Backtesting 0DTE Options Strategies

Zero days to expiration (0DTE) options have exploded in popularity, and correspondingly, there's an ocean of overfit backtests claiming huge returns. The vast majority of these backtests are wrong in ways that can't be recovered from — the math simply doesn't work in real markets.

The first problem is intraday data. A daily bar backtest cannot handle 0DTE. You need at least minute-level options chains including bid/ask for every strike throughout the day. Getting this data historically is expensive and error-prone. Free sources don't exist. Expect to pay for CBOE or similar quality data.

The second problem is gamma. 0DTE options have massive gamma near expiration — small underlying moves produce huge option moves. Your backtest must model this properly by pricing options from the actual implied volatility surface, not from Black-Scholes with a constant vol input.

Slippage on 0DTE is brutal. Even highly liquid SPX 0DTE options have wide effective spreads once you account for market impact. A common realistic assumption: buy at the ask, sell at the bid, then subtract another few pennies for adverse selection. This eats most theoretical edges.

Finally, be extremely skeptical of any 0DTE strategy that requires precision timing. If the backtest requires entering "at 9:35 AM" or "when VIX crosses 15.2" — the real world doesn't give you those exact prices. Assume you'll enter 15 minutes late at a worse price. If the strategy still works with that haircut, maybe. If not, you were curve-fitting.