Hey everyone,
I’ve been backtesting an earnings strangle strategy—selling short-dated (2–9 DTE) strangles right before earnings and closing them the following morning. My expectation was that IV crush would outweigh the directional move, but the results show a negative expectancy.

Even though most trades are small winners, the occasional large move wipes out multiple gains, creating a fat-tailed P&L distribution. It’s similar to what Euan Sinclair calls “selling insurance without properly pricing the tail.”

I’ve tested across multiple tickers and years, with deltas around ±16–20, and even when filtering for higher IV percentile, the outcome doesn’t improve much. It looks like the mean is dragged down by rare, extreme post-earnings moves—essentially a short-volatility strategy that’s under-compensated for tail risk.

Would love input from anyone who’s tackled this. Have you found ways to hedge or reshape the payoff curve—like using ratio spreads, wings, or adaptive position sizing—to make the distribution more Gaussian or reduce tail losses?

I’ve attached the backtest JSON if anyone wants to explore the data. Appreciate any insights or similar experiences!