The thesis
Mean-reversion runs one of the oldest ideas in technical trading: buy weakness, sell strength. Concretely, it buys names whose RSI drops below 30 (oversold) and sells when RSI pushes above 70 (overbought). It operates over a broad, liquid universe of 24 US large-caps spanning tech, financials, healthcare, staples, and energy — AAPL, MSFT, NVDA, JPM, UNH, KO, XOM and the like. The strategy is currently live.
Backtest performance
Over 451 days the backtest returned 14.73% (final equity $11,473 on a $10k base), a CAGR of 7.98%. The headline strength is a 70.59% win rate across 38 trades — most positions closed green. The headline weakness is that this consistency didn't translate into risk-adjusted quality: the Sharpe is only 0.58, with a 15.64% max drawdown. In other words, the wins are frequent but the losses, when they come, are large enough to flatten the ratio. Turnover is heavy at 879%, though fees stayed trivial ($38 total) and there was no FX cost.
Validation: the part that matters
This is where the picture turns cautious. Walk-forward validation did not pass. Across four folds, three were positive — but the most recent fold (2025-12-16 to 2026-05-29) returned −2.84% with a −0.33 Sharpe and a 14.96% drawdown. That negative fold is also the out-of-sample result the validator reports, so the strategy's true forward estimate is a loss, not the 14.73% full-sample figure.
The probabilistic metrics agree. The Probabilistic Sharpe Ratio sits at a reasonable 0.785, but the Deflated Sharpe Ratio is just 0.304 across 6 trials — once you penalise for the number of configurations tested, the edge is no longer statistically convincing. A high in-sample win rate that degrades out-of-sample is the classic signature of a strategy partly fit to its own history.
Recent activity
The live book tells its own story: nothing is happening. The last six scheduled runs (Oct 7–14) all report 0 executed, 0 rejected — no RSI extreme has triggered an entry or exit. The most recent executed trades were buys in late September (MCD, HD, BAC) and a WMT buy back in May. Cash is pinned at $329.71, so the account is effectively fully invested and waiting. Meanwhile total equity has drifted around $9,440–$9,703, sitting below the $10k base — the live portfolio is currently underwater even as the backtest shows gains.
Verdict
Mean-reversion is intuitive, cheap to run, and wins often. But a failed validation, a negative out-of-sample fold, a deflated Sharpe near 0.3, and a live book that is both idle and below water are four independent reasons to treat the backtest's 14.73% as optimistic. Promising mechanics; unproven edge.