Earnings gap statistics: what the data actually says

What the earnings data actually looks like

Earnings day is when a company reports the prior quarter's revenue, profit, and guidance. It happens four times a year per stock, almost always either before the market opens or after it closes — which means the price reaction is delivered as a gap. The size of that gap is what we need to talk about. Most retail traders dramatically underestimate it until they get caught on the wrong side once.

Across the S&P 500, about 70% of stocks move more than 2% on their earnings day. Roughly 15-20% move more than 5%. Mid-cap and small-cap names tend to have even wider reactions because they have less analyst coverage and thinner liquidity. A 10% gap is unremarkable for a small-cap on a bad print. A 20-30% gap is not rare.

Here is the harder truth. The average earnings-day return across the universe is approximately zero. The data is roughly symmetric — about as many big up moves as big down moves over a long sample. So when you hold a stock through earnings expecting a beat, you are taking a high-variance bet with no statistical edge unless you have a real, calibrated, repeatable thesis. 'I like the company' is not a thesis. 'I think they'll beat' is not a thesis. 'Their channel checks suggest revenue is tracking 15% above consensus and the stock is priced for in-line' — that's a thesis, and even then you're playing fat tails.

What to do with this. Most retail traders should default to flat into earnings on positions where they don't have a specific earnings thesis. If you hold long-term and want to ride through, that's a different game (you're an investor on that position, not a trader). If you want to trade earnings reactions, the cleanest plays are usually after the print — fading the open on a stretched gap, riding post-earnings drift after a beat-and-raise, or pairing earnings winners against losers in the same sector.

Recap: ~70% of S&P 500 stocks move more than 2% on earnings day, the distribution is fat-tailed, average return is roughly zero. Without a real thesis, default to flat. The cleaner trades are after the print, not through it.