8-pair basket) Intermarket / sector-rotation	A macro / intermarket reading that times the market	DOESN'T WORK Eight intermarket ratios (defensives/cyclicals, gold/stocks, bonds/stocks, gold-miners, semis, biotech, oil) tested as SPY gates -- all noise. The flashiest (semis momentum, O…	F +0.23 edge vs. buy & hold
What it's meant to do A macro / intermarket reading that times the market — Intermarket / sector-rotation What we found Eight intermarket ratios (defensives/cyclicals, gold/stocks, bonds/stocks, gold-miners, semis, biotech, oil) tested as SPY gates -- all noise. The flashiest (semis momentum, the later years +0.56) is an the later years mirage with a negative in-sample. May still work for SECTOR rotation; doesn't time the index. The numbers best (Semis/SPY 63d-mom) the later years +0.56 but early-years -0.33 = mirage; all 8 ratios noise as gates · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked unused_data_sweep.py; 36 series, both directions x deciles, shuffled-null floor +0.33 (95th) -- none cleared honestly.
▸Excess CAPE Yield (as a timing signal) Valuation timers	Excess CAPE Yield tells you when to be in or out of the market	DOESN'T WORK The single best long-run return FORECASTER on the bench -- the Excess CAPE Yield predicts the forward DECADE at +0.60 correlation. But as an in/out timing signal it is noise, …	F edge as an in/out timing gate
What it's meant to do Excess CAPE Yield tells you when to be in or out of the market — Valuation timers What we found The single best long-run return FORECASTER on the bench -- the Excess CAPE Yield predicts the forward DECADE at +0.60 correlation. But as an in/out timing signal it is noise, in the modern era and across 100 years of Shiller data. Valuation tells you what to EXPECT, never WHEN. A market can stay expensive for a decade; going to cash because the premium is thin means sitting out the bull. The numbers +0.06 in/out edge (vs +0.41 floor); noise in both eras · edge as an in/out timing gate Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked index_timing_sweep.py; each index as a pull-to-cash gate, best threshold/direction, the later years edge vs a shuffle floor. Forecasting != timing.
▸CAPE / P-E10 (as a timing signal) Robert Shiller / valuation crowd	CAPE / P-E10 tells you when to be in or out of the market	DOESN'T WORK High CAPE genuinely predicts low long-run returns -- and genuinely cannot time entries and exits. Zero edge as a buy/sell gate over a century. The most expensive mistake in ma…	F edge as an in/out timing gate
What it's meant to do CAPE / P-E10 tells you when to be in or out of the market — Robert Shiller / valuation crowd What we found High CAPE genuinely predicts low long-run returns -- and genuinely cannot time entries and exits. Zero edge as a buy/sell gate over a century. The most expensive mistake in markets is confusing 'expensive' with 'about to fall.' They are not the same thing. The numbers +0.08 in/out edge (vs floor); noise over 100 years · edge as an in/out timing gate Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked index_timing_sweep.py; each index as a pull-to-cash gate, best threshold/direction, the later years edge vs a shuffle floor. Forecasting != timing.
▸Equity Risk Premium / Fed model (as a timing signal) The 'Fed model' crowd	Equity Risk Premium / Fed model tells you when to be in or out of the market	DOESN'T WORK Earnings yield minus the bond yield is a fine long-horizon gauge of whether stocks are cheap vs bonds. As a market-timing trigger it's worthless -- it flickers for years befor…	F edge as an in/out timing gate
What it's meant to do Equity Risk Premium / Fed model tells you when to be in or out of the market — The 'Fed model' crowd What we found Earnings yield minus the bond yield is a fine long-horizon gauge of whether stocks are cheap vs bonds. As a market-timing trigger it's worthless -- it flickers for years before anything happens. Forecast, not a timer. The numbers +0.00 in/out edge; noise · edge as an in/out timing gate Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked index_timing_sweep.py; each index as a pull-to-cash gate, best threshold/direction, the later years edge vs a shuffle floor. Forecasting != timing.
▸Risk Ratios (SPY:TLT + 5 sector RS) StockCharts / sector-rotation chartists	A chart-based signal that times the market	DOESN'T WORK The popular 'Risk On / Risk Off' ratio panel -- SPY:TLT plus five sector relative-strength ratios -- run as 50-day-EMA crossover gates on SPY. The famous SPY:TLT is the WORST:…	F +0.05 edge vs. buy & hold
