The Thesis
The mean-reversion strategy applies a classic signal: buy when RSI falls below 30 (oversold), sell when it climbs above 70 (overbought). The universe spans 24 large-cap US equities across technology, financials, healthcare, consumer staples, energy, and industrials — a deliberate bias toward liquid, analyst-covered names where dislocations are more likely to snap back than to trend.
The logic is sound in theory. Large-caps attract institutional arbitrage, which tends to compress extreme moves. RSI-based entries exploit short-term sentiment overreaction without requiring a view on fundamentals.
Backtest Performance
Over 451 days through late May 2026, the strategy returned 14.73% (7.98% annualised), executing 38 trades at a win rate of 70.6%. Those are genuinely encouraging headline numbers — a 70%+ win rate in any systematic strategy warrants attention.
The risk picture is more mixed. The Sharpe ratio of 0.58 is modest; it clears zero but falls well short of the 1.0 threshold most practitioners expect before scaling. Max drawdown reached 15.64%, which is meaningful relative to the total return. High turnover (879%) signals the strategy cycles through positions aggressively, making fee and slippage assumptions load-bearing even when individual trade commissions are small.
Fold-by-Fold Breakdown
The cross-validated decomposition reveals an uneven ride:
| Fold | Period | Return | Sharpe | Max DD | Trades |
|---|---|---|---|---|---|
| 1 | Aug 2024 – Jan 2025 | +2.06% | 0.57 | 4.24% | 8 |
| 2 | Jan 2025 – Jul 2025 | +11.10% | 1.32 | 10.53% | 10 |
| 3 | Jul 2025 – Dec 2025 | +2.21% | 0.46 | 8.80% | 6 |
| 4 | Dec 2025 – May 2026 | −2.84% | −0.33 | 14.96% | 10 |
Fold 2 is the standout — a 1.32 Sharpe and 11.1% return in a single period. But Fold 4, the most recent and therefore the most predictive of current regime conditions, is the worst: negative return, negative Sharpe, and the highest drawdown of any window. Three of four folds are profitable, but the trajectory is the wrong direction.
Validation Gate: Failed
The strategy did not pass Headmars' automated validation. The out-of-sample return sits at −2.84% with a Sharpe of −0.33. Two probabilistic metrics reinforce the concern: the PSR (Probabilistic Sharpe Ratio) of 0.785 — meaning roughly a 78.5% chance the true Sharpe is positive — sounds reasonable in isolation, but the DSR (Deflated Sharpe Ratio) of 0.304 adjusts for the six trials run against this strategy. After multiple-testing correction, there is only a ~30% probability the observed edge is real. That is a low bar.
Recent Activity: Idle
Since the last executed trade — a 21-share purchase of WMT at $115.75 on May 31 — the strategy has run daily without firing a single order. Six consecutive daily logs (July 2–9) show zero executions. Cash stands at $7,569.25 against a total portfolio value oscillating around $9,900–$9,942.
Quiet periods are not inherently bad; a high-RSI-threshold strategy should sit out trending markets. But the extended inactivity, combined with a portfolio value that has declined from the ~$11,473 backtest final equity, is consistent with the Fold 4 drawdown pattern continuing into live operation.
Strengths and Risks
Strengths: High win rate (70.6%) provides psychological resilience and steady small gains. The universe is liquid and well-diversified. The signal is simple and not overfit to a single sector.
Risks: The most recent fold is the weakest, suggesting the regime may have shifted against mean-reversion in this universe. The DSR of 0.304 means the historical edge may be largely noise. High turnover amplifies real-world friction. The strategy is currently live despite failing validation — a decision worth revisiting if Fold 4 deterioration continues.
Bottom line: Mean-reversion has a plausible thesis and a solid historical win rate, but the validation data points to overfitting risk and possible regime drift. It warrants monitoring rather than scaling until out-of-sample performance stabilises.