Does ATR Expansion help RSI OB/OS?
Crypto · 213 backtest runs · As of 19 Oct 2026 · Methodology
Enough runs, but they disagree: the effect's 90% interval straddles the ±1 percentage-point line. Claiming a direction would assert more precision than the data supports.
CAGR Comparison
Risk-Adjusted Performance
How does this filter work?
Signals only during ATR expansion (trend onset)
Multiple-testing threshold
▼
Testing several filters on the same data raises the chance that one looks good by luck. SR̂₀ = 0.215 is the Sharpe the best of 11 trials (no filter + every filter of this strategy) would reach by chance alone — computed from the spread of these trials' Sharpe values (extreme-value formula per Bailey & López de Prado 2014).
The filter passes when its Sharpe (-0.082) lies above SR̂₀. This is a threshold test, not a probability: a true Deflated Sharpe would need each run's period count, skew and kurtosis. All Sharpe values are annualized over 365 calendar days.
Full analysis in Edge Library
Per-asset breakdown, trade-level data, Sharpe comparison — available after sign-in.