What it's meant to do A chart-based signal that times the market — StockCharts / sector-rotation chartists What we found The popular 'Risk On / Risk Off' ratio panel -- SPY:TLT plus five sector relative-strength ratios -- run as 50-day-EMA crossover gates on SPY. The famous SPY:TLT is the WORST: as a timer it underperforms buy-and-hold (edge -0.16, -0.39 out-of-sample). The sector gates (XLY, XLI) flatter in the rear-view (+0.13, +0.18) and fade to roughly zero out-of-sample. The >=4-of-6 composite cuts drawdown (-55%->-22%) but only matches buy-and-hold on risk-adjusted terms (+0.05 the later years) while handing back half the return. An intuitive panel with no out-of-sample timing edge. The numbers SPY:TLT actively hurts (the later years -0.39); sector RS gates +0.13/+0.18 in-sample fade to ~0 the later years; composite +0.05 · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked chartist_timers.py; total-return SPY gates vs 3mo T-bill, edge vs buy-hold, the later years=2014+. The slow trend filter beats both as overlays too.
▸Recession Monitor (3M/10Y + 2Y/10Y inversion) Yield-curve / recession watchers	A chart-based signal that times the market	DOESN'T WORK The Fed-staple yield-curve signal -- 3-month/10-year AND 2-year/10-year both inverted -- tested as an equity gate. As a market timer it UNDERPERFORMS buy-and-hold in every ver…	F -0.09 edge vs. buy & hold
What it's meant to do A chart-based signal that times the market — Yield-curve / recession watchers What we found The Fed-staple yield-curve signal -- 3-month/10-year AND 2-year/10-year both inverted -- tested as an equity gate. As a market timer it UNDERPERFORMS buy-and-hold in every version (out-while-inverted -0.09, the ~14-month recession window -0.12), worse out-of-sample. The one inversion in our window (2022) preceded NO recession and a rising market -- the 'nine of the last two recessions' problem. It is a recession FORECASTER, not an equity timer; the forecasting claim itself needs longer history and NBER dates to test fairly (our spread copy starts 2010 -- a single inversion). The numbers Both-inverted as an equity gate HURTS: edge -0.09 (-0.17 the later years); the 14-month recession window worse (-0.12/-0.25) · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_recession_timers.py; total-return SPY gate vs 3mo T-bill, edge vs buy-hold, the later years=2018+.
▸McClellan Oscillator Sherman & Marian McClellan / breadth tradition	Times the market via breadth momentum (19/39-day EMA of advances minus declines)	DOESN'T WORK The famous McClellan Oscillator LOSES to buy-and-hold as a risk gate and collapses out-of-sample -- a fast, jumpy reading that whipsaws. Our Risk Reducer beats it decisively (…	F -0.31 edge vs. buy & hold
What it's meant to do Times the market via breadth momentum (19/39-day EMA of advances minus declines) — Sherman & Marian McClellan / breadth tradition What we found The famous McClellan Oscillator LOSES to buy-and-hold as a risk gate and collapses out-of-sample -- a fast, jumpy reading that whipsaws. Our Risk Reducer beats it decisively (+0.48 vs -0.31). The numbers edge -0.31 (early-years +0.02 -> the later years -0.53, collapses out-of-sample) · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked McClellan-vs-Risk-Reducer bake-off; EMA19-EMA39 of real $ADSPD; null floor +0.03, not cleared.
▸McClellan Summation Index McClellan / breadth tradition	Marks bull/bear regimes via the cumulative McClellan Oscillator	DOESN'T WORK Every anchor-free rule on the Summation Index (rising, versus rolling-mean, level) scores negative and fails the noise floor. A famous indicator, a losing gate.	F -0.25 edge vs. buy & hold
What it's meant to do Marks bull/bear regimes via the cumulative McClellan Oscillator — McClellan / breadth tradition What we found Every anchor-free rule on the Summation Index (rising, versus rolling-mean, level) scores negative and fails the noise floor. A famous indicator, a losing gate. The numbers edge -0.25 to -0.42 across every rule (rising / vs-mean / level); all noise · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked McClellan-vs-Risk-Reducer bake-off; none of the variants cleared the +0.03 luck threshold.
▸Cross-Sectional Dispersion Dispersion / correlation tradition	Low cross-sectional return dispersion (stocks moving together calmly) signals risk-on	DOESN'T WORK As a simple low-dispersion = risk-on equity gate, cross-sectional return dispersion is noise (below the noise floor). NOTE: this is ONE construction -- IMPLIED dispersion and …	F +0.04 edge vs. buy & hold
What it's meant to do Low cross-sectional return dispersion (stocks moving together calmly) signals risk-on — Dispersion / correlation tradition What we found As a simple low-dispersion = risk-on equity gate, cross-sectional return dispersion is noise (below the noise floor). NOTE: this is ONE construction -- IMPLIED dispersion and realized-correlation regime are different signals with different uses (vol trading), and are not tested here. So this is 'return-dispersion-as-equity-gate fails,' not 'dispersion fails.' The numbers edge -0.25 (early-years -0.80 -> the later years +0.04); fails the luck threshold as a standalone equity gate · edge vs. buy & hold Performance F doesn't do what it claims Rigor C 3/5 honesty checks passed How honestly we tested it – survivorship-free · ✗ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_constructions.py; cross-sectional return dispersion from the survivorship-free panel.
