The headline test
Does the earnings gap predict the next move?
No — not in a way you could trade. When a stock gaps up or down on its earnings report, the size of that gap has no reliable relationship with where the stock goes over the next 1, 5, 20, or 60 trading days, once the analysis is done honestly.
What “done honestly” means here
- Excess returns. Forward returns are measured relative to the market, so a “signal” that’s really just the whole market rising doesn’t count.
- Out-of-sample. The sample is split, so a pattern that only shows up in the data it was found in gets caught.
- Costs. A realistic transaction-cost assumption is applied before anything is called tradeable.
- Multiple testing. The result is judged against how many cuts of the data were checked.
Apply all four and the apparent edge in the raw numbers is gone. The initial gap turns out to be almost entirely general market movement, not something gap-specific.
Why publish a null result
“This popular-sounding setup doesn’t work” is as useful as a positive finding, and it’s reported here as plainly as one would be. The data-building work — identifying the true reaction day when vendor timestamps are ambiguous — is the same either way.
Look at any company’s history
The company explorer shows how any S&P 500 stock has behaved around its earnings — the gap, the drift, the volatility — with a continuation-versus-reversal scatter for each name.
Open the explorer →Common questions
So the earnings gap tells you nothing about what comes next?
That's the finding here. In raw data, big gaps look like they're followed by continued drift — but most of that is just the overall market moving, and it disappears once you measure returns as excess over the market, split the sample for out-of-sample testing, and subtract realistic trading costs.
Isn't there a well-known post-earnings drift anomaly?
Post-earnings announcement drift is about the earnings surprise — actual EPS versus what analysts expected — and it's strongest in small, illiquid stocks. This test asks something narrower and more practical: does the price gap alone, without the surprise data, carry that information for large-cap stocks? Largely, no.
What about specific sectors or gap sizes?
The analysis segments by sector, gap size, and market-cap tercile. No subgroup shows a robust, tradeable effect that survives the corrections applied to the whole sample.
Educational research, not advice. Past performance does not predict future results.