The thesis
SilentDip rests on a clean, falsifiable idea: a sharp drop without a negative headline is usually noise and tends to revert, while the same drop with genuine bad news keeps drifting lower. So the strategy only buys when two conditions line up — oversold technicals and an absence of salient negative sentiment — then exits as the name mean-reverts toward its SMA20 or RSI normalizes. It draws on Clinch, He, Landsman & Li (SSRN working paper, Nov 2025), which extends Chan's classic 2003 work on price reactions to news versus no-news. The academic lineage matters here: the edge is a documented behavioural asymmetry, not a curve-fit artifact.
The universe is deliberately boring — 24 US large caps (AAPL, MSFT, JPM, JNJ, XOM and peers). Liquid, well-covered names are exactly where a "no salient news" classifier is most trustworthy, so the universe choice is consistent with the thesis rather than a reach for volatility.
Validation
This is the strongest part of the file. Across four walk-forward folds spanning Sep 2021 to Oct 2026, all four were positive (+4.96%, +1.91%, +26.26%, +16.75%), and the out-of-sample Sharpe in the most recent folds climbed to 1.84 and 1.91. The Probabilistic Sharpe Ratio of 0.991 is excellent. Crucially, the report also carries a Deflated Sharpe Ratio of 0.563 across 42 trials — and that is the number to respect. After penalising for selection across all those trials, the deflated edge is still positive, but only modestly above the 0.5 breakeven. The strategy survives the multiple-testing haircut; it does not sail through it.
Performance, in context
The headline +52.62% total return is real but needs its timeframe: over 1,279 trading days that is a CAGR of just 8.69%, with a full-sample Sharpe of 1.03 and a contained 10.38% max drawdown. A 70.55% win rate across 294 trades is attractive and fits a mean-reversion profile — many small wins. Turnover of ~4,192% is high, though fees total only $294, so frictions look manageable in backtest.
The live reality
Here balance demands honesty. The live paper book sits at roughly $9,985 against a $10k start — fractionally underwater — with $8,515 in cash. The only recent position is a single BAC buy (29 shares @ $51 on Oct 14), currently near flat. The last six scheduled runs (Oct 20–27) each executed zero trades: the signal simply isn't firing. That is arguably the system working as designed — a quiet market offers few qualifying quiet dips — but it means the live record is too thin to confirm the backtest.
Verdict
SilentDip is one of the better-validated strategies in the lab: an academically grounded thesis, four positive folds, and a deflated edge that survives 42 trials. The risks are equally clear — a CAGR under 9%, a DSR only narrowly above breakeven, and a dependency on the sentiment filter correctly distinguishing "no news" from "unflagged bad news." For now it is mostly cash and mostly patience. The signal to watch is not the next green day but the next real selloff — that is when this thesis earns its keep.