▸Advance/Decline Ratio Classic market-breadth canon	Advance/Decline Ratio times market turns	DOESN'T WORK Advancers vs decliners, the textbook breadth ratio. the later years +0.28 -- below the ~+0.40 luck threshold. Doesn't beat buy-hold once you account for luck.	F +0.28 edge vs. buy & hold
What it's meant to do Advance/Decline Ratio times market turns — Classic market-breadth canon What we found Advancers vs decliners, the textbook breadth ratio. the later years +0.28 -- below the ~+0.40 luck threshold. Doesn't beat buy-hold once you account for luck. The numbers the later years edge +0.28 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸Percent of Stocks Above the 200-Day Average Classic market-breadth canon	Percent of Stocks Above the 200-Day Average times market turns	DOESN'T WORK A breadth-nerd favorite. the later years +0.18, under the floor -- it looks meaningful on a chart but adds nothing a coin-flip wouldn't.	F +0.18 edge vs. buy & hold
What it's meant to do Percent of Stocks Above the 200-Day Average times market turns — Classic market-breadth canon What we found A breadth-nerd favorite. the later years +0.18, under the floor -- it looks meaningful on a chart but adds nothing a coin-flip wouldn't. The numbers the later years edge +0.18 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸Percent of Stocks Above the 50-Day Average Classic market-breadth canon	Percent of Stocks Above the 50-Day Average times market turns	DOESN'T WORK The faster cousin. the later years +0.09 -- noise. Whatever it's telling you, the price already said it.	F +0.09 edge vs. buy & hold
What it's meant to do Percent of Stocks Above the 50-Day Average times market turns — Classic market-breadth canon What we found The faster cousin. the later years +0.09 -- noise. Whatever it's telling you, the price already said it. The numbers the later years edge +0.09 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸The High-Low Index (highs minus lows) Classic market-breadth canon	The High-Low Index (highs minus lows) times market turns	DOESN'T WORK The most popular high-low construction. the later years +0.01 -- essentially zero edge. Subtracting the upside extreme from the downside one dilutes the half that actually carries the dan…	F +0.01 edge vs. buy & hold
What it's meant to do The High-Low Index (highs minus lows) times market turns — Classic market-breadth canon What we found The most popular high-low construction. the later years +0.01 -- essentially zero edge. Subtracting the upside extreme from the downside one dilutes the half that actually carries the danger. The numbers the later years edge +0.01 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸Up/Down (Advancing) Volume Classic market-breadth canon	Up/Down (Advancing) Volume times market turns	DOESN'T WORK Volume flowing into advancers vs decliners. the later years +0.10 -- below the floor. A confirming indicator that doesn't lead.	F +0.10 edge vs. buy & hold
What it's meant to do Up/Down (Advancing) Volume times market turns — Classic market-breadth canon What we found Volume flowing into advancers vs decliners. the later years +0.10 -- below the floor. A confirming indicator that doesn't lead. The numbers the later years edge +0.10 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸Zweig Breadth Thrust Classic market-breadth canon	Zweig Breadth Thrust times market turns	DOESN'T WORK The famous rare 'thrust' buy signal. the later years +0.13 -- it fires too rarely and too late to clear the noise floor as a systematic gate.	F +0.13 edge vs. buy & hold
What it's meant to do Zweig Breadth Thrust times market turns — Classic market-breadth canon What we found The famous rare 'thrust' buy signal. the later years +0.13 -- it fires too rarely and too late to clear the noise floor as a systematic gate. The numbers the later years edge +0.13 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸Equal-Weight vs Cap-Weight Divergence Classic market-breadth canon	Equal-Weight vs Cap-Weight Divergence times market turns	DOESN'T WORK When the average stock diverges from the cap-weighted index. the later years +0.41 -- it pokes at the noise floor but doesn't clear it once you control for data-mining. A good story, not …	F +0.41 edge vs. buy & hold
What it's meant to do Equal-Weight vs Cap-Weight Divergence times market turns — Classic market-breadth canon What we found When the average stock diverges from the cap-weighted index. the later years +0.41 -- it pokes at the noise floor but doesn't clear it once you control for data-mining. A good story, not a tradeable gate. The numbers the later years edge +0.41 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸The Arms Index (TRIN) Classic market-breadth canon	The Arms Index (TRIN) times market turns	DOESN'T WORK The classic 'is volume rushing into rising or falling stocks' panic gauge. the later years +0.16 -- noise. A crowd-mood meter that doesn't actually time anything.	F +0.16 edge vs. buy & hold
