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 large-cap U.S. equities — the megacap tech complex (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV), and staples/industrials (PG, KO, WMT, COST, CAT, XOM). It is currently flagged live.
Backtest: Encouraging on the Surface
Over 451 trading days the strategy returned 14.73%, ending at $11,473 on a $10k base — a 7.98% CAGR. The win rate is genuinely high at 70.59% across 38 trades, and the max drawdown of 15.64% is tolerable for an equity strategy. But the Sharpe ratio of 0.58 is the tell: a strong hit rate paired with a mediocre risk-adjusted return means the average loss is meaningfully larger than the average win — classic mean-reversion behaviour, where you're right often but occasionally caught holding a falling knife. Turnover of 879% is heavy; at this trade count fees stayed modest ($38 total), but that churn is a fragility to watch.
Validation: The Warning Light
This is where enthusiasm should cool. The strategy failed walk-forward validation. Across four folds, three were positive — but the ordering matters:
- Fold 1 (Aug 2024–Jan 2025): +2.06%
- Fold 2 (Jan–Jul 2025): +11.10%, Sharpe 1.32
- Fold 3 (Jul–Dec 2025): +2.21%
- Fold 4 (Dec 2025–May 2026): −2.84%, Sharpe −0.33
Almost all the reported edge lives in fold 2. The most recent out-of-sample window is a loss, and the headline OOS figures (−2.84% return, −0.33 Sharpe) confirm the edge is decaying, not compounding. The probabilistic scorecard is mixed: a PSR of 0.785 is respectable, but the deflated Sharpe of 0.304 — which penalises the 6 trials run — sits well below the confidence you'd want before scaling. The gate was right to fail it.
Recent Activity: Quietly Sidelined
Live behaviour mirrors the caution. The last executed trade was a WMT buy (21 shares @ $115.75) on May 31. Every scheduled run since — six consecutive sessions from July 13 through July 20 — logged 0 executed, 0 rejected. With RSI signals simply not firing, the book is 76% cash ($7,569 of a $9,930 total). Notably, that total sits below the $10k starting stake, so the live account is modestly underwater even as the backtest shows gains — a reminder that in-sample curves and real equity diverge.
Verdict
Mean-reversion has a coherent thesis, a high win rate, and contained drawdowns — real strengths. But the concentration of returns in a single fold, a negative most-recent window, and a deflated Sharpe near 0.3 argue against trusting it with size. Idle-and-in-cash is arguably the correct posture right now. Keep it live for observation, keep the allocation small, and let another clean out-of-sample fold decide whether the edge is real or a curve-fit ghost.