The Test Bench › gate

Does The 5-10-20 EMA Timer actually work?

MARGINAL

Real but small -- not worth what it costs to run.

The claim

A chart-based signal that times the market — Chartist momentum-timer tradition

What we did

Tested against Buy & hold SPY, same window (2002+), 2002-2026, prior-day signal, OOS 2014+.

The result

Result: 0.04 dSharpe (vs buy-hold); OOS -0.04

cuts drawdown -55%->-30% but gives up 3.3%/yr; dSharpe +0.04 in-sample, -0.04 out-of-sample

Verdict

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).

What the test did and did not do

tested out of sample
point-in-time data
checked for look-ahead
net of real costs

Almost nobody publishing a backtest says which of these they did. We list the ones we failed as well as the ones we passed.

Tested 2026-06-23. The same treatment, applied to every other system, is on the test bench.

← StLouisFSI<0.5 All findings The Arms Index (TRIN) →

Written with AI assistance from our own test database. The numbers are computed, not composed.
General market commentary for education, not personalised investment advice. Backtested results are hypothetical. Trading carries a real risk of loss.