The thesis
Bollinger-reversion is a textbook mean-reversion play: buy when price closes below the lower Bollinger band, sell when it pushes above the upper band. It runs on a 24-name large-cap universe spanning tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), staples (PG, KO, WMT, COST) and healthcare (JNJ, UNH, PFE, ABBV). The premise is simple and defensible — liquid mega-caps tend to snap back from short-term extremes — but simplicity is no guarantee of edge.
The backtest
On paper, the numbers flatter. Over 451 days the strategy returned 17.55% (9.46% CAGR), lifting a notional $10k to $11,755. Win rate is a respectable 63.89% across 76 trades, and the Sharpe of 0.66 is modest but positive. The catch sits in the risk column: a 20.57% max drawdown is steep for a strategy leaning on staples, and turnover of 1,721% means the book churns roughly seventeen times over. Fees stayed trivial ($76), so cost isn't the problem — capacity for whipsaw is.
Why validation says no
This is where the story turns. Walk-forward testing produced four folds, and all four were positive — encouraging. But the trend is unmistakably downward:
- Fold 1 (Aug 2024–Jan 2025): +7.67%, Sharpe 1.78
- Fold 2: +1.42%, Sharpe 0.25, with a 20.56% drawdown
- Fold 3: +0.77%, Sharpe 0.20
- Fold 4 (Dec 2025–May 2026): +0.40%, Sharpe 0.14
The most recent out-of-sample window returned just 0.4% at a Sharpe of 0.14 — effectively flat. The Deflated Sharpe Ratio lands at 0.342 against a Probabilistic Sharpe of 0.814, and with six trials in the search, the framework marks the strategy failed. The reading is straightforward: the headline 17.55% is front-loaded by an early regime, and the edge has decayed toward zero as the sample moved forward. That is the classic signature of a strategy fit more to history than to a durable market inefficiency.
Live activity
Live behaviour reinforces the caution. The last executed trades were in early August (a single CAT buy, a DIS round-trip). Every scheduled run since — six sessions from 24–31 August — logged zero executions and one-to-two rejections. Portfolio value has drifted from $9,816 down to $9,663, sitting below the $10k line with $2,752 idle in cash. The band conditions simply aren't triggering clean fills, and the equity curve is bleeding gently rather than compounding.
Verdict
Strengths are real: a coherent thesis, a high win rate, and four consecutive positive folds. But the risks dominate — punishing drawdowns, extreme turnover, monotonic out-of-sample decay, and a validation failure that the numbers earn honestly. Bollinger-reversion looks less like an alpha engine and more like a strategy that captured one favourable stretch and has been coasting on its memory since. It stays worth watching, but not worth trusting until an out-of-sample window earns its keep.