BB + RSI Entry with ADX Filter Passes All Gates

We tested a mean-reversion strategy that buys when price touches the lower Bollinger Band (20-period) and RSI drops below 30, but only when ADX is below 50 to avoid strong trending markets. It passed all six gates with a 27.88 deflated Sharpe ratio (a risk-adjusted return metric corrected for multiple testing) and a 0.0% probability of being a statistical fluke after accounting for the thousands of candidates we tested in parallel.

The strategy exits on a fixed net profit target of 1.5% and was tested across ten coins (BTC, ETH, SOL, XRP, DOGE, BNB, ADA, AVAX, LINK, LTC) on 15-minute bars. In both walk-forward training and validation, it returned 1.101% average profit per trade with a 90% win rate. Those numbers are identical across training and validation, which is unusual and suggests either very stable behavior or a small sample size. We do not have per-gene trade counts in this data export, so treat the win rate with appropriate caution until you see the full breakdown on the strategy page.

The ADX filter is the key design choice here. By avoiding markets with ADX above 50, the strategy sidesteps the problem that kills most mean-reversion systems: getting run over during breakouts. The tradeoff is fewer opportunities. Low-ADX regimes are quieter by definition, so this will sit idle during volatile, directional moves that trend-followers love.

Where to be skeptical: we do not yet have forward-test data (live paper trading shows zero trades so far), and the identical training and validation stats raise a yellow flag about sample size. The strategy also locks in at a 1.5% net profit target, which means it never lets winners run. If the next regime shift brings sustained trends, this will underperform, and that is by design.

You can see the full regime breakdown, trade-by-trade results, and live forward-test updates on the survivor registry. If you want to test your own mean-reversion idea with custom filters, head to StratProof Prove and run it through the same six-gate gauntlet.