FINANCIAR
What PEAD is — why a share price keeps moving after the earnings announcement
The idea, in one paragraph
PEAD (post-earnings announcement drift) is the tendency of a share price to keep moving in the direction of an earnings surprise after announcement day. Better news than expected: the price tends to rise further. Worse news: it tends to fall further. In a perfectly efficient market, public information should enter the price at once and the later move should not be predictable. That is why PEAD was called an anomaly.
Where it comes from: Ball and Brown, 1968
The starting point is “An Empirical Evaluation of Accounting Income Numbers”, published in 1968. Ray Ball and Philip Brown wanted to know something else: whether the profit figure has any value to investors. On 261 US firms, across fiscal years 1957–1965, they found that annual income captures half or more of the information about a firm that appears in a year, but that 85–90% of this content is in the price before the announcement month. They also noticed that the abnormal performance indexes keep going for about a month after the report.
The authors repeated the exercise over half a century later, on daily US data, 1971–2017. A theoretical portfolio long good news and short bad news separates returns by 27.88% in the 360 days before the announcement, by 1.20% on the day and by 2.16% in the following 180 days. The proportions matter: most of the move comes first, and the drift is a small tail. In 16 other markets the drift was positive everywhere and statistically significant in 15.
What a “surprise” is
The surprise is the gap between reported and expected profit. The expectation comes either from a statistical model of past profits or from analyst consensus. The choice is not neutral. When the surprise is stated as a percentage of consensus, a very small base inflates the result. An arithmetic example, not a real firm: consensus of $0.02 a share, result of $0.05, a surprise of 150%, even though the news may be trivial.
Why it might exist: the proposed mechanism
Bernard and Thomas (1989) asked directly whether the drift is a delayed reaction or a risk premium, and their conclusion leaned towards delayed reaction. Their 1990 study went further: prices appear not to reflect fully what current earnings say about future earnings, and the three-day reactions around later announcements are related to the autocorrelation of earnings.
How solid are the links? The surprise and the announcement-day reaction are measured directly. Delayed reaction is a documented interpretation, not an observation: it explains portfolio averages, not what a particular investor thought. The fading of the drift over time is, as shown below, disputed.
Is it dead? The current debate
Martineau (2022) argues that the drift has not existed for large stocks since 2006 and has recently vanished for microcaps too, where the analyst-surprise analysis stops it in 2016. A plausible explanation comes from McLean and Pontiff (2016): across 97 return predictors, results fall by 26% out of sample and by 58% after publication, a sign that investors learn from research.
Others disagree, and the 2026 UCLA Anderson review shows that much of the dispute concerns very small firms. Subrahmanyam (2025) finds, over February 2001 to December 2024, a t-statistic of 2.18 on all stocks and of 1.43 without microcaps. Those are about 3% of market value. In other words, an effect can be statistically real and still hard to use: small firms cost more to trade.
What we found (olivLaw's own results)
This section is not academic literature. It is an internal paper study, with no real money and no external review. We applied the classic rule to the US market: the top decile of surprise against consensus (at least 62.9%, a threshold chosen on 2017–2021 only), a 60-session hold, long only, a minimum price of $5, trading costs included. The rule was sealed before the