What it's meant to do The Arms Index (TRIN) times market turns — Classic market-breadth canon What we found The classic 'is volume rushing into rising or falling stocks' panic gauge. the later years +0.16 -- noise. A crowd-mood meter that doesn't actually time anything. The numbers the later years edge +0.16 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸Percent of Stocks Above the 20-Day Average Classic market-breadth canon	Percent of Stocks Above the 20-Day Average times market turns	DOESN'T WORK The short-term cousin of the percent-above-MA family. the later years -0.03 -- the worst of the lot, actively negative. Fast breadth is mostly fast noise.	F -0.03 edge vs. buy & hold
What it's meant to do Percent of Stocks Above the 20-Day Average times market turns — Classic market-breadth canon What we found The short-term cousin of the percent-above-MA family. the later years -0.03 -- the worst of the lot, actively negative. Fast breadth is mostly fast noise. The numbers the later years edge -0.03 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸Small-Cap Breadth (Russell % above 50-Day) Classic market-breadth canon	Small-Cap Breadth (Russell % above 50-Day) times market turns	DOESN'T WORK Small-cap participation -- 'the generals vs the troops.' the later years +0.20 -- noise. Makes a great narrative and a useless signal.	F +0.20 edge vs. buy & hold
What it's meant to do Small-Cap Breadth (Russell % above 50-Day) times market turns — Classic market-breadth canon What we found Small-cap participation -- 'the generals vs the troops.' the later years +0.20 -- noise. Makes a great narrative and a useless signal. The numbers the later years edge +0.20 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸Nasdaq Advance/Decline Classic market-breadth canon	Nasdaq Advance/Decline times market turns	DOESN'T WORK Advancers minus decliners for the growth side of the tape. the later years +0.14 -- noise, same as its NYSE sibling.	F +0.14 edge vs. buy & hold
What it's meant to do Nasdaq Advance/Decline times market turns — Classic market-breadth canon What we found Advancers minus decliners for the growth side of the tape. the later years +0.14 -- noise, same as its NYSE sibling. The numbers the later years edge +0.14 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸The Bullish Percent Index (Point & Figure) Classic market-breadth canon	The Bullish Percent Index (Point & Figure) times market turns	DOESN'T WORK The chartists' favorite breadth gauge -- the percent of stocks on a P&F buy signal -- which we built ourselves from the whole universe (no scraping). the later years +0.10, noise. Even th…	F +0.10 edge vs. buy & hold
What it's meant to do The Bullish Percent Index (Point & Figure) times market turns — Classic market-breadth canon What we found The chartists' favorite breadth gauge -- the percent of stocks on a P&F buy signal -- which we built ourselves from the whole universe (no scraping). the later years +0.10, noise. Even this one doesn't time the market once you control for luck. The numbers the later years edge +0.10 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
▸The Hindenburg Omen Classic market-breadth canon	The Hindenburg Omen times market turns	DOESN'T WORK The famous 'crash' signal -- it fires when too many stocks hit fresh extremes in BOTH directions at once. the later years +0.16 -- noise. It has 'predicted' far more crashes than have eve…	F +0.16 edge vs. buy & hold
What it's meant to do The Hindenburg Omen times market turns — Classic market-breadth canon What we found The famous 'crash' signal -- it fires when too many stocks hit fresh extremes in BOTH directions at once. the later years +0.16 -- noise. It has 'predicted' far more crashes than have ever happened. The numbers the later years edge +0.16 vs buy-hold; luck threshold ~+0.40 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked breadth_sweep.py; the later years edge vs a shuffle null floor.
Technical Indicators — the TA Canon, Tested · 22 tested
Tested as S&P 500 timing signals. An indicator that's noise on the index can behave differently on another asset (oil, bonds, a single stock) — which is exactly what our weekly tests explore. We scope every claim to what we actually tested.
▸Rate of Change (12) Technical-analysis canon	Rate of Change (12) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Rate of Change (12) scored +0.17 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and ad…	F +0.17 edge vs. buy & hold
