Thesis
Mean-reversion runs one of the oldest ideas in systematic trading: prices overshoot, then snap back. It buys names that look oversold on the 14-day RSI (below 30) and sells those that look overbought (above 70), operating across a 24-name universe of US large caps spanning tech, financials, healthcare, staples, and energy. The appeal is a high hit rate — you are betting on the average, and the average usually shows up.
Backtest performance
On paper the strategy looks respectable. Over 451 days it turned an initial book into $11,473 of final equity — a 14.73% total return, or roughly 7.98% CAGR. The win rate is a strong 70.59% across 38 trades, and fees were negligible at $38 with no FX drag.
But the quality metrics temper the headline. The Sharpe ratio is only 0.58, and the maximum drawdown is 15.64% — meaning the ride to that return was bumpier than the return itself justifies. The 879% turnover is also high: this is an active strategy churning the book many times over, which raises slippage and execution risk that a clean backtest understates.
Validation: the honest verdict
This is where the strategy earns its caution flag. Validation failed. Walk-forward testing over four folds showed three positive periods and one negative — but the negative one is the most recent (Dec 2025–May 2026), returning -2.84% with a -0.33 Sharpe and a 14.96% drawdown. Out-of-sample performance mirrored that weakest fold.
The deflated metrics tell the same story. The Probabilistic Sharpe Ratio of 0.785 is reasonable, but the Deflated Sharpe Ratio — which penalises for the six trials run — drops to 0.304. A DSR below 0.5 means we cannot confidently reject the possibility that the backtest edge is a product of selection rather than skill. Fold 2's excellent 1.32 Sharpe is doing a lot of the heavy lifting for the full-period number.
Recent activity
The live book reflects this uncertainty in the most literal way possible: it isn't doing much. The last executed trade was a WMT buy (21 shares at $115.75) back on 31 May. Every scheduled run since early September has printed the same line — 0 executed, 0 rejected — with $7,569 sitting in cash against a total portfolio value hovering between roughly $9,788 and $9,872. In other words, no RSI signal has fired, and the strategy is patiently parked, slightly underwater on its live mandate.
Balance sheet on the idea
Strengths: a coherent, well-understood thesis; a high win rate; low fees; and, right now, discipline — it isn't forcing trades without signals.
Risks: the failed validation is the headline. A negative most-recent fold, a modest Sharpe, high turnover, and a deflated DSR all suggest the historical edge may not survive forward. Mean-reversion also carries a structural tail risk — it wins often but can lose big when an oversold name keeps falling.
Our read: keep it live and observed, but do not scale it. The next few signal-bearing runs matter more than the backtest ever did.