RSI + SMA Long in Low-Volatility Windows

We tested a long-only strategy that buys when RSI(14) drops below 35 and price sits above the 200-period simple moving average, then exits at a fixed profit target. It only trades when ADX is below 30, filtering out strong trending markets. The strategy graduated to our survivor library with an average gain of 1.065% per trade and an 87.5% win rate across walk-forward training and validation on 15-minute bars covering BTC, ETH, SOL, and XRP.

The deflated Sharpe ratio came in at 5.17, which sounds impressive. But the probability this result is a fluke after correcting for multiple testing is 47.1%. That sits just below our 50% gate, meaning it passed on a technicality. Nearly half the time, luck alone could produce numbers this good when you test enough variants. We graduated it because it cleared all six gates, but you should treat this as a borderline case, not a high-conviction signal.

The regime filter is doing most of the work here. By only entering when ADX is low, the strategy avoids getting chopped up in strong trends and waits for mean-reversion setups in quieter markets. The RSI threshold at 35 is slightly less extreme than the classic 30 oversold level, and the SMA(200) requirement ensures we are buying dips in an uptrend, not catching falling knives. The net profit exit mode locks in the 1.5% target without a stop loss, which works well when win rate is high but leaves you exposed to the occasional blowup if a trade goes badly wrong.

The biggest red flag is zero forward trades. This strategy has not taken a single live position since graduation, so all we have are backtest results. Until it logs real trades in changing market conditions, we cannot confirm the edge holds. The 47% fluke probability and the lack of forward data make this a watch-and-wait case, not a deploy-today case.

Check the full survivor registry at stratproof.com/survivors to see how this strategy compares to others, or visit stratproof.com/prove to backtest your own low-volatility mean-reversion ideas with custom thresholds.