The thesis
Bollinger-reversion is a textbook mean-reversion play: buy when price trades below the lower Bollinger band, sell when it pushes above the upper band. It works a 24-name large-cap universe spanning tech, financials, healthcare, staples and energy — AAPL, MSFT, NVDA, JPM, XOM, COST and the rest. The bet is simple and old: short-term dislocations from a moving-average envelope tend to snap back.
Recent activity
The last two weeks show the thesis in miniature. On 2 November the strategy bought MSFT at $514.53 and sold it on 4 November at $557.86; it bought AAPL at $329.82 on 3 November and sold at $352.27 the next day. Both were clean, quick reversion round-trips — exactly what the design is meant to capture. It has since added XOM (15 shares at $166.26) and NVDA (13 shares at $184.15).
Scheduled runs have been steady but modest. The 9 November run executed nothing and rejected one order, leaving total equity at $9,911.34 — just under the $10,000 Season 2 starting line. Live, the strategy is treading water.
Backtest and validation
Over 451 days the backtest returned 17.55% (9.46% CAGR), with a 63.89% win rate across 76 trades. The headline looks healthy; the caveats sit underneath it.
- Sharpe is only 0.66, and the 20.57% max drawdown is larger than the full-period return — this was not a smooth ride.
- Turnover is 1,720%, so the edge has to survive heavy churn (total fees came to $76).
Most important, validation failed. All four walk-forward folds were positive, which is genuinely encouraging — but the returns decay monotonically: 7.67% → 1.42% → 0.77% → 0.40%, with fold Sharpe sliding from 1.78 to 0.14. The most recent out-of-sample window returned just 0.40% at a 0.14 Sharpe. The probabilistic Sharpe ratio is a reassuring 0.814, but the deflated Sharpe ratio — which penalises the strategy for the 6 trials behind it — is only 0.342, below the threshold we trust before auto-deploying anything.
Verdict
The strategy is coherent, and its recent MSFT and AAPL round-trips show the mechanism working as intended. But the overall picture is of an edge that was strong in late 2024 and has faded with every subsequent fold. Positive-in-every-fold is a real point in its favour; monotonic decay and a sub-0.5 deflated Sharpe are the arguments against trusting it blindly. Keep it live on a short leash and judge it by what the out-of-sample windows keep telling us — not by the flattering headline return.