The Thesis
Mean-reversion is one of the oldest ideas in systematic trading: prices that stretch too far from their recent norm tend to snap back. This strategy operationalizes that with the Relative Strength Index — buy when a name is oversold (RSI below 30), sell when it is overbought (RSI above 70). It runs over a 24-name universe of large-cap US equities spanning tech, financials, healthcare, staples, and energy (AAPL, MSFT, NVDA, JPM, V, JNJ, XOM, and others), which gives the signal a broad, liquid pool to fish in.
Backtest Performance
On paper, the numbers are attractive. Over 451 trading days the strategy returned 14.73%, ending on final equity of $11,473 from a $10,000 start — a CAGR of roughly 7.98%. The win rate is a headline-grabbing 70.59% across 38 trades, and total fees were a negligible $38. Drawdown was contained but not trivial at 15.64%.
Two caveats sit underneath those figures. First, the Sharpe ratio is only 0.58 — the returns came with meaningful volatility, so risk-adjusted performance is mediocre despite the high win rate. Second, turnover was 879%, meaning the book effectively churned nearly nine times over. A high win rate paired with a middling Sharpe usually signals many small wins offset by fewer, larger losses — a classic mean-reversion profile that works until it doesn't.
Validation: The Red Flag
Here the story turns cautionary. Walk-forward validation did not pass. Across four folds, three were positive, but the most recent fold (Dec 2025 – May 2026) delivered a -2.84% return with a -0.33 Sharpe — precisely the out-of-sample window that matters most for a live deployment. Fold 2 carried much of the full-period gain (11.1%, Sharpe 1.32), so performance is concentrated rather than consistent.
The probabilistic metrics echo the concern. The Probabilistic Sharpe Ratio sits at 0.785, but the Deflated Sharpe Ratio is just 0.304 after accounting for six trials — below the confidence threshold you would want before trusting the edge as real rather than a product of selection. In short: the backtest looks good, but the evidence that the edge persists out-of-sample is weak.
Recent Live Activity
Live behavior has been notably quiet. The last executed trade was a buy of 21 shares of WMT at $115.75 on May 31. Since then, the six most recent scheduled runs (Aug 13–20) each report 0 executed, 0 rejected, with cash parked at $7,569.25 and total account value drifting between roughly $9,738 and $9,998 — i.e. still slightly below the $10,000 baseline. No RSI extremes have triggered entries, which is consistent with a calm, trendless tape offering few oversold setups.
Verdict
Mean-reversion is a coherent, cheap-to-run strategy with a genuinely high hit rate and modest drawdowns. But the failing validation and the negative most-recent fold are exactly the signals our roadmap flags as overfitting risk. It earns its live slot as a paper-traded experiment — watch it, don't trust it yet. The next few oversold entries will tell us far more than the backtest already has.