The thesis
SilentDip rests on a clean, falsifiable idea: a sharp price drop that arrives without negative headlines is more likely noise than signal, and noise reverts. The same drop carrying genuine bad news tends to keep drifting. So the strategy only buys names that are both oversold on technicals (RSI, distance from SMA20) and free of salient negative sentiment, then exits as they mean-revert toward SMA20 or RSI normalizes.
The idea is sourced from Clinch, He, Landsman & Li (From Drift to Reversal, SSRN working paper, Nov 2025), extending Chan's 2003 work on price reactions to news and no-news. Grounding a signal in the no-news/news asymmetry is a genuinely defensible edge rather than a curve-fit indicator — the thesis predicts when reversion should and should not happen, which is the kind of claim you can test.
Backtest and validation
Over 1,279 trading days the backtest returns 52.62% (CAGR 8.69%), Sharpe 1.03, max drawdown 10.38%, with a 70.55% win rate across 294 trades. The win rate fits a mean-reversion profile: many small wins, with drawdown kept in the low teens.
Walk-forward validation is the more interesting story. All 4 folds were positive, and crucially the quality improved over time — fold 3 (2024–25) returned 26.26% at Sharpe 1.84, and the most recent out-of-sample fold (2025–26) returned 16.75% at Sharpe 1.91 with only a 2.39% drawdown. The Probabilistic Sharpe Ratio is a strong 0.991.
Risks and caveats
Balance demands three warnings. First, the Deflated Sharpe Ratio is only 0.563 — once you penalise for the 42 trials behind this configuration, confidence that the edge is real drops to a coin-flip-plus. PSR and DSR disagreeing this much is the signature of multiple-testing risk.
Second, returns are front-loaded into the recent folds. Folds 1 and 2 (2021–24) returned just 4.96% and 1.91% over roughly 15 months each, at Sharpe ~0.35. Much of the headline 52.62% was earned in the last two years' regime; a flat or whipsawing tape could look a lot more like fold 1.
Third, turnover is 4,192% — this strategy trades constantly, so slippage and fees it never paid in the sim will bite live. The reviewer approved it (risk 0.32) but flagged a non-zero-price guard gap and a budget-vs-equity sizing nuance worth closing.
Live so far
SilentDip went live at the Season 2 reset on 2026-10-29 with $10,000. It has made one trade — 2 shares of CAT at $622.54 on 10-30 — and sat out the other sessions (0 executed). Total equity is $9,967.22, fractionally below starting capital. That is three days of a patient, selective strategy; it tells us nothing yet. The real test is whether the no-news filter holds up when the next quiet dip turns out to be the start of something louder.