Event risk
Does buying stock before earnings work?
No — not as an edge you can count on. Holding a stock into its earnings report is a bet on a single overnight move that, across 10,603S&P 500 earnings events in this dataset, has gone up 51.8% of the time and down 47.4% of the time, averaging out to +0.08%— economically indistinguishable from a coin flip. And once that move happens, the days and weeks afterward don’t reliably make up for the risk either.
What you’re actually risking
The average earnings-day gap, up or down, is 3.8% in absolute terms (median 2.5%). About a quarter of all events (25.3%) land in the “big” buckets — a move of more than 5% in either direction — using the same gap definition and reaction-day detection described in the methodology. That’s the size of the bet you’re placing by holding through the report, regardless of which way it goes.
The direction is close to a coin flip
Of those 10,603 events, 5,495 gapped up, 5,021 gapped down, and 87 landed exactly flat. The average signed move — +0.08% — is small enough that it’s more plausibly explained by this dataset’s five-year window skewing toward a bull market (see the methodology’s caveats) than by earnings reports having a genuine net-positive bias. There’s no evidence here that simply owning a stock through its report is a positive-expectancy trade on its own.
The days after don’t rescue it
The obvious rebuttal is that the initial gap is only half the story — maybe the follow-through in the days and weeks afterward makes up for the coin-flip risk on report day. This project tested exactly that, and it doesn’t: once forward returns are measured against the market instead of in raw terms, only 4 of 20 bucket × horizon combinations stay significant, and none of them replicate in an out-of-sample test. Nothing survives a realistic transaction-cost assumption either. See the full null-result test for the details.
See the full statistical breakdown
The findings page has the complete gap-bucket, sector, and market-cap analysis — raw returns, market-adjusted excess returns, and the out-of-sample and cost checks behind the numbers above.
Read the findings →Common questions
If I'm confident about a beat, isn't holding through earnings still worth it?
The size of an earnings surprise does correlate with the size of the gap that follows it (Spearman ρ = 0.35 in-sample, 0.30 out-of-sample, both highly significant) — so being right about a beat does show up in the reaction. What it doesn't do is predict what happens after that gap: the correlation between surprise size and forward excess returns is much weaker (ρ in the 0.01–0.09 range) and flips significance depending on which period you test. Being right about the beat mostly pays off (or doesn't) in the gap itself, not in a further edge afterward.
Does the risk differ by sector?
Yes, substantially. Information Technology has averaged a 5.9% absolute earnings-day move in this dataset, versus 1.2% for Utilities. See the full sector breakdown for the complete ranking — but a bigger average move doesn't mean a more predictable one.
Can I use the options market to size the risk instead of guessing?
Yes — the options market prices its own expected move via the at-the-money straddle on the nearest expiration after the report, and this site shows a live snapshot of that for upcoming earnings in the company explorer. One honest caveat: free historical options data isn't available, so there's no way to check here whether that implied number has actually been well-calibrated over time. It's shown as a live estimate, not a backtested one.
Educational research, not advice. Past performance does not predict future results.