SMA Cross With ADX Regime Filter Passes All Gates

We tested a simple moving average crossover (SMA 21 crossing above SMA 50) that only trades when ADX sits below 45, filtering out strong trending regimes. The strategy graduated with 2.198% average profit per trade in both training and validation, a 100% win rate, and a deflated Sharpe ratio of 11.55. After correcting for multiple testing, the probability this is a statistical fluke stands at 0.2%, well below our 50% gate.

The regime filter is the interesting piece here. Most golden cross systems struggle in choppy markets and shine in trends. This one inverts that logic, entering only when ADX signals low directional conviction. The backtest suggests mean reversion or range-bound behavior dominates those regimes across the ten coins we tested: BTC, ETH, SOL, XRP, DOGE, BNB, ADA, AVAX, LINK, and LTC on 15-minute bars.

The 100% win rate and identical training and validation P&L should raise your eyebrows. We raise ours too. Perfect scores in backtest often mean the strategy fired rarely enough that every trade happened to land in favorable conditions, or the exit logic (net profit target in this case) cut losses before they registered as losing trades. The data block above does not include total trade count at the gene level, but validation trade count was not high enough to give us confidence in the win rate's stability.

Forward testing has recorded zero trades so far. The strategy is live in paper mode, but ADX has not dipped below 45 during the right SMA alignment since graduation. This is not a kill signal, but it does mean we have no out-of-sample evidence yet. Treat the backtest numbers as a hypothesis, not a forecast.

If you are drawn to regime-filtered trend systems, head to /prove and test your own ADX or volatility gates. The full survivor registry lives at /survivors, where you can track this strategy's forward performance as trades arrive.