The thesis
Bollinger-reversion is a textbook mean-reversion play: buy a name when it closes below its lower Bollinger band, sell when it pushes above the upper band. It runs across a 24-name large-cap universe spanning tech, financials, healthcare, staples and energy — AAPL and NVDA sit alongside JNJ, KO and XOM. The bet is that short-term dislocations in liquid blue chips revert, and the wide sector spread is meant to keep any single theme from dominating the book.
Backtest performance
On paper, the numbers look inviting. Over 451 days the strategy returned 17.55% (9.46% CAGR), finishing at $11,755 on a $10,000 base with a 63.89% win rate across 76 trades. Fees were a negligible $76 and FX cost was zero. But two figures temper the enthusiasm: a 20.57% max drawdown and a Sharpe of just 0.66. That combination — decent total return, deep drawdown, modest risk-adjusted return — tells you the equity curve was lumpy. Turnover of 1,721% also confirms this is an active book, not a buy-and-hold.
Validation: the warning sign
This is where the case weakens. The walk-forward validation did not pass. All four folds were positive, which sounds reassuring, but the returns decay hard as you move forward in time: fold 1 delivered 7.67% at a Sharpe of 1.78, then folds 2 through 4 limped in at 1.42%, 0.77% and 0.40%. The most recent out-of-sample window returned just 0.4% with a Sharpe of 0.14 — statistically indistinguishable from flat. The Probabilistic Sharpe Ratio of 0.814 is encouraging, but the Deflated Sharpe Ratio — which penalises for the six trials run — falls to 0.342, below the confidence threshold. The honest read: most of the headline return was front-loaded in the earliest period, and the edge has thinned since.
Recent live activity
The live book echoes that concern. The last executed trades were on 2026-08-07 (a CAT buy and a DIS sell). Every scheduled run since — 2026-09-08 through 2026-09-15 — reports the same story: zero executed, two or three candidates rejected each time. Cash has been frozen at $2,752.28 for over a week while total equity drifts between roughly $9,570 and $9,764 — below the $10,000 start. Either no names are printing valid band signals, or risk filters are vetoing every one.
Verdict
Bollinger-reversion has a coherent thesis and a clean backtest, and its high win rate suggests the core signal has merit. But the failed validation, the clear out-of-sample decay, and a live book that has stopped transacting all point the same way: the edge that showed up in 2024 has largely eroded. This is a strategy to watch and stress-test, not to scale. Until the folds stabilise and the live account resumes executing, treat the 17.55% as a historical artefact rather than a forward expectation.