Post-Earnings Announcement Drift

Methodology

Post-Earnings Announcement Drift (PEAD) is one of the oldest anomalies in asset pricing, first documented by Ball and Brown in 1968. After a company beats or misses the consensus estimate, its stock tends to keep moving in that direction, relative to the market, for weeks afterward. That is surprising because the surprise is public the moment it's reported, so an efficient market would price it in right away. This page runs that test on the S&P 500's most recently reported quarter. It is not investment advice.

Surprise %. Each company's most recently reported quarter, (actual EPS − consensus estimate EPS) / |consensus estimate EPS|, comes from Earnings Surprise History. That page is now a one-time snapshot (last refreshed 09/16/2026) and no longer updates weekly. As it ages, fewer report dates fall inside this page's ~100-trading-day price window. The "In Coverage Window" figure above is that count. This page's own price data still refreshes weekly.

Prices. Yahoo Finance daily adjusted closes for the last ~100 trading days, for the S&P 500 plus SPY. That's enough for a +1/+5/+10/+20-day window, and it's the same setup as Relative Strength Leaders/Laggards. For each company with a report date inside the window, the anchor day is the first close on or after the report date, never before, so no pre-announcement price leaks into the test. Excess returns vs. SPY use (1+stock)/(1+SPY) − 1, as on Relative Strength Leaders/Laggards and International vs. US. A company that reported too recently to reach a horizon is left blank for that horizon only.

Tests. The regression chart is the main test: each company's surprise % against its +20-trading-day excess return. Pearson (linear) and Spearman (rank) correlations are both shown, as on Factor Analysis. Because the surprise is public before the return window opens, there is no lookahead bias. The beat-vs-miss comparison below it is a simpler version of the same question: mean and median +20-day excess return for each group, plus a Welch t-test on the difference. This uses a t-test instead of a proportion test because the outcome is a continuous return. Earnings Surprise History uses a two-proportion z-test for its beat/miss rates.

Sector medians are shown only for sectors with at least 5 companies in both the beat and miss groups. Smaller samples are too noisy to be meaningful. Company names and sectors come from Sector Beeswarm.

Prices refresh weekly. Each run recomputes the drift panel from the latest quarter and price window, so there is no accumulating history (same as Spin-Off Performance Tracker).