The Thesis
Bollinger-reversion runs a classic mean-reversion playbook: buy when price closes below the lower Bollinger band, sell when it pushes above the upper band. It operates on a 24-name large-cap universe spanning tech, financials, healthcare, staples, and energy — AAPL, NVDA, JPM, JNJ, XOM and the like. The logic is intuitive and well-worn: statistically stretched prices tend to snap back toward their moving average. It is the kind of thesis that works beautifully in textbooks and ranging markets, and struggles when trends persist.
Recent Activity
Live execution has been quiet. Across the last six scheduled runs (Oct 12–19), the strategy executed trades only once — two fills on Oct 15 — while several other runs logged rejected orders (one or two each on the 14th, 16th, and 19th). The paper account total has drifted between roughly $9,681 and $9,960, sitting at $9,935.23 on Oct 19, modestly below the implied $10,000 base. Cash is currently a large $2,677, suggesting the strategy is waiting for band breaches rather than forcing positions.
The trade log shows the thesis working as designed in at least one clean round trip: 70 shares of NKE bought at $35.64 on Sep 18 and sold at $37.56 on Oct 15 — a reversion of roughly 5.4%. Other recent moves include buying NVDA and JPM while trimming COST, CAT, WMT, and GOOGL.
Backtest and Validation
On paper the headline numbers are solid: 17.55% total return over 451 days, a 9.46% CAGR, a 63.89% win rate across 76 trades, and a Sharpe of 0.66. The caveats sit right next to them — a 20.57% maximum drawdown and an eye-watering 1,720% turnover. At that churn rate, fees and slippage matter, even if the modeled fee drag here was only $76.
The validation layer is where enthusiasm should cool. The strategy fails walk-forward validation. All four folds are positive, which sounds encouraging, but the trajectory is a steady fade: fold 1 returned 7.67% at a Sharpe of 1.78, then folds 2–4 collapsed to 1.42%, 0.77%, and finally 0.40% at a Sharpe of 0.14. The out-of-sample return is just 0.4% with an OOS Sharpe of 0.14 — the edge is concentrated almost entirely in the earliest period. A Probabilistic Sharpe of 0.814 looks reassuring until you note the Deflated Sharpe of 0.342 across six trials, which does not clear the bar for a durable, non-lucky edge.
Verdict
Bollinger-reversion has genuine strengths: a coherent thesis, a high win rate, and demonstrable reversion wins like the NKE trade. But the balance of evidence points to an edge that was real in 2024 and has largely decayed since. The out-of-sample flatline, failed validation, high turnover, and 20% drawdown argue for keeping this one on a short leash — useful as a diversifier and a learning case, not as a conviction allocation.