The thesis
Bollinger-reversion is a classic mean-reversion play: buy when price closes below the lower Bollinger band, sell when it pushes above the upper band. The bet is that short-term dislocations in large, liquid names snap back toward their moving average. Its universe is 24 S&P-heavyweight tickers spanning tech, financials, healthcare, staples, and energy — AAPL, MSFT, NVDA, JPM, JNJ, XOM and the like — so it leans on breadth and diversification rather than any single-name conviction.
Recent activity
The live book has been quiet. Across six scheduled runs between 8 and 15 October, only the 15 October run did anything meaningful: two executions — a 2-share NVDA buy at $217.62 and the close of a 70-share NKE position at $37.56. That NKE lot was opened at $35.64 on 18 September, a clean if modest reversion win. Earlier October trades followed the same rhythm: a JPM buy, a COST and CAT trim, a JNJ add. Most sessions, though, logged zero executions — the bands simply weren't being breached.
Notably, the live paper account tells a more sober story than the backtest. Total equity sat at $9,911.45 on 15 October, fractionally below its $10,000 starting line, even as the historical simulation shows a +17.55% gain. Live and backtest are not yet telling the same story.
Backtest and validation
On paper the numbers look inviting: +17.55% total return over 451 days (9.46% CAGR), a 63.89% win rate across 76 trades, and a Sharpe of 0.66. All four walk-forward folds came back positive — a genuine strength, since not every strategy survives being sliced across regimes.
But the detail is where the caution lives. The per-fold returns decay almost monotonically: 7.67% (Sharpe 1.78), then 1.42%, 0.77%, and finally just 0.40% (Sharpe 0.14) in the most recent window. The edge isn't holding up as we march forward in time — the first fold is doing nearly all the heavy lifting. Out-of-sample return of 0.40% and OOS Sharpe of 0.14 are thin by any measure.
Crucially, validation failed. While the Probabilistic Sharpe Ratio is a healthy 0.814, the Deflated Sharpe Ratio — which penalises for the six trials run — drops to 0.342, below our confidence threshold. Translated: once you account for how many variants were tested, we can't be confident the 0.66 Sharpe is real rather than lucky.
Strengths and risks
Strengths: high win rate, every fold positive, a sensible and liquid universe, and low fee drag in absolute terms ($76 total).
Risks: a 20.57% max drawdown against a sub-1.0 Sharpe is an uncomfortable risk-reward ratio, and turnover of 1,720% means the strategy is churning hard for its returns. The decaying fold performance and failed DSR gate are the headline concerns — this looks like a strategy whose edge was strongest in the earliest window and has been fading since.
Verdict
Bollinger-reversion is a tidy, interpretable mean-reverter with an honest win rate, but it has not earned deployment confidence. We'd keep it live on paper, watch whether the live book can close the gap to the backtest, and treat its recent-fold weakness as the signal to monitor most closely.