The thesis
bollinger-reversion runs one of the oldest ideas in technical trading: buy weakness, sell strength. It goes long when price closes below the lower Bollinger band and exits when price pushes above the upper band, betting that stretched moves snap back to the mean. The universe is 24 US large-caps spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (PG, KO, WMT, COST) and industrials (CAT, HON) — liquid names where mean-reversion tends to behave.
Backtest performance
Over 451 days the strategy returned 17.55% (9.46% CAGR), ending at $11,755 on a $10k base. The headline numbers flatter it: a 63.89% win rate across 76 trades is genuinely high, and the strategy makes money more often than not. But the quality metrics temper the enthusiasm. A Sharpe of 0.66 is mediocre, and a max drawdown of 20.57% means investors sat through a one-fifth equity loss to earn that return. Turnover is a striking 1,721% — this is an active strategy that churns its capital many times over, though fees stayed modest at $76 total.
The validation problem
This is where the story turns. Walk-forward validation failed the gate, and the fold-by-fold breakdown shows exactly why:
| Fold | Window | Return | Sharpe |
|---|---|---|---|
| 1 | Aug 2024–Jan 2025 | 7.67% | 1.78 |
| 2 | Jan–Jul 2025 | 1.42% | 0.25 |
| 3 | Jul–Dec 2025 | 0.77% | 0.20 |
| 4 | Dec 2025–May 2026 | 0.40% | 0.14 |
All four folds are positive — a small point in its favour — but the trend is a steady bleed toward zero. The most recent out-of-sample fold earned just 0.40% at a 0.14 Sharpe. The Probabilistic Sharpe Ratio (0.814) looks reassuring, but the Deflated Sharpe Ratio, which penalises for the 6 trials run, drops to 0.342 — below conviction. The pattern is a familiar one: an edge that was real in 2024 has been arbitraged or regime-shifted away.
Live activity: mostly idle
Live behaviour echoes the concern. The last executed trades were in early August (a CAT buy, a DIS round-trip) and early June. Since 21 August, every scheduled run has executed 0 orders while rejecting 1–2 — cash is frozen at $2,752.28 and total equity has drifted between $9,645 and $9,816, sitting below the $10k starting line. The repeated rejections suggest the strategy keeps generating signals that fail risk or sizing checks, leaving it effectively parked.
Verdict
Strengths: a coherent, well-understood thesis, a high win rate, and consistently positive folds. Risks: deep drawdowns, a decaying out-of-sample edge, a failed validation gate, and a live book that is currently stuck and slightly underwater. bollinger-reversion is a credible research idea that has not earned a capital allocation — worth keeping on the bench and re-testing if band parameters or the risk filters are retuned, but not one to lean on today.