The thesis
Bollinger-reversion runs one of the oldest ideas in technical trading: buy weakness below the lower Bollinger band, sell strength above the upper band, and collect the snap back to the mean. It operates on a 24-name large-cap universe spanning tech (AAPL, MSFT, NVDA), financials (JPM, V, MA), staples (PG, KO, COST) and industrials (CAT, HON). It is live today, and the logic is clean enough that its trade log reads like a lesson plan — the June buy of DIS at $99.18 and the August sell at $105.15 is exactly the band-to-band round-trip the thesis promises.
Recent activity
The strategy trades on a scheduled daily cadence, and the last two weeks have been quiet. Of the six most recent runs, four executed nothing while flagging a single rejected order each — a signal was generated but blocked, likely by cash or position constraints. Only 6 October produced a fill (one sell of COST at $934.52). Earlier in the window it rotated more actively: a buy of JNJ (9 shares at $255.36), a sell of CAT, and a chunky 70-share NKE buy at $35.64.
The live paper book is sitting in the $9,500–$9,700 range, with cash swinging between roughly $900 and $2,800 as positions turn over. The run of rejected orders is worth watching — it suggests the model wants to act more often than the book's capital allows.
Backtest and validation
On paper the headline is encouraging: 17.55% total return over 451 days, a 9.46% CAGR, a 63.89% win rate across 76 trades, and modest fees ($76). That is a believable profile for a mean-reversion system — frequent small wins, as the high hit-rate confirms.
The risk side is less comfortable. The Sharpe is a thin 0.66, the max drawdown is a steep 20.57%, and turnover is a punishing 1,720%, meaning the whole book churns roughly seventeen times a year. In live trading that churn is where edge quietly leaks away.
Crucially, the validation gate fails. All four walk-forward folds are positive, which looks reassuring — but they decay almost monotonically: fold 1 returned 7.67% at a 1.78 Sharpe, then 1.42%, then 0.77%, and the most recent fold just 0.40% at a 0.14 Sharpe. The out-of-sample Sharpe of 0.14 is effectively flat. After adjusting for the 6 trials behind this configuration, the deflated Sharpe (DSR) lands at 0.342 — below the bar that would rule out luck — even though the probabilistic Sharpe (PSR) reads an optimistic 0.814.
The verdict
The strength is real: a coherent, well-understood edge with a high win rate and every historical fold in the black. But the shape of those folds tells the honest story — the alpha front-loaded into 2024 and has faded to a rounding error since. Combined with the failed DSR and the heavy turnover, this reads as a strategy whose backtest flatters a mean that recent markets have stopped reverting to. Keep it live and small as a learning signal; don't scale it until an out-of-sample fold earns the confidence back.