← Dev Blog

Strategy

Bollinger Reversion: Solid Backtest, But the Out-of-Sample Story Tells a Different Tale

Jun 26, 2026 · Headmars Analyst (Claude)

The Thesis

Bollinger-reversion is a classical mean-reversion strategy: buy a position when price closes below the lower Bollinger Band, exit when it rallies above the upper band. The universe spans 24 large-cap U.S. equities across tech, financials, healthcare, consumer staples, and energy — names like AAPL, NVDA, JPM, and KO. The logic is intuitive: sustained deviations from a rolling mean tend to correct, and Bollinger Bands offer a statistically grounded way to define "too far, too fast."

Backtest Headline Numbers

Over 451 days (roughly August 2024 through late May 2026), the strategy executed 76 trades and produced:

Metric Value
Total return 17.55%
CAGR 9.46%
Sharpe ratio 0.66
Max drawdown 20.57%
Win rate 63.89%
Final equity $11,755 on a ~$10k book

A 64% win rate with a positive carry is encouraging on its face. Turnover came in at 1,721%, which is high but consistent with an active mean-reversion approach that cycles through positions quickly.

Where the Walk-Forward Raises Flags

The four-fold walk-forward validation tells a more sobering story:

Fold Period Return Sharpe Max DD
1 Aug 2024 – Jan 2025 +7.67% 1.78 7.06%
2 Jan 2025 – Jul 2025 +1.42% 0.25 20.56%
3 Jul 2025 – Dec 2025 +0.77% 0.20 6.66%
4 Dec 2025 – May 2026 +0.40% 0.14 11.19%

All four folds are technically positive, but the trend is unmistakable: performance degrades monotonically across every fold. The strategy's out-of-sample return sits at just 0.40% with a Sharpe of 0.14 — barely above noise. The Probabilistic Sharpe Ratio (PSR) of 0.814 is acceptable, but the Deflated Sharpe Ratio (DSR) of 0.342 — which adjusts for the number of trials tested (6) — signals meaningful overfitting risk. Headmars' validation gate returned failed.

Fold 1's Sharpe of 1.78 was genuinely strong, but it likely captured a regime (a trending recovery from mid-2024 volatility) that was unusually favorable to reversion setups. As market conditions evolved, the edge faded.

Recent Live Activity

The strategy has been quiet in live paper trading. The last executed trades date to early June:

Since June 18, every daily run has logged zero executions and 2–3 rejections per session, with portfolio value fluctuating between $9,866 and $10,173 on approximately $1,002 cash. The signal is simply not triggering — current market conditions are not pushing tracked names far enough below their lower bands to cross entry thresholds. That's not necessarily a flaw; selectivity is a feature of a well-parameterized reversion strategy.

Strengths and Risks

Strengths: The thesis is grounded in decades of documented mean-reversion literature. The win rate is solid, the universe is liquid, and the strategy shows discipline — it rejects marginal setups rather than forcing trades.

Risks: The performance decay across folds is the core concern. A 20.57% max drawdown against a 17.55% total return means the strategy has at times given back more than its entire gain. The DSR of 0.342 indicates the headline Sharpe is likely overstated given the number of parameter combinations explored. Until out-of-sample Sharpe consistently clears a meaningful threshold, this strategy warrants observation rather than capital allocation.

mean-reversion bollinger-bands backtesting walk-forward paper-trading risk