The thesis
PanicFade v4 bets that markets overreact to bad news. When a name gets hit by negative sentiment (below -0.4) and is already technically oversold — RSI14 under 35 and trading below its 20-day SMA — the strategy buys into the fear and sizes into the panic. It exits on reversion (RSI14 back above 50 or price reclaiming the SMA20) and hard-stops on fresh lows under continued negativity, treating that as a genuine fundamental break rather than noise. The universe is deliberately boring: 24 mega-cap, liquid names (AAPL, MSFT, JPM, UNH, XOM and peers), which keeps the overreaction-reversion mechanism cleaner than it would be in small caps. The approach is grounded in recent sentiment-feedback literature (arXiv:2509.11970, 2025) on asymmetric, mean-reverting responses to negative media.
Recent activity
The live account has been quiet. Across the last week of scheduled runs (Oct 19–26) most days executed zero trades — unsurprising for a strategy that only acts when a specific oversold-plus-negative-sentiment setup appears. The one burst of activity was Oct 23, when it bought 3 GOOGL at $321.67 and 1 CAT at $688.79. Earlier in October it rotated through BAC (bought 23, having sold 21 on Oct 1), trimmed MCD, HD, NKE and UNH, and re-entered UNH and MCD. The paper portfolio's total value has hovered narrowly around $9,780–$9,820, with cash swinging between roughly $6,978 and $8,642 as positions open and close.
Validation and performance
This is where PanicFade earns attention. The four-fold walk-forward validation passed cleanly: all four folds were positive, and performance improved across time — 1.6%, 3.88%, 10.14% and 21.7% — with the most recent out-of-sample fold also posting the best Sharpe (2.44) and smallest drawdown (2.6%). A Probabilistic Sharpe Ratio of 0.985 is strong, and the full backtest shows a 62.88% win rate over 530 trades with a contained 9.3% max drawdown.
The risks
Balance demands scepticism. The headline 37.93% return is spread over 1,233 days, so the CAGR is a modest 6.79%, and the overall Sharpe of 0.96 sits below 1. The Deflated Sharpe Ratio is only 0.56 — after accounting for 27 trials, the edge is plausible but far from certain, a real multiple-testing caveat. Turnover is extreme at 5,807%, meaning the strategy trades constantly; its live-quote-only cash accounting and per-trade fees make it genuinely execution- and cost-sensitive. Fold 1, spanning the 2022 drawdown, barely cleared zero (1.6%, Sharpe 0.19) — a reminder that "fade the panic" struggles when the panic is justified. And the live account is currently flat-to-slightly-soft, so the backtest's strength has yet to prove itself in production.
Verdict
PanicFade v4 is a well-constructed, literature-grounded mean-reversion strategy with unusually honest validation. The evidence is encouraging but not conclusive: a borderline DSR, sub-1 Sharpe, and punishing turnover are the price of admission. Worth running live — and worth watching closely.