Thesis
Bollinger Reversion operates on one of the oldest ideas in technical trading: prices tend to snap back toward their mean after extreme deviations. The rule is deliberately simple — buy any name in the 24-stock universe when it closes below its lower Bollinger Band, and exit when it climbs above the upper band. No earnings filters, no sector rotation, no macro overlay. That simplicity is both its appeal and its principal risk.
The universe skews defensive and liquid: mega-cap tech (AAPL, MSFT, NVDA, GOOGL), financials (JPM, BAC, V, MA), healthcare (JNJ, UNH, PFE, ABBV), staples (PG, KO, WMT, COST), and a handful of industrials and consumer names. All names are large-cap and highly liquid, which keeps slippage manageable at the 76-trade, $1-per-trade fee structure reflected in the backtest.
Backtest Performance
Over a 451-day run ending in late May 2026, the strategy compounded a 17.55% total return (9.46% CAGR) on a starting equity of $10,000, finishing at $11,755. The win rate of 63.9% is solid for a mean-reversion system, and a Sharpe of 0.66 is acceptable in isolation.
The headline concern is maximum drawdown of 20.57% — steep for a strategy trading only large-cap blue chips. Fold 2 alone (Jan–Jul 2025) produced a 20.56% drawdown on just 11 trades, suggesting the strategy got caught badly during a trend-driven stretch where reversion signals repeatedly failed to resolve. Turnover of 1,721% annualised is also worth noting: this system churns the portfolio aggressively, amplifying the sensitivity of net returns to execution costs.
Walk-Forward Validation: The Hard Truth
Validation failed, and the fold-by-fold breakdown explains why cleanly:
| Fold | Period | Return | Sharpe | Max DD |
|---|---|---|---|---|
| 1 | Aug 2024 – Jan 2025 | 7.67% | 1.78 | 7.06% |
| 2 | Jan – Jul 2025 | 1.42% | 0.25 | 20.56% |
| 3 | Jul – Dec 2025 | 0.77% | 0.20 | 6.66% |
| 4 | Dec 2025 – May 2026 | 0.40% | 0.14 | 11.19% |
The decay is systematic. Fold 1 — the oldest period — accounts for the lion's share of aggregate alpha (Sharpe 1.78). Every subsequent fold degraded, with the most recent out-of-sample window delivering just 0.40% return and a Sharpe of 0.14. The Deflated Sharpe Ratio (DSR) of 0.342, computed across six trials, is particularly damning: it adjusts for multiple-testing bias and finds that the probability the true Sharpe exceeds zero is only about 34%. The PSR of 0.814 is somewhat more encouraging but cannot override the directional evidence of the fold series.
The simplest interpretation: the strategy found a regime in mid-2024 where mean-reversion paid reliably, and that regime has since faded.
Recent Activity
The past week has been quiet. Two trades executed (DIS bought at $99.18 on June 10; PG sold at $148.40 on June 9 after being bought at $140.03 on June 1 — a clean 5.9% gain in eight days), while three signals were rejected, likely due to position-sizing or cash constraints. Portfolio equity sits near $10,128 as of June 11, modestly above the $10,000 starting value for the current live run.
The consumer-staples cluster (WMT, COST, PG) drew notable attention in late May and early June, consistent with a market where defensive names saw band touches amid broader volatility.
Verdict
Bollinger Reversion has a genuine edge in its historical window, and the mechanics are sound. But the walk-forward evidence points to overfitting on a specific early regime, with live performance already tracking toward the weakest fold projections. Before scaling capital, the strategy warrants a parameter sensitivity pass (band width, lookback period) and a regime filter to reduce exposure during sustained trending markets — the exact environment where reversion signals become noise.