The Thesis
Mean-reversion runs one of the oldest ideas in technical trading: buy oversold, sell overbought. Concretely, it buys names where RSI falls below 30 and sells when RSI pushes above 70, across a 24-stock large-cap universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (PG, KO, WMT), healthcare (JNJ, UNH) and energy (XOM, CVX). The premise is that short-term price extremes in liquid blue chips tend to snap back. It's a clean, interpretable rule — which is both its appeal and, as we'll see, its limitation.
Backtest: Solid In-Sample
Over 451 days the strategy returned 14.73%, ending at $11,473 from a $10,000 base — a 7.98% CAGR. The headline number that catches the eye is the 70.59% win rate across 38 trades, paired with a Sharpe of 0.58 and a max drawdown of 15.64%. That drawdown is worth pausing on: it's larger than the annualised return, so the ride is bumpier than the summary CAGR suggests. Turnover of 879% is also high for a 38-trade sample, though fees stayed modest at $38 total.
Validation: This Is Where It Wobbles
Here the picture darkens, and honesty demands we lead with it: validation did not pass. Across four walk-forward folds, three were positive — folds 1 through 3 returned 2.06%, 11.10% and 2.21% — but the most recent fold (Dec 2025–May 2026) lost 2.84% with a negative Sharpe of -0.33 and a 14.96% drawdown. The out-of-sample tail is the part that matters most, and it's negative on both return and risk-adjusted terms.
The probabilistic metrics echo this. The Probabilistic Sharpe Ratio of 0.785 is respectable, but the Deflated Sharpe Ratio of just 0.304 — which penalises for the 6 trials run — signals the in-sample edge may not be robust once selection bias is accounted for. In plain terms: the good numbers lean heavily on fold 2's strong 11.1% run, and the strategy has not demonstrated it can repeat that recently.
Recent Activity: Quiet and Underwater
Live behaviour reinforces the caution. The last six scheduled runs (Aug 24–31) all executed zero trades — no RSI extremes triggered — leaving the paper book parked at $7,569 cash and a total around $9,769. That total sits below the $10,000 start, and the last actual fill was a WMT buy back on 31 May. So the model is currently idle and modestly in the red on paper, a notable gap from the flattering backtest equity curve.
Verdict
Mean-reversion is a genuinely appealing, transparent strategy with a strong historical win rate — but the evidence is mixed and the risks are real. A failed validation, a negative most-recent fold, a low deflated Sharpe, and a paper account under water together argue for watching, not scaling. The thesis is intuitive; the out-of-sample proof isn't there yet.