The thesis
bollinger-reversion runs one of the oldest ideas in technical trading: buy weakness, sell strength. When a name in its 24-stock universe — mega-cap tech, financials, staples, and a few industrials — dips below its lower Bollinger band, the strategy buys; when price pushes above the upper band, it sells. The universe is deliberately liquid and boring (AAPL, MSFT, JPM, PG, KO, WMT, CAT and friends), which suits a reversion approach that needs orderly, range-bound behaviour rather than runaway trends.
What the backtest says
On paper, the numbers are inviting. Over 451 trading days the strategy returned 17.55% (a 9.46% CAGR), finishing at $11,755 on a $10k base. The 63.89% win rate across 76 trades is genuinely high, and folds were positive in all four validation windows.
The caveats sit right next to those wins. The Sharpe ratio of 0.66 is modest — return came with a bumpy ride, including a 20.57% max drawdown. Turnover of 1,720% is heavy churn for a $76 fee drag, and it signals a strategy that trades often enough that execution assumptions matter a lot.
Why validation failed
This is the part that deserves the most attention. The walk-forward validation did not pass, and the reason is a textbook decay pattern:
| Fold | Return | Sharpe |
|---|---|---|
| 1 (Aug 24–Jan 25) | 7.67% | 1.78 |
| 2 (Jan–Jul 25) | 1.42% | 0.25 |
| 3 (Jul–Dec 25) | 0.77% | 0.20 |
| 4 (Dec 25–May 26) | 0.40% | 0.14 |
The edge is almost entirely front-loaded. Out-of-sample return collapses to 0.4% with a Sharpe of 0.14 in the most recent fold. The Deflated Sharpe Ratio of 0.342 — well below the 0.5 confidence threshold — tells the same story once you account for the 6 trials behind this configuration: the headline Sharpe is not distinguishable from luck. The Probabilistic Sharpe Ratio of 0.814 looks friendlier, but DSR is the honest number here.
Live activity: quiet, and slightly red
The live paper account reflects that fading edge. The last executed trades were on 7 August (a CAT buy, a DIS round-trip). Every scheduled run since — 31 Aug through 7 Sep — reports "0 executed, 2 rejected", with the book frozen at $2,752 cash and total equity drifting between $9,638 and $9,740. In other words, the strategy is currently underwater against its $10k start and can't get its recent signals filled.
Verdict
bollinger-reversion is a well-behaved idea with a real historical win rate, but it carries two honest warning flags: a validation failure driven by out-of-sample decay, and a live account that is both underwater and struggling to execute. The high win rate is a strength; the shrinking per-fold Sharpe and the stalled order flow are the risks. This is one to keep on a short leash and watch for whether reversion conditions return — not one to size up.