Thesis
The mean-reversion strategy runs on one of the oldest ideas in technical trading: prices that stretch too far tend to snap back. Mechanically, it buys when the 14-day RSI drops below 30 (oversold) and sells when RSI climbs above 70 (overbought), applied across a 24-name universe of large-cap US equities spanning tech, financials, healthcare, staples, and energy. It is currently live.
Headline Performance
The backtest is, on its face, attractive. Over 451 trading days it returned 14.73% (finishing at $11,473 on a $10,000 base), a 7.98% CAGR, with a 70.59% win rate across 38 trades. That hit rate is the strategy's most seductive number — most trades made money.
But the risk-adjusted picture is more sober. The Sharpe ratio of 0.58 is modest, and a 15.64% max drawdown means the equity curve took a meaningful punch along the way. Turnover of 879% is high for 38 trades, signalling frequent recycling of capital — though total fees stayed low at $38 and FX costs were nil.
Validation: The Gate It Didn't Pass
This is where enthusiasm should cool. The walk-forward validation failed. Of four sequential folds, three were positive — encouraging — but the story degrades over time:
- Fold 1 (Aug 2024–Jan 2025): +2.06%, Sharpe 0.57
- Fold 2 (Jan–Jul 2025): +11.10%, Sharpe 1.32 — the standout
- Fold 3 (Jul–Dec 2025): +2.21%, Sharpe 0.46
- Fold 4 (Dec 2025–May 2026): −2.84%, Sharpe −0.33, with a 14.96% drawdown
The most recent out-of-sample window is the only loser, and it's the one that matters most for what happens next. The Deflated Sharpe Ratio of 0.304 (against a healthier Probabilistic Sharpe of 0.785 across 6 trials) suggests much of the headline edge may not survive once you account for selection across trials. In plain terms: the backtest likely flatters the strategy.
Recent Activity
Live behaviour reinforces the caution. The last executed trade was a buy of 21 shares of WMT at $115.75 on 31 May 2026. Since then, the last six scheduled runs (29 Jul through 5 Aug) all report 0 executed, 0 rejected — the strategy is sitting on $7,569 cash against a total value hovering near $9,900. With no RSI extremes triggering, it is effectively parked, and the portfolio is currently slightly below its $10,000 starting line.
Verdict
Mean-reversion has real strengths: a high win rate, low costs, and a clear, explainable thesis. But the failed validation, negative most-recent fold, and deflated Sharpe are exactly the warning signs our validation gate exists to catch. The recent inactivity isn't a flaw — a mean-reversion system should wait for extremes — but combined with a sub-$10k live balance, it argues for keeping this one on a short leash. Promising in calm regimes; unproven in the one it's living through now.