What it's meant to do Rate of Change (12) signals when to be in or out of the market — Technical-analysis canon What we found As a mechanical buy/sell timing signal on the index, Rate of Change (12) scored +0.17 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and adds nothing a coin-flip wouldn't. The numbers the later years edge +0.17 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸On-Balance Volume (slope) Joe Granville	On-Balance Volume (slope) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, On-Balance Volume (slope) scored +0.16 out-of-sample -- under the +0.35 luck threshold. A running volume tally. Confirm…	F +0.16 edge vs. buy & hold
What it's meant to do On-Balance Volume (slope) signals when to be in or out of the market — Joe Granville What we found As a mechanical buy/sell timing signal on the index, On-Balance Volume (slope) scored +0.16 out-of-sample -- under the +0.35 luck threshold. A running volume tally. Confirms moves; never leads them. The numbers the later years edge +0.16 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸CCI (20) Donald Lambert	CCI (20) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, CCI (20) scored +0.14 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and adds nothing …	F +0.14 edge vs. buy & hold
What it's meant to do CCI (20) signals when to be in or out of the market — Donald Lambert What we found As a mechanical buy/sell timing signal on the index, CCI (20) scored +0.14 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and adds nothing a coin-flip wouldn't. The numbers the later years edge +0.14 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Price vs 50-day MA Technical-analysis canon	Price vs 50-day MA signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Price vs 50-day MA scored +0.10 out-of-sample -- under the +0.35 luck threshold. The trend filter everyone uses. As a r…	F +0.10 edge vs. buy & hold
What it's meant to do Price vs 50-day MA signals when to be in or out of the market — Technical-analysis canon What we found As a mechanical buy/sell timing signal on the index, Price vs 50-day MA scored +0.10 out-of-sample -- under the +0.35 luck threshold. The trend filter everyone uses. As a return signal: nothing. The numbers the later years edge +0.10 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸MACD histogram Gerald Appel	MACD histogram signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, MACD histogram scored +0.08 out-of-sample -- under the +0.35 luck threshold. The derivative of a lagging signal. Faster…	F +0.08 edge vs. buy & hold
What it's meant to do MACD histogram signals when to be in or out of the market — Gerald Appel What we found As a mechanical buy/sell timing signal on the index, MACD histogram scored +0.08 out-of-sample -- under the +0.35 luck threshold. The derivative of a lagging signal. Faster, and just as noisy. The numbers the later years edge +0.08 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸True Strength Index Technical-analysis canon	True Strength Index signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, True Strength Index scored +0.05 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and ad…	F +0.05 edge vs. buy & hold
What it's meant to do True Strength Index signals when to be in or out of the market — Technical-analysis canon What we found As a mechanical buy/sell timing signal on the index, True Strength Index scored +0.05 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and adds nothing a coin-flip wouldn't. The numbers the later years edge +0.05 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Coppock Curve E.S.C. Coppock	Coppock Curve signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Coppock Curve scored +0.05 out-of-sample -- under the +0.35 luck threshold. A weighted average of two rate-of-change li…	F +0.05 edge vs. buy & hold
What it's meant to do Coppock Curve signals when to be in or out of the market — E.S.C. Coppock What we found As a mechanical buy/sell timing signal on the index, Coppock Curve scored +0.05 out-of-sample -- under the +0.35 luck threshold. A weighted average of two rate-of-change lines. Smooth, and smoothly useless. The numbers the later years edge +0.05 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Awesome Oscillator Bill Williams	Awesome Oscillator signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Awesome Oscillator scored +0.05 out-of-sample -- under the +0.35 luck threshold. Two moving averages of the midpoint. A…	F +0.05 edge vs. buy & hold
What it's meant to do Awesome Oscillator signals when to be in or out of the market — Bill Williams What we found As a mechanical buy/sell timing signal on the index, Awesome Oscillator scored +0.05 out-of-sample -- under the +0.35 luck threshold. Two moving averages of the midpoint. As awesome as that sounds. The numbers the later years edge +0.05 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Detrended Price Osc (20) Technical-analysis canon	Detrended Price Osc (20) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Detrended Price Osc (20) scored +0.04 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart a…	F +0.04 edge vs. buy & hold
