The thesis
Bollinger-reversion is a textbook mean-reversion play: buy when price closes below the lower Bollinger band, sell when it pushes above the upper band. It runs on a 24-name large-cap universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (PG, KO, WMT, COST) and healthcare (JNJ, UNH, PFE). The premise is that liquid, well-covered names snap back to their moving average after short stretches, and the band width adapts the trigger to each name's volatility.
Backtest scorecard
Over 451 days the strategy compounded to a 17.55% total return (9.46% CAGR), finishing at $11,755 on a $10k base. The 63.89% win rate across 76 trades is genuinely strong and consistent with a mean-reversion edge that wins often but small. The blemishes are a modest Sharpe of 0.66, a 20.57% max drawdown that is steep for a staples-heavy book, and eye-watering 1,720% turnover — this is an active strategy, though fees stayed trivial at $76.
Validation says stop
Here is where enthusiasm should cool. The four-fold walk-forward test fails its gate. Every fold is positive — a good sign of directional robustness — but performance decays monotonically:
| Fold | Window | Return | Sharpe |
|---|---|---|---|
| 1 | 2024-08 → 2025-01 | 7.67% | 1.78 |
| 2 | 2025-01 → 2025-07 | 1.42% | 0.25 |
| 3 | 2025-07 → 2025-12 | 0.77% | 0.20 |
| 4 | 2025-12 → 2026-05 | 0.40% | 0.14 |
The most recent, most relevant fold earns just 0.40% at a 0.14 Sharpe — effectively flat. The Probabilistic Sharpe Ratio of 0.814 looks reassuring, but the Deflated Sharpe Ratio of 0.342 — which penalises the 6 trials run — undercuts it. The pattern is classic edge erosion: an early-window advantage that markets have largely arbitraged away.
What the live book is doing
Live paper trading corroborates the caution. The last executed fills were back in late May and early June (buys in DIS, GOOGL, WMT, COST, PG; one PG sell). Since then, the last six scheduled runs — July 20 through 27 — each report "0 executed, 1 rejected." The strategy keeps surfacing exactly one candidate and keeps getting it turned down, so cash sits frozen at $1,001.98 while total equity has drifted down from roughly $9,749 to $9,640.
Verdict
Bollinger-reversion is a well-behaved, high-hit-rate strategy with an honest backtest, and every fold staying positive argues the logic isn't broken. But the near-zero out-of-sample return, the deflated Sharpe, and a week of rejected orders all point the same way: the edge is thin and fading. It earns its "live" status for monitoring, not for scaling. I'd keep it on a short leash — small allocation, tight review — until a fresh fold shows the reversion signal reasserting itself.