The Thesis
Mean-reversion is one of the oldest ideas in systematic trading: assets that have sold off sharply tend to snap back. This strategy operationalises that intuition with RSI — entering long when RSI drops below 30 (oversold) and exiting when RSI climbs above 70 (overbought). The universe covers 24 large-cap US names across technology, financials, healthcare, consumer, and energy, giving the strategy enough breadth to find signals without straying into illiquid territory.
Backtest: Encouraging on the Surface
Over the 451-day backtest window, the strategy compounded to a 14.73% total return (7.98% annualised), with a 70.6% win rate across 38 trades. Those are respectable numbers for a rules-based system with no parameter optimisation beyond a single RSI threshold pair.
Fee drag was minimal — flat $1 per trade, $38 total — and FX costs were zero given the all-USD universe. Turnover ran high at 879%, which is worth watching if the strategy ever moves to live execution with real commissions and spread.
The Sharpe of 0.58 is modest but positive, and the max drawdown of 15.64% is within acceptable bounds for an equity-long strategy. On raw backtest numbers alone, this looks like a workable foundation.
Validation: Where the Story Gets Complicated
The four-fold walk-forward validation tells a more sobering story. Three of the four folds were profitable:
| Fold | Period | Return | Sharpe | Drawdown |
|---|---|---|---|---|
| 1 | Aug 2024 – Jan 2025 | +2.06% | 0.57 | 4.24% |
| 2 | Jan – Jul 2025 | +11.10% | 1.32 | 10.53% |
| 3 | Jul – Dec 2025 | +2.21% | 0.46 | 8.80% |
| 4 | Dec 2025 – May 2026 | -2.84% | -0.33 | 14.96% |
Fold 2 is the outlier that flatters the aggregate — remove it and the picture is largely flat. More concerning, Fold 4 is the most recent period, meaning the strategy's live deployment begins exactly where its out-of-sample performance turned negative.
The probabilistic Sharpe ratio (PSR) of 0.785 suggests the Sharpe is above zero with moderate confidence, but the deflated Sharpe ratio (DSR) of 0.304 — which adjusts for the number of trials tested — is the figure that matters most here. With six trials and a DSR below 0.5, there is a meaningful probability the observed edge is attributable to selection rather than genuine alpha. The validation gate correctly flagged this as a fail.
Live Activity So Far
The strategy deployed on 31 May 2026 with $10,000, immediately buying 21 shares of WMT at $115.75 — a textbook oversold entry on a defensive consumer name. Since then, five consecutive daily runs have found no qualifying signals: RSI conditions on the remaining 23 names have not triggered entry thresholds. As of 5 June, total portfolio value sits at $10,094.92 ($7,569.25 cash + open WMT position), marginally above water.
The extended silence is not unusual for a high-threshold RSI strategy in a market that has recovered from recent lows — there simply are not many names oversold enough to qualify.
Risk Factors to Watch
- Recency of underperformance: The strategy's worst fold is its most recent. If the market regime that caused Fold 4's losses persists, the live run starts in a headwind.
- Overfitting risk: A DSR of 0.304 with six trials warrants caution. The win rate and return numbers may partly reflect the specific threshold values chosen.
- High turnover: At 879% annualised, scaling this to larger capital with real transaction costs would materially compress net returns.
- Single-factor signal: RSI alone, without volume, trend-filter, or sector context, can generate false reversals in trending markets.
Bottom Line
Mean-reversion is a coherent, well-understood strategy with a clean backtest and an easy-to-audit signal. The concerns are real but not disqualifying — the DSR and recent fold failure are reasons to watch closely, not reasons to dismiss the approach outright. The next 60 days of live paper trading, particularly how it navigates any market dislocation that produces genuine oversold readings, will be far more informative than the backtest ever was.