What it's meant to do Detrended Price Osc (20) signals when to be in or out of the market — Technical-analysis canon What we found As a mechanical buy/sell timing signal on the index, Detrended Price Osc (20) scored +0.04 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and adds nothing a coin-flip wouldn't. The numbers the later years edge +0.04 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸MACD line Gerald Appel	MACD line signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, MACD line scored +0.03 out-of-sample -- under the +0.35 luck threshold. A lagging crossover of two lagging averages -- …	F +0.03 edge vs. buy & hold
What it's meant to do MACD line signals when to be in or out of the market — Gerald Appel What we found As a mechanical buy/sell timing signal on the index, MACD line scored +0.03 out-of-sample -- under the +0.35 luck threshold. A lagging crossover of two lagging averages -- it calls the move once it's already happened. The numbers the later years edge +0.03 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Price vs 200-day MA Technical-analysis canon	Price vs 200-day MA signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Price vs 200-day MA scored +0.02 out-of-sample -- under the +0.35 luck threshold. The most-watched line in markets. It …	F +0.02 edge vs. buy & hold
What it's meant to do Price vs 200-day MA signals when to be in or out of the market — Technical-analysis canon What we found As a mechanical buy/sell timing signal on the index, Price vs 200-day MA scored +0.02 out-of-sample -- under the +0.35 luck threshold. The most-watched line in markets. It marks regime; it doesn't time it. The numbers the later years edge +0.02 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Bollinger %B (20,2) John Bollinger	Bollinger %B (20,2) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Bollinger %B (20,2) scored +0.02 out-of-sample -- under the +0.35 luck threshold. Bands measure volatility; they say no…	F +0.02 edge vs. buy & hold
What it's meant to do Bollinger %B (20,2) signals when to be in or out of the market — John Bollinger What we found As a mechanical buy/sell timing signal on the index, Bollinger %B (20,2) scored +0.02 out-of-sample -- under the +0.35 luck threshold. Bands measure volatility; they say nothing about which way price goes next. The numbers the later years edge +0.02 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Accum/Dist (slope) Technical-analysis canon	Accum/Dist (slope) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Accum/Dist (slope) scored +0.01 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and add…	F +0.01 edge vs. buy & hold
What it's meant to do Accum/Dist (slope) signals when to be in or out of the market — Technical-analysis canon What we found As a mechanical buy/sell timing signal on the index, Accum/Dist (slope) scored +0.01 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and adds nothing a coin-flip wouldn't. The numbers the later years edge +0.01 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Chaikin Money Flow (20) Marc Chaikin	Chaikin Money Flow (20) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Chaikin Money Flow (20) scored +0.01 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart an…	F +0.01 edge vs. buy & hold
What it's meant to do Chaikin Money Flow (20) signals when to be in or out of the market — Marc Chaikin What we found As a mechanical buy/sell timing signal on the index, Chaikin Money Flow (20) scored +0.01 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and adds nothing a coin-flip wouldn't. The numbers the later years edge +0.01 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸RSI (14) J. Welles Wilder	RSI (14) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, RSI (14) scored +0.01 out-of-sample -- under the +0.35 luck threshold. Overbought/oversold is a story the chart tells a…	F +0.01 edge vs. buy & hold
What it's meant to do RSI (14) signals when to be in or out of the market — J. Welles Wilder What we found As a mechanical buy/sell timing signal on the index, RSI (14) scored +0.01 out-of-sample -- under the +0.35 luck threshold. Overbought/oversold is a story the chart tells after the move, not before. The numbers the later years edge +0.01 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Stochastic %K (14) George Lane	Stochastic %K (14) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Stochastic %K (14) scored -0.01 out-of-sample -- under the +0.35 luck threshold. Where price sits in its recent range -…	F -0.01 edge vs. buy & hold
