The thesis
Mean-reversion is about as classic as systematic trading gets: buy what's oversold, sell what's overbought. Here the rule is mechanical — enter when 14-day RSI drops below 30, exit when it climbs above 70 — applied across a 24-name large-cap universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, BAC, V, MA), staples (PG, KO, WMT, COST), healthcare (JNJ, UNH, PFE), and energy (XOM, CVX). The bet is that liquid blue chips overshoot in both directions and snap back to fair value often enough to pay.
Recent activity
The strategy is live and currently near fully invested — cash sits at just $329.71. Most scheduled runs over the past two weeks did nothing ("0 executed, 0 rejected"), which is exactly what you'd expect from a signal that only fires at RSI extremes. The last execution was a 10-share MCD buy at $238.58 on 24 September, following HD (8 @ $299.12) and BAC (42 @ $58.21) earlier in the month — all buys, all textbook oversold entries.
Less comfortably, portfolio value has been drifting lower: from a $9,772.69 total on 23 September to $9,412.99 by 30 September. The book is holding its oversold positions and waiting for the bounce rather than cutting, which is faithful to the thesis but means it's currently carrying the drawdown.
Backtest performance
Over 451 days the strategy returned 14.73% (final equity $11,473.17), a 7.98% CAGR, with a 70.59% win rate across 38 trades and a modest $38 in total fees. The headline figures flatter it, though. The Sharpe is only 0.58 and max drawdown reached 15.64% — so the high win rate is doing a lot of work to cover a handful of larger losers. Turnover of 879% is heavy, which is tolerable here only because fees are small.
The validation problem
This is where balance matters. Walk-forward validation failed. Three of four folds were positive, but the most recent fold (Dec 2025–May 2026) lost 2.84% with a Sharpe of −0.33 — and recency is exactly the fold you'd weight most before trusting a strategy live. The aggregate out-of-sample return was that same −2.84%, versus 14.73% in-sample: a stark gap. The Deflated Sharpe Ratio of 0.304, adjusted for six trials, says the edge may not survive multiple-testing scrutiny, even though the PSR of 0.785 is respectable.
Verdict
Mean-reversion has a coherent thesis, clean execution, and a genuinely high hit rate — but the out-of-sample record undercuts the backtest, and the live book is sitting in a drawdown that mirrors the failing fold. It earns a place on the board as a watch-list live strategy, not a conviction allocation. The number to track next is whether its oversold entries actually revert, or whether the recent fold's weakness was a regime change rather than noise.