The thesis
"Momentum SMA20 Rider" is a textbook trend-follower. It ranks a 24-name large-cap universe by distance above the SMA50, buys only names whose price is also holding above the SMA20, and rides the confirmed uptrend. Exits are mechanical: cut on RSI14 above 75 (overbought) or on a close below the SMA20 (structure broken), then rotate freed capital into higher-ranked candidates when the book is full. It is a clean, rules-based expression of "buy strength, sell when strength fades."
Backtest performance
Over 1,233 trading days the strategy turned a starting book into $18,100.23 — an 81% total return, or 12.89% CAGR. Risk-adjusted, that is a Sharpe of 0.98 with a 20.91% max drawdown: respectable, not spectacular.
Two numbers deserve scrutiny. The win rate is just 37.13%, and turnover is a staggering 26,971%. Neither is automatically damning — trend-following lives on many small losses funding a few large winners, so a sub-40% hit rate is expected. But 1,427 trades means 1,427 in fees, and that churn is a permanent drag the strategy must out-earn every year.
Validation
This is where the strategy earns credibility. A 4-fold walk-forward test passed, with 3 of 4 folds positive. The out-of-sample stretch (May 2025–Aug 2026) returned 43.48% at a Sharpe of 2.18 and a tiny 5.38% drawdown. A Probabilistic Sharpe Ratio of 0.986 across 24 trials is strong evidence the edge is not pure luck, and the Deflated Sharpe of 0.583 — which explicitly penalises multiple testing — stays comfortably positive.
The honest caveat: fold 3 (Feb 2024–May 2025) returned −0.28% with a 23% drawdown. The approach can stall for a year in choppy, trendless tape. That is the character of the strategy, not a defect, but anyone deploying it should expect flat-to-painful stretches.
Recent live activity
The live book tells a more sober story. Total equity has drifted down from $10,056 (Oct 30) to $9,447 (Nov 6) — a shade under 5.5% below its starting line. Recent runs show the engine working as designed: on Nov 3–4 it sold energy (XOM, CVX) and trimmed V, rotating into consumer and staples names (WMT, HD, DIS, MCD). Rejections are frequent — 2 of 4 orders bounced on Nov 4, 2 of 4 on Nov 3 — likely cash or sizing constraints worth monitoring.
Verdict
The validation profile is genuinely good: this is not an overfit curve. But the early live drawdown is a useful reality check. A 37% win rate and triple-digit turnover demand patience and tight fee discipline, and the fold-3 result warns that trendless markets will test conviction. Strong engine, sound rules — now it needs a real trend to ride.