RSI(14) Oversold + EMA(200) Filter Passes All Six Gates

We tested a simple mean-reversion strategy that buys when RSI(14) drops below 30 on 15-minute bars and exits when the trade hits 1% net profit. It passed all six gates with a 0.673% average P&L per trade, a 90.5% win rate in walk-forward validation, and a deflated Sharpe ratio of 24.22. The probability this result is a statistical fluke after correcting for multiple testing is 0.0%.

The entry logic waits for RSI(14) to fall below the 30 oversold threshold, signaling short-term weakness. The strategy also includes an ADX filter that blocks trades when ADX exceeds 60, effectively avoiding strong trending environments where mean-reversion setups are more likely to fail. The exit is purely profit-based: once the position gains 1% net of fees, it closes. There is no time-based or technical exit, so losing trades can drift longer than winners.

We ran this across ten coins (BTC, ETH, SOL, XRP, DOGE, BNB, ADA, AVAX, LINK, LTC) on 15-minute data. Both training and validation walks returned identical metrics, which is unusual and worth noting. The 90.5% win rate is high, but the strategy takes a 1% profit target every time, so most trades that mean-revert even slightly will close profitably. The ADX ceiling likely helps by filtering out environments where price continues falling after the RSI spike.

The main caveat is trade count. The strategy has logged zero trades in forward paper testing so far, which means we have no live data yet. This is not a red flag on day one, but it does mean the real-world behavior is still unproven. The 15-minute bar resolution and the requirement for RSI to drop below 30 while ADX stays under 60 can produce long stretches without setups, especially in calm or strongly trending markets.

The deflated Sharpe ratio of 24.22 is exceptional and survives the haircut applied for multiple testing. The strategy is now live in paper mode and will either confirm these metrics or reveal regime-specific brittleness.

You can see this strategy and the rest of the survivor registry at stratproof.com/survivors, or test your own mean-reversion idea at stratproof.com/prove.