The thesis
Bollinger-reversion is textbook mean reversion: buy when price closes below the lower Bollinger band, sell when it pushes above the upper band. It trades a 24-name large-cap universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (PG, KO, COST) and industrials (CAT, HON). The logic is interpretable and the hypothesis is honest — bands stretch, bands snap back. The question, as always, is whether that snap-back survives contact with real walk-forward data.
Backtest: the headline looks good
Over 451 days the strategy returned 17.55% (final equity $11,755), a 9.46% CAGR, with a 63.89% win rate across 76 trades. Winning nearly two trades in three is genuinely attractive for a reversion book. But the supporting cast is weaker: the Sharpe is a middling 0.66, max drawdown reaches 20.57%, and turnover is a punishing 1,720%. That churn matters — reversion signals fire often, and even at a modest $76 in total fees here, high turnover is exactly where live slippage quietly erodes a clean backtest.
Validation: the gate says no
This is where the picture turns. Validation failed. All four walk-forward folds are positive — a point in the strategy's favour — but the trend is unmistakable decay:
| Fold | Return | Sharpe | Max DD |
|---|---|---|---|
| 1 | 7.67% | 1.78 | 7.06% |
| 2 | 1.42% | 0.25 | 20.56% |
| 3 | 0.77% | 0.20 | 6.66% |
| 4 | 0.40% | 0.14 | 11.19% |
The most recent out-of-sample fold earned 0.4% at a Sharpe of 0.14 — effectively flat. The deflated Sharpe ratio (DSR 0.342), which penalises the 6 trials run during search, confirms the suspicion: the glossy 17.55% is largely an in-sample artefact. PSR of 0.814 is respectable, but DSR is the number that counts once you account for selection, and it is low.
Live activity: flat and often blocked
Live behaviour echoes the validation warning. The paper book sits at roughly $9,971 total — marginally below its $10k start — after recent scheduled runs. Execution is choppy: of the last six runs, several rejected every order they attempted (0 executed, 1–2 rejected), likely cash or sizing constraints. Recent fills include buying CAT (3 @ $737) and NVDA (2 @ $217.62) against sells of MCD, NKE and COST.
The balance
Strengths: a clear, defensible thesis, a strong win rate, and four consecutive positive folds. Risks: the strategy is failing its own validation gate, out-of-sample edge has eroded to near zero, drawdowns are heavy, and turnover is a structural headwind. Bollinger-reversion is a credible idea that the data does not yet credit with a durable edge. It belongs in observation, not conviction.