The thesis
SilentDip rests on a clean, testable distinction: a stock that falls hard without a negative headline is probably overreacting to noise and will revert, while the same drop carrying genuine bad news tends to keep drifting lower. The strategy therefore demands two conditions together — oversold technicals and the absence of salient negative sentiment — before buying, then exits as the name mean-reverts to its SMA20 or as RSI normalizes. The idea is grounded in recent academic work (Clinch, He, Landsman & Li, SSRN, Nov 2025), itself extending Chan's 2003 study of price reactions to news versus no-news. It trades a focused universe of 24 U.S. large caps (AAPL, MSFT, JPM, XOM, and the like), which keeps liquidity high and headline coverage dense enough to make the "no-news" filter meaningful.
Recent activity
The live book has been quiet to the point of hibernation. Of the last six scheduled runs, five executed nothing at all; the only trade was a buy of 29 shares of BAC at $51 on 14 October. As of 21 October the account sits at roughly $9,991 total against $8,516 in cash — essentially flat, and fractionally below its $10k starting line. This idleness is by design: the entry conditions are strict, so dry spells are expected. It does, however, mean the strategy is largely unproven in the live environment so far, with one open position doing most of the talking.
Backtest and validation
On paper the record is strong. Over 1,279 days the backtest returned 52.62% (8.69% CAGR) at a Sharpe of 1.03, with a contained 10.38% max drawdown and a 70.55% win rate across 294 trades. Turnover is heavy at 4,192%, but fees stayed trivial ($294 total, ~$1 a trade).
More reassuring is the walk-forward validation: all four folds were positive, and performance improved out of sample, with the most recent fold (Jul 2025–Oct 2026) delivering 16.75% at a 1.91 Sharpe and just a 2.39% drawdown. The probabilistic Sharpe ratio is a confident 0.991.
Strengths and risks
The strengths are real: positive folds throughout, a high win rate, and the best risk-adjusted numbers arriving most recently rather than being front-loaded by a lucky early period.
The risks deserve equal billing. The deflated Sharpe ratio is only 0.563 — meaning that once you account for the 42 trials run during development, confidence that the edge is genuine falls well short of the ~0.95 you'd want. The first two folds were weak (Sharpe near 0.35), so the headline numbers lean on two strong recent regimes that favoured mean reversion. And the near-dormant live book offers little independent evidence yet. Promising, grounded, and honestly validated — but not yet confirmed.