What it's meant to do Stochastic %K (14) signals when to be in or out of the market — George Lane What we found As a mechanical buy/sell timing signal on the index, Stochastic %K (14) scored -0.01 out-of-sample -- under the +0.35 luck threshold. Where price sits in its recent range -- and the range has no memory. The numbers the later years edge -0.01 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Williams %R (14) Larry Williams	Williams %R (14) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Williams %R (14) scored -0.01 out-of-sample -- under the +0.35 luck threshold. Stochastic with a minus sign. Same non-e…	F -0.01 edge vs. buy & hold
What it's meant to do Williams %R (14) signals when to be in or out of the market — Larry Williams What we found As a mechanical buy/sell timing signal on the index, Williams %R (14) scored -0.01 out-of-sample -- under the +0.35 luck threshold. Stochastic with a minus sign. Same non-edge. The numbers the later years edge -0.01 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Vortex (14) Botes & Siepman	Vortex (14) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Vortex (14) scored -0.01 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and adds nothi…	F -0.01 edge vs. buy & hold
What it's meant to do Vortex (14) signals when to be in or out of the market — Botes & Siepman What we found As a mechanical buy/sell timing signal on the index, Vortex (14) scored -0.01 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and adds nothing a coin-flip wouldn't. The numbers the later years edge -0.01 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Money Flow Index (14) Technical-analysis canon	Money Flow Index (14) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Money Flow Index (14) scored -0.03 out-of-sample -- under the +0.35 luck threshold. Volume-weighted RSI. The volume doe…	F -0.03 edge vs. buy & hold
What it's meant to do Money Flow Index (14) signals when to be in or out of the market — Technical-analysis canon What we found As a mechanical buy/sell timing signal on the index, Money Flow Index (14) scored -0.03 out-of-sample -- under the +0.35 luck threshold. Volume-weighted RSI. The volume doesn't rescue it. The numbers the later years edge -0.03 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸ADX (14) J. Welles Wilder	ADX (14) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, ADX (14) scored -0.04 out-of-sample -- under the +0.35 luck threshold. It measures trend STRENGTH, not direction -- it …	F -0.04 edge vs. buy & hold
What it's meant to do ADX (14) signals when to be in or out of the market — J. Welles Wilder What we found As a mechanical buy/sell timing signal on the index, ADX (14) scored -0.04 out-of-sample -- under the +0.35 luck threshold. It measures trend STRENGTH, not direction -- it times nothing on its own. The numbers the later years edge -0.04 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸Stochastic RSI (14) Technical-analysis canon	Stochastic RSI (14) signals when to be in or out of the market	DOESN'T WORK As a mechanical buy/sell timing signal on the index, Stochastic RSI (14) scored -0.06 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and ad…	F -0.06 edge vs. buy & hold
What it's meant to do Stochastic RSI (14) signals when to be in or out of the market — Technical-analysis canon What we found As a mechanical buy/sell timing signal on the index, Stochastic RSI (14) scored -0.06 out-of-sample -- under the +0.35 luck threshold. It looks meaningful on a chart and adds nothing a coin-flip wouldn't. The numbers the later years edge -0.06 vs buy-hold; luck threshold +0.35 (below it = noise) · edge vs. buy & hold Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep.py; best threshold/direction per indicator, the later years edge vs a shuffle null floor. Tested as index-timing signals.
▸The Entire TA Library (pandas-ta, 257 indicators) Technical-analysis canon	Surely SOME technical indicator can time the market	DOESN'T WORK We ran EVERY indicator in the pandas-ta library -- all 257 -- through the same test. Zero provide a real market-timing edge. The one that appeared to (the Detrended Price Osci…	F indicators that actually worked
What it's meant to do Surely SOME technical indicator can time the market — Technical-analysis canon What we found We ran EVERY indicator in the pandas-ta library -- all 257 -- through the same test. Zero provide a real market-timing edge. The one that appeared to (the Detrended Price Oscillator, +2.84) is a LOOKAHEAD ARTIFACT: the standard DPO centers its moving average and peeks at future prices. Built point-in-time, with no peeking, it's noise like all the rest. The one exception we caught is the rule -- and a demonstration of why we check for look-ahead. The numbers 257 indicators tested; 0 provide a real timing edge (the lone apparent winner is a lookahead artifact) · indicators that actually worked Performance F doesn't do what it claims Rigor A 5/5 honesty checks passed How honestly we tested it ✓ survivorship-free · ✓ net of real cost · ✓ out-of-sample · ✓ point-in-time · ✓ look-ahead checked indicator_sweep_full.py; full pandas-ta-classic library, best threshold/direction per indicator, the later years edge vs a shuffle null floor.

