The Test Bench › indicator

Does The Entire TA Library (pandas-ta, 257 indicators) actually work?

DEAD

No edge left once tested on a fair footing.

The claim

Surely SOME technical indicator can time the market — Technical-analysis canon

What we did

Tested against Buy & hold SPY, the full pandas-ta-classic library on SPY 1993-2026, OOS>=2014, null-floor controlled.

The result

Result: 257 indicators tested; 0 provide a real timing edge (the lone apparent winner is a lookahead artifact)

257 indicators tested; 0 provide a real timing edge (the lone apparent winner is a lookahead artifact)

Verdict

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.

What the test did and did not do

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

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.

← Stochastic RSI (14) All findings True Strength Index →

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.