Thesis and Universe
Bollinger-reversion operates on a classic mean-reversion premise: enter long when a stock trades below its lower Bollinger Band — a statistical signal that price has stretched beyond a normal range — and exit when it recovers above the upper band. The universe spans 24 liquid large-caps across technology, financials, healthcare, consumer staples, and energy, giving the strategy broad diversification and ample rotation opportunities without venturing into thin or volatile small-cap territory.
Backtest at a Glance
Over 451 days of historical simulation, the strategy accumulated a 17.55% total return (~9.5% annualised), executing 76 trades at a 63.9% win rate. Those headline numbers read well for a pure long-side, rules-based system. The Sharpe ratio of 0.66 is modest — respectable enough for a strategy this simple, but leaving meaningful room for improvement. The sticking point is the 20.6% maximum drawdown, which is steep relative to the return generated; a risk-adjusted lens softens the headline story.
Turnover is notably high at 1,720%, reflecting frequent re-entry around band touches across a diversified universe. With $1 in flat fees per trade, total friction was $76 — negligible at this scale, though that changes quickly in live execution with slippage.
Walk-Forward Validation: The Warning Sign
The validation gate did not pass, and the fold-by-fold breakdown explains why:
| Fold | Period | Return | Sharpe | Max DD |
|---|---|---|---|---|
| 1 | Aug 2024 – Jan 2025 | 7.67% | 1.78 | 7.1% |
| 2 | Jan 2025 – Jul 2025 | 1.42% | 0.25 | 20.6% |
| 3 | Jul 2025 – Dec 2025 | 0.77% | 0.20 | 6.7% |
| 4 | Dec 2025 – May 2026 | 0.40% | 0.14 | 11.2% |
The degradation is monotonic and pronounced. Fold 1 — likely the in-sample anchor — delivered a Sharpe of 1.78 and a clean 7.1% drawdown. Every subsequent out-of-sample window is weaker, with the most recent period essentially flat at 0.4% return and a 0.14 Sharpe. The out-of-sample Sharpe (0.14) against a PSR of 0.814 and a Deflated Sharpe Ratio of 0.342 across six trials signals that a meaningful portion of the backtest performance may reflect regime luck rather than durable alpha. The validation framework is flagging exactly the risk it was designed to catch: a curve-fit edge fading as market conditions evolve.
Early Live Performance
Deployed May 31 with $10,000, the strategy immediately found two entries — WMT (21 shares at $115.75) and COST (2 shares at $956.32) — and added GOOGL and PG the following session. Since then, four consecutive daily runs have executed nothing, with one or zero rejected signals each day. The paper portfolio stood at $10,206 as of June 5, a modest positive start, but the sample is far too small to draw signal from.
The pattern of rejections suggests the position-sizing or risk rules are correctly throttling over-concentration — a healthy sign for capital preservation even if it limits upside in a quiet market.
Balanced Assessment
Strengths: Simple, interpretable logic; solid win rate; diversified universe; low transaction cost profile; all four folds were net-positive.
Risks: Monotonically declining fold performance points to regime sensitivity — the strategy appears most potent in mean-reverting, lower-volatility environments (Fold 1) and struggles when trends persist or volatility spikes (Fold 2's 20.6% drawdown). The low DSR (0.34) with six trials suggests the full-period Sharpe is likely overstated.
Bollinger-reversion is not broken, but it is operating outside its optimal conditions. The prudent posture is continued paper trading with tighter drawdown monitoring rather than capital allocation until the live record either validates or refutes the walk-forward signal.