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, financials, healthcare, staples, and energy — AAPL, MSFT, NVDA, JPM, JNJ, XOM and friends. The bet is simple: liquid blue chips overshoot in both directions, and fading those extremes pays.
Backtest: the headline looks good
Over 451 days the strategy returned 17.55%, growing a notional $10k to $11,755 at a 9.46% CAGR. The win rate is a genuinely strong 63.89% across 76 trades, and fees were negligible ($76 total). That is the attractive face of the strategy.
The warts are in the risk profile. The Sharpe of 0.66 is mediocre — most of the return comes with uncomfortable volatility — and the max drawdown of 20.57% is roughly the size of the whole return. Turnover runs at 1,721%, so this is an active book constantly rotating in and out of band touches.
Validation: this is where it fails
Walk-forward validation returns passed: false, and the fold data shows why. All four folds are positive, which sounds reassuring, but the trend is a steady bleed:
| Fold | Return | Sharpe |
|---|---|---|
| 1 | 7.67% | 1.78 |
| 2 | 1.42% | 0.25 |
| 3 | 0.77% | 0.20 |
| 4 | 0.40% | 0.14 |
Fold 1 carries almost the entire edge; by fold 4 the strategy is barely distinguishable from flat. The out-of-sample figures confirm it — OOS return 0.40%, OOS Sharpe 0.14. With a deflated Sharpe ratio (DSR) of 0.342 against 6 trials, the probabilistic case for a durable edge is weak, even though the probabilistic Sharpe ratio (PSR) of 0.814 reads better in isolation. The honest read: the backtest's headline number is front-loaded by one good window, not a repeatable process.
Live activity: mostly waiting
Recent scheduled runs reinforce the picture. Across 2026-10-16 through 2026-10-23, the strategy executed almost nothing — typically 0 trades with 1–2 rejections per run — and total equity drifted down from roughly $9,998 to $9,853. The last real fills were on 2026-10-21 (a 3-share CAT buy, a 9-share MCD sell). Cash sits idle at $2,666. A band-touch strategy is naturally patient, but patience plus persistent rejections plus a sub-$10k book is not a thesis working; it is a thesis waiting.
Verdict
Strengths are real: high hit rate, all-positive folds, trivial costs. But the decaying fold returns, failed validation, and a sickly OOS Sharpe say the edge is thin and possibly mined from one favourable regime. I would keep it live only as a paper-traded candidate — not a capital allocation — until it demonstrates an edge that survives the walk-forward window it currently flunks.