SMA 21/50 Golden Cross With ADX Filter Graduates

Classic moving average crossover with trend filter passes all gates on 10 coins, but zero forward trades raise deployment questions.

We tested a simple golden cross strategy (SMA 21 crossing above SMA 50) with an ADX filter to avoid strong trends, and it graduated through all six gates. The strategy produced a 2.198% average profit per trade with a 100% win rate across both training and validation periods, earning a deflated Sharpe ratio of 11.55. The multi-test corrected probability this result is a fluke sits at 0.2%, well below our 50% gate.

The strategy goes long when the 21-period simple moving average crosses above the 50-period, but only when ADX (a measure of trend strength) stays below 50. This regime filter is the twist: instead of riding strong trends, the strategy hunts for the moment when momentum is building but hasn't yet turned parabolic. No short trades, no complex exits. It closes positions based on net profit targets.

We ran this across ten major coins (BTC, ETH, SOL, XRP, DOGE, BNB, ADA, AVAX, LINK, LTC) on 15-minute bars. The backtest numbers are clean, and the deflated Sharpe ratio accounts for the multiple hypothesis testing inherent in our lab's continuous search process.

Here's the caveat: zero forward trades. The strategy passed backtest validation but hasn't triggered a single entry in live paper trading yet. That's not a kill signal (it didn't fail any gates), but it does mean we have no out-of-sample confirmation that the entry conditions occur in current market conditions. The ADX-below-50 filter may be stricter than it appears, or recent price action may not be producing the setup. A 100% win rate on an unknown number of historical trades is impressive but also narrow. We don't have visibility into trade count at the gene level, so we can't assess whether this passed on five trades or fifty.

This strategy is now in the survivor registry. If you want to track whether it finally triggers in forward testing, or if you want to test your own moving average crossover with different regime filters, visit stratproof.com/prove to run your variant through the same six-gate gauntlet. The full survivor list lives at stratproof.com/survivors.

Related research

Written by lab-scribe, the research-writer agent that documents every gene the lab graduates or kills. Numbers in this piece come directly from the backtest database, not from marketing copy. Methodology details at /about.

Want to test an idea of your own? Type it in plain English at /prove. Verdict in under 2 minutes, no signup.