The thesis
Mean-reversion is about as classic as systematic equity trading gets: buy the oversold (RSI below 30), sell the overbought (RSI above 70). The bet is that short-term price extremes in large, liquid names snap back toward a local mean. The strategy runs on a 24-name universe of blue chips — AAPL, MSFT, NVDA, JPM, V, JNJ, WMT, XOM and the like — so there is no small-cap illiquidity hiding in the results.
What the backtest says
Over 451 days the backtest turned $10,000 into $11,473.17 — a 14.73% total return, or roughly 7.98% CAGR. The headline number traders love is the 70.59% win rate across 38 trades. Drawdown topped out at a manageable 15.64%, and fees were trivial ($38 total, no FX cost).
So far, so appealing. But two figures temper the story. The Sharpe ratio is only 0.58 — a lot of that 14.73% came with meaningful volatility, and a high win rate paired with a middling Sharpe usually means the occasional loser is large. And turnover is 879%, so the book churns itself nearly nine times over the test window. A high win rate on high turnover is fragile: it leans on many small wins that transaction costs and slippage can quietly erode.
Validation: it failed
This is the part worth dwelling on. Walk-forward validation across four folds returned a verdict of failed. Three of the four folds were positive — folds two and three delivered 11.1% and 2.21% — but the most recent fold (2025-12-16 to 2026-05-29) lost -2.84% with a -0.33 Sharpe. That final, out-of-sample stretch is the one that matters most, because it is the regime closest to today.
The statistical gatekeepers agree. The Probabilistic Sharpe Ratio sits at a respectable 0.785, but the Deflated Sharpe Ratio is just 0.304 across six trials — once you penalise for the number of configurations tried, the edge largely evaporates. The strategy looks good in-sample and shaky out-of-sample.
Live activity
The live book tells a consistent story. The last six scheduled runs (01–08 October) executed zero trades, with a couple of rejections, and total equity has drifted between $9,356 and $9,624 on just $329.71 of cash. The most recent executions — buys in MCD, HD and BAC in September, WMT back in May — left the portfolio almost fully deployed and then idle. At $9,623.88 the live book sits below the $10,000 the backtest started from.
Verdict
Mean-reversion has a coherent thesis, a genuinely high hit rate, and contained drawdowns — real strengths. But a failed validation, a negative latest fold, a deflated Sharpe near zero, and a flat, underwater live book are not noise to wave away. This is a strategy to watch, not to scale. We'd want to see the next out-of-sample fold turn positive before trusting the backtest's 14.73% as anything other than in-sample optimism.