The thesis
PanicFade v4 bets that markets overreact to bad news. When a name in its 24-stock large-cap universe gets hit by a negative headline shock (sentiment below -0.4) and is already technically oversold (RSI14 under 35 and trading below its 20-day average), the strategy buys and sizes into the fear. It then exits on reversion — RSI14 back above 50 or price reclaiming the SMA20 — and hard-stops fresh lows on continued negativity, treating those as a genuine fundamental break rather than a panic to fade. The design is grounded in recent sentiment-overreaction literature (arXiv:2509.11970, 2025), which lends it a defensible economic rationale rather than a purely data-mined edge.
Recent activity
The live book has been busy but quiet in P&L terms. Across runs from 30 Oct to 6 Nov, each scheduled session executed one to three trades with zero rejections, and total equity drifted down from $9,956.61 to $9,709.35 — a touch below the $10k Season 2 baseline. Much of that activity was concentrated in CAT: the strategy bought near $622, then sold progressively lower at $613, $586, and $559 across repeated cycles. That pattern is the thesis working as designed but not as hoped — CAT's weakness kept extending, so the stops fired instead of the reversion paying off. GOOGL and NVDA round-trips and a 20-share BAC buy at $50.04 round out the window.
Backtest and validation
Over 1,233 days the backtest returned 37.93% (final equity $13,793), a 6.79% CAGR, with a 0.96 Sharpe, a contained 9.3% max drawdown, and a healthy 62.88% win rate across 530 trades. The validation is the genuine highlight: four walk-forward folds, all four positive, with out-of-sample return of 21.7% and an out-of-sample Sharpe of 2.44 in the most recent fold. The probabilistic Sharpe ratio of 0.985 is strong, and passing a 27-trial gate suggests the result survived meaningful multiple-testing scrutiny.
Strengths and risks
The strengths are real: a coherent, literature-backed thesis; consistent positive folds; a high win rate; and a shallow drawdown profile. But balance demands caution on three fronts. First, the deflated Sharpe of 0.56 is far more sober than the 0.985 PSR — once you adjust for those 27 trials, confidence in a persistent edge is moderate, not overwhelming. Second, the out-of-sample strength leans heavily on fold 4 (Sharpe 2.44 vs 0.19 in fold 1), raising the question of regime dependence. Third, turnover of 5,807% is enormous; even at the modest fee load booked here, this is a strategy whose edge can be eroded by frictions, and the CAT churn is a live example of costs stacking up without reversion.
Verdict: a well-constructed, honestly validated mean-reversion strategy worth running — but one to watch closely, because the gap between its backtest polish and its flat-to-soft live open is exactly where overfitting hides.