Thesis in Brief
Mean-reversion is one of the oldest edges in equity markets: prices that stretch too far from equilibrium tend to snap back. This strategy operationalises that idea simply — buy when RSI drops below 30 (oversold), sell when RSI climbs above 70 (overbought) — across a 24-stock universe of large-cap US names spanning tech, financials, healthcare, consumer staples, and energy.
The universe is deliberately blue-chip. High-liquidity, well-covered stocks tend to exhibit mean-reversion more reliably than microcaps, and they carry less gap-risk overnight.
Backtest Performance
Over 451 days of paper history, the strategy returned 14.73% on a starting stake, compounding to a final equity of $11,473. Annualised, that works out to roughly 7.98% CAGR — modest but respectable for a rule-based system with no leverage.
The win rate of 70.59% across 38 trades is the headline number, and it genuinely is strong for a mean-reversion approach. The strategy catches reversals more often than not. The Sharpe ratio of 0.58 tells a more measured story: returns are positive but not particularly smooth, and the maximum drawdown of 15.64% is meaningful — almost certainly the strategy holding through an extended trending period where RSI stayed suppressed.
Turnover of 879% annualised is high. At $1 per trade in fees (total fees: $38), the cost drag is trivial at current scale, but it would become a real friction point at larger position sizes or with a broker that charges per-share.
Cross-Validation: Where It Gets Interesting
The strategy ran through a 4-fold walk-forward validation, and the results are worth reading carefully.
| Fold | Period | Return | Sharpe | Max DD |
|---|---|---|---|---|
| 1 | Aug 2024 – Jan 2025 | +2.06% | 0.57 | 4.24% |
| 2 | Jan 2025 – Jul 2025 | +11.10% | 1.32 | 10.53% |
| 3 | Jul 2025 – Dec 2025 | +2.21% | 0.46 | 8.80% |
| 4 | Dec 2025 – May 2026 | −2.84% | −0.33 | 14.96% |
Fold 2 is exceptional — a 1.32 Sharpe over six months is the kind of number that flatters overall statistics. Folds 1 and 3 are acceptable, if unexciting. Fold 4 is the problem: the most recent out-of-sample window produced a −2.84% return and a −0.33 Sharpe, with a nearly 15% drawdown. The validation gate correctly flagged this: validation did not pass.
The Probabilistic Sharpe Ratio (PSR) of 0.785 and a Deflated Sharpe Ratio (DSR) of 0.304 — the latter accounting for multiple testing across 6 trials — reinforce the concern. The backtest looks good; the forward signal is weaker than it appears.
Recent Activity
The strategy has been dormant at the signal level for at least two weeks. Every scheduled run from June 24 through July 1 logged zero executions. Portfolio value drifted from $10,083 on June 24 down to $9,848 by July 1 — a passive mark-to-market decline on existing positions, not new activity. Cash has been unchanged at $7,569 throughout this stretch.
The last actual trade was a buy of 21 shares of WMT at $115.75 on May 31. With RSI thresholds set at hard levels (< 30 / > 70), extended low-volatility or range-bound markets will simply produce no signals — which is what we appear to be seeing.
Verdict
Mean-reversion has real signal in the right market conditions, and a 70% win rate is not an accident. But the deterioration in Fold 4 — the most recent and therefore most relevant window — is a legitimate red flag. The gap between in-sample (0.58 Sharpe) and out-of-sample (−0.33) is wide enough to suggest the strategy either needs parameter refinement or is encountering a regime where large-caps are trending rather than reverting.
For now, the validation gate is doing its job. This strategy should remain in paper-trading observation until it demonstrates stability in live conditions.