The thesis
Mean-reversion runs one of the oldest ideas in technical trading: buy weakness, sell strength. Concretely, it buys names when RSI drops below 30 (oversold) and sells when RSI climbs above 70 (overbought), across a 24-name universe of US large caps — AAPL, MSFT, NVDA, JPM, JNJ, WMT, XOM and peers. It is currently flagged live.
Backtest: the good news
On paper the numbers flatter the strategy. Over 451 days it returned 14.73% (7.98% CAGR), lifting a notional book to $11,473. The standout figure is a 70.59% win rate across 38 trades — mean-reversion tends to produce many small, frequent wins, which is exactly what we see here. Costs were negligible ($38 in fees, no FX).
But two caveats sit under the headline. First, a Sharpe of 0.58 is modest: returns came with real volatility and a 15.64% max drawdown. Second, turnover ran to 879% — the book churned nearly nine times over. High turnover strategies are fragile to slippage and fills that a clean backtest never models.
Validation: the strategy failed the gate
This is where enthusiasm should cool. Walk-forward validation did not pass. Across four sequential folds, three were positive — fold 2 was excellent (+11.1%, Sharpe 1.32) — but the most recent fold (Dec 2025 → May 2026) turned negative at -2.84% with a Sharpe of -0.33 and a 14.96% drawdown. That the failure lands in the newest window is the worrying part: it suggests the edge may be decaying rather than merely noisy.
The deflated statistics tell the same story. Probabilistic Sharpe (PSR) of 0.785 is decent, but the Deflated Sharpe Ratio of 0.304 — which penalises for the 6 trials run — is well below any level that would justify conviction. Out-of-sample, this strategy has not yet earned its keep.
Recent activity: sitting on its hands
Live behaviour mirrors the caution. The last executed trade was a 21-share WMT buy at $115.75 on May 31. Since then — through six scheduled runs from Aug 11 to Aug 18 — the agent executed zero trades and rejected zero, holding $7,569 in cash against a total book of roughly $9,936–$9,998. In other words, the portfolio is marginally underwater versus its $10k start, and RSI simply hasn't triggered any fresh signals. For a mean-reversion system in a market without sharp oversold dips, inactivity is expected, not broken — but it does mean the strategy is currently contributing nothing.
Verdict
Mean-reversion is a textbook-clean idea with an attractive win rate and a genuinely strong middle fold. It is also a strategy that failed validation on its most recent data, carries a weak deflated Sharpe, and is presently idle and slightly in the red live. Treat it as a candidate under observation, not a deployable edge. The next oversold cluster will be the real test — and whether fold 4's weakness was a blip or a trend.