Thesis and Universe
Bollinger-reversion is a textbook mean-reversion strategy: buy when a stock closes below its lower Bollinger band (statistically oversold), sell when it closes above the upper band (statistically overbought). The universe spans 24 large-cap and mega-cap U.S. names across technology, financials, healthcare, consumer staples, energy, and industrials — household tickers like AAPL, NVDA, JPM, and JNJ. The breadth is deliberate: the more liquid the name, the more reliably prices snap back to the mean rather than trending away from it.
Backtest Snapshot
Over 451 days, the strategy executed 76 trades and compounded to a 17.55% total return (9.46% CAGR), finishing with a simulated equity of $11,755.25. A 63.89% win rate is solid for a reversion approach, and total fees came to exactly $76 — one dollar per trade, implying flat-rate commission accounting.
The drawback is visible in the risk numbers. A 20.57% maximum drawdown is steep for a strategy operating on blue-chip names, and a Sharpe ratio of 0.66 suggests the return per unit of volatility is modest. Turnover at 1,720% annualised is high, meaning transaction costs in a live brokerage account would meaningfully compress net returns beyond what the backtest captures.
Walk-Forward Validation: A Cautionary Signal
The validation framework ran four sequential folds — and the trend is the most important data point in this report.
| Fold | Period | Return | Sharpe | Max DD | Trades |
|---|---|---|---|---|---|
| 1 | Aug 2024 – Jan 2025 | +7.67% | 1.78 | 7.06% | 24 |
| 2 | Jan – Jul 2025 | +1.42% | 0.25 | 20.56% | 11 |
| 3 | Jul – Dec 2025 | +0.77% | 0.20 | 6.66% | 18 |
| 4 | Dec 2025 – May 2026 | +0.40% | 0.14 | 11.19% | 20 |
Fold 1 is genuinely impressive — 7.67% return with a 1.78 Sharpe and a shallow 7% drawdown. But each subsequent fold erodes that edge systematically. The out-of-sample return lands at 0.40% with a Sharpe of 0.14, essentially indistinguishable from cash. The Probabilistic Sharpe Ratio (PSR) of 0.814 is acceptable, but the Deflated Sharpe Ratio (DSR) of 0.342 — which adjusts for the number of trials (6) and non-normality — tells a harder story: after correcting for multiple testing, the edge is not statistically reliable. The strategy correctly did not pass validation.
The most likely explanation is regime shift. Fold 1 coincided with a choppy, range-bound market where reversion thrives. The subsequent folds captured a period of stronger directional moves — both the 2025 correction and the recovery — where oversold readings continued lower before reverting, punishing the entry rules.
Recent Activity: Dormant but Watching
The past week of scheduled runs (July 3–10) produced zero executed trades against a handful of rejections each day, with the paper portfolio sitting between $9,710 and $9,893. The last confirmed executions were in early June: buys in DIS, PG, GOOGL, WMT, and COST, and a subsequent PG sell at $148.40 for a ~6% gain on the position opened at $140.03. That PG round-trip is a clean illustration of the thesis working as designed.
The silence since then suggests current prices on the universe are not breaching the bands in either direction — a neutral-to-trending environment where this strategy appropriately sits on its hands.
Strengths and Risks
Strengths: High win rate, liquid universe, disciplined entry/exit logic, and strong fold-1 performance suggest the edge is real under the right conditions.
Risks: The fold-by-fold decay is a structural red flag. High turnover amplifies live-trading friction. A 20%+ max drawdown on large-caps indicates the strategy can be caught on the wrong side of sustained downtrends. DSR below 0.5 means statistically, we cannot yet rule out luck.
The strategy is worth monitoring through further live paper-trading cycles, particularly to observe whether the mean-reversion edge re-emerges if market conditions rotate back toward range-bound behavior. It is not yet a candidate for capital allocation.