The thesis
The Momentum SMA20 Rider is a textbook trend-follower with a clear rulebook. It ranks a 24-name large-cap universe — think AAPL, MSFT, NVDA, JPM, XOM, COST — by how far price sits above its SMA50, then buys the leaders only once price is also holding above the SMA20. Exits are mechanical: trim when RSI14 pushes past 75 (overbought) or when a close breaks back below the SMA20 (trend structure gone). When the book is full, weaker names rotate out for stronger ranks. It's a coherent "buy strength, cut on first weakness" doctrine, and the exit logic is refreshingly unambiguous.
Recent activity
The strategy has been active almost daily. Over the last two weeks it executed buys in COST, V, WMT, DIS and XOM, and sells in V, AAPL, MSFT, DIS and NVDA — plenty of churn, including a round-trip in both V and DIS. The scheduled-run comments show a recurring pattern of one or two fills with a rejection alongside, suggesting the sizing or cash constraint regularly bites. Account value has drifted modestly lower across the window, from about $9,753 on 14 October to $9,699 on 21 October, with cash swinging between roughly $550 and $2,500 as positions rotate. That is roughly flat-to-slightly-down live performance over a short, noisy sample — worth watching, not yet worth worrying about.
Backtest and validation
The headline numbers are genuinely good: 81% total return, a 12.89% CAGR over 1,233 days, Sharpe 0.98, and a contained 20.91% max drawdown. Crucially, validation passed — a 4-fold walk-forward with 3 of 4 folds positive, an out-of-sample return of 43.48% at Sharpe 2.18, a PSR of 0.986 and a deflated Sharpe (DSR) of 0.583 across 24 trials. The DSR surviving multiple-testing deflation is the number I trust most here.
Strengths and risks
The strength is discipline: hard exits on RSI and SMA20 breaks keep losers small, which is exactly why a 37.13% win rate still compounds to 81% — this is an asymmetric, fat-right-tail profile, not a high-accuracy one.
The risks are equally plain. First, turnover is enormous — roughly 27,000% across the test, meaning fees and slippage matter a lot in live trading, and the daily rejections hint sizing isn't frictionless. Second, the record is uneven: Fold 1 (2021–22) returned just 4.77% and Fold 3 (2024–25) was essentially flat at −0.28% with a 23% drawdown. Most of the edge came from Fold 2 and Fold 4. A low win rate plus concentrated returns means a few missed winners could meaningfully dent results.
Verdict: a well-constructed, validated momentum engine whose math depends on letting winners run and killing losers fast. Respect the backtest, but size for the drawdown and watch whether live turnover costs erode the edge.