🛡️ The Risk Reducer Effect

Our flagship signal isn't only a market-timing gate — it's a risk overlay that bolts onto any premium strategy. Below: every trade above, run ungated vs. gated by the Risk Reducer (step aside when it flags caution), on real cross-the-spread fills. The worst trade and the max drawdown shrink across the board.

strategy	worst trade			max drawdown
	ungated	gated	cut	ungated	gated	cut
Big Lizard	-6929	-2121	−69%	-14659	-4000	−73%
Short Straddle	-6550	-2097	−68%	-13286	-6551	−51%
Short Put	-4930	-790	−84%	-7368	-1169	−84%
Short Strangle	-4859	-1182	−76%	-9222	-3291	−64%
Jade Lizard	-2530	-2316	−8%	-5228	-5895	−-13%
Iron Fly	-1091	-655	−40%	-17436	-8266	−53%
Iron Condor	-677	-677	−0%	-6797	-4408	−35%

Same strategies, same fills — the only change is standing aside when the Risk Reducer says Not Worth The Risk. It slashes the tail exactly where the danger lives — the naked-risk sells (short put −84%, strangle −81%). On already-defined-risk trades it's redundant, as it should be. It helps most where the danger is worst. Members see the live read the day it changes.

💲 The Liquidity Tax

We ran the same option strategies on the three most liquid ETF option markets on Earth, crossing the real bid/ask each time. Watch the edge melt as the spread widens (half-spread on traded strikes: SPY 0.8% → QQQ 1.3% → IWM 1.7%). Each number is expectancy per trade after crossing the spread; the small figure is how much of the patient-mid edge survived.

strategy	SPY0.8%	QQQ1.3%	IWM1.7%
Big Lizard	+10773%	+5252%	+2648%
Short Straddle	+6260%	+2741%	+1335%
Short Put	+4894%	+2886%	+651%
Short Strangle	+3179%	+1254%	+224%
Jade Lizard	+2766%	+01%	+426%
Iron Condor	-23gone	-38gone	-15-732%
Iron Fly	-64gone	-62-664%	-9-33%

The short put nets +$48 on SPY and +$6 on IWM — and IWM is still a top-5 options market. A single-stock weekly runs 5–10× SPY's spread. The edge lives in the most liquid market on the planet and evaporates as you walk away from it.

💀 The Fund Graveyard

The part no fund marketer will show you: 97 thesis ETFs that were liquidated and deleted — about 1 in 5 of every theme fund ever launched. Every "best thematic ETFs" list on the internet structurally hides these, because dead funds vanish from the database. We keep them. Here are the 20 worst burials (total return from launch to the day they were shut down):

VANECK GREEN INFRASTRUCTURE ETFRNEW	Infrastructure	2.7 yrs	-99%
GLOBAL X CANNABIS ETFPOTX	Cannabis	4.4 yrs	-96%
AXS CANNABIS ETFTHCX	Cannabis	4.6 yrs	-92%
ADVISORSHARES POSEIDON DYNAMIC CANNABIS ETFPSDN	Cannabis	1.8 yrs	-90%
VIRIDI BITCOIN MINERS ETFRIGZ	Crypto Price (BTC/ETH)	1.5 yrs	-83%
VOLT CRYPTO INDUSTRY AND EQUITY ETFBTCR	Blockchain & Web3 Equity	1.2 yrs	-77%
CAMBRIA CANNABIS ETFTOKE	Cannabis	6.7 yrs	-74%
SIMPLIFY VOLT FINTECH DISRUPTION ETFVFIN	Fintech	1.5 yrs	-71%
VANECK DIGITAL ASSETS MINING ETFDAM	Blockchain & Web3 Equity	1.1 yrs	-70%
PROSHARES ULTRAPRO NASDAQ BIOTECHNOLOGYUBIO	Genomics & Biotech	4.2 yrs	-66%
STKD 100% MSTR & 100% COIN ETFAPED	Crypto Price (BTC/ETH)	0.8 yrs	-64%
SIMPLIFY VOLT CLOUD AND CYBERSECURITY DISRUPTION ETFVCLO	Cybersecurity	2.3 yrs	-61%
GLOBAL X SOLAR ETFRAYS	Clean / Renewable Energy	4.0 yrs	-58%
DIREXION MOONSHOT INNOVATORS ETFMOON	Disruptive / Innovation	3.7 yrs	-58%
REX-OSPREY(TM) ETH + STAKING ETFESK	Crypto Price (BTC/ETH)	0.7 yrs	-57%
SIMPLIFY VOLT POP CULTURE DISRUPTION ETFVPOP	Disruptive / Innovation	1.5 yrs	-57%
DIREXION HYDROGEN ETFHJEN	Clean / Renewable Energy	3.3 yrs	-53%
CANNABIS GROWTH ETFBUDX	Cannabis	0.6 yrs	-51%
GLOBAL X EMERGING MARKETS INTERNET & E-COMMERCE ETFEWEB	Internet & Social	3.0 yrs	-50%
GLOBAL X WIND ENERGY ETFWNDY	Clean / Renewable Energy	3.9 yrs	-50%

Survivorship bias, made visible. When someone shows you a theme's "average return," ask where the bodies are.

The Grade (A–F) on each row is Performance — whether it actually works, judged against the right benchmark; the small number under it only compares within a category. Open any row to see its Rigor grade — how honestly we tested it: one point each for survivorship-free (dead companies left in), net of real cost, out-of-sample, point-in-time data (no peeking ahead), and look-ahead checked — alongside what the system was meant to do and what we learned.

The Test Bench

🛡️ If your fund had used the Risk Reducer

What to own — the report card

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do

What we found

The numbers

How honestly we tested it

What it's meant to do