The thesis
Mean-reversion is one of the oldest ideas in systematic trading, and this strategy states it plainly: buy the oversold (RSI below 30), sell the overbought (RSI above 70). It runs across a 24-name universe of US large caps — the megacap tech complex (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare, staples, and a handful of industrials and energy names. The premise is that short-term price extremes in liquid, well-covered stocks tend to snap back. It's a clean, testable hypothesis, which is exactly why it's worth scrutinising rather than taking on faith.
Recent activity
The strategy is live and trading on schedule. Over the sessions from 18–25 September it was mostly quiet: several runs executed zero trades, with single buys landing on 21 and 24 September. Recent fills are all on the long side — MCD (10 shares at $238.58), HD (8 at $299.12), and BAC (42 at $58.21) — consistent with a strategy hunting oversold entries. Notably, there are no sells in the recent tape, so the book has been accumulating rather than rotating.
The live account tells a sober story. Total equity drifted from roughly $9,841 on 22 September to $9,692 by 25 September, leaving the paper portfolio slightly underwater against its $10,000 base. That's a small sample and not a verdict, but it does not yet corroborate the backtest's optimism.
Backtest and validation
On paper the full-sample numbers look attractive: 14.73% total return over 451 days, a 7.98% CAGR, a 70.59% win rate across 38 trades, and a manageable 15.64% max drawdown. Fees were negligible ($38, no FX cost). The catch is turnover of 879% — this is an active strategy, and slippage assumptions matter more than the fee line suggests.
The headline is that validation failed. Across four walk-forward folds, three were positive, but the results are lopsided. Fold 2 (Jan–Jul 2025) did the heavy lifting with an 11.1% return and a 1.32 Sharpe; folds 1 and 3 were merely fine (+2.06%, +2.21%); and the most recent fold (Dec 2025–May 2026) lost 2.84% with a negative −0.33 Sharpe and a near-15% drawdown. The out-of-sample Sharpe is negative, and while the probabilistic Sharpe ratio (0.785) is respectable, the deflated Sharpe ratio of 0.304 — which penalises the six trials run — is not convincing. In plain terms, the full-sample return leans heavily on one favourable window, and performance has been decaying into the present.
The balance
The strengths are real: a high win rate, a coherent and well-understood thesis, low direct costs, and a liquid universe. But the risks line up against the recent record. The negative latest fold, the weak deflated Sharpe, the long-only accumulation with no exits firing, and a live account that's slightly down all point the same direction. Mean-reversion may still earn its keep, but the evidence favours running it small and watching whether the fold-4 weakness was noise or the start of a regime the RSI rules don't handle. Right now, conviction should track the validation — which says not yet.