The Thesis
PanicFade v4 is a contrarian mean-reversion strategy built on a simple behavioural premise: negative headline shocks are routinely overreacted to, and prices tend to revert toward fundamentals once the panic fades. It buys large caps hit by a sentiment reading below −0.4 that are already technically oversold — RSI14 under 35 and trading below their 20-day moving average — and sizes into the weakness. Exits are symmetric and rules-based: it sells on reversion (RSI14 back above 50, or price reclaiming the SMA20) and hard-stops on fresh lows under continued negativity, treating that as evidence of a genuine fundamental break rather than noise.
The approach is grounded in 2025 sentiment-feedback literature (arXiv:2509.11970, corroborated by work in JRFM), which documents asymmetric, retail-driven overreaction to negative media. Confining the universe to 24 mega-cap names — AAPL, MSFT, JPM, UNH, XOM and peers — is a sensible guardrail: reversion logic is far safer in deeply liquid, well-covered stocks than in the small caps where "panic" can prove permanent.
Recent Activity
The live book has been quiet. Of the six scheduled runs between 12 and 19 October, four executed zero trades — the signal simply isn't firing in a calm tape, which is exactly how a panic-fade should behave when there's no panic. The most recent run (19 Oct) left the account at $9,802.98 total with $8,642.40 in cash: heavily defensive and mostly uninvested.
Recent round-trips were mixed. MCD was bought at $231.79 and sold at $242.65 — a clean winner. NKE (bought $36.10, sold $35.53) and HD (bought $299.12, sold $293.95) were small losers, exited on the rules rather than left to rot. That churn is characteristic: backtest turnover is a striking 5,807%, though with roughly $1 of fees per trade the frictional drag stays modest ($530 total across 530 trades).
Validation and Backtest
The numbers that matter most are the walk-forward results. Across four out-of-sample folds, all four were positive — 1.6%, 3.9%, 10.1% and 21.7% — with the most recent fold posting a 2.44 Sharpe at just a 2.6% drawdown. A 0.985 probabilistic Sharpe ratio and a 0.56 deflated Sharpe (which penalises the 27 trials run) suggest the edge is unlikely to be pure overfitting. For a validation gate, that is a genuinely strong pass.
The Balance
Strengths are real: robust validation, a defensible behavioural thesis, disciplined exits, and a tight 9.3% max drawdown. But temper expectations. Full-sample Sharpe is a modest 0.96 and CAGR just 6.79% — this reads as a steady, low-beta complement, not a return engine. The improving fold-by-fold returns could reflect a strengthening edge, or simply a 2025–26 regime that happened to reward dip-buying. And the live paper book, sub-$9,900 and parked largely in cash, is a reminder of the trade-off at the heart of the strategy: one that waits for panic earns nothing while markets stay calm. Patience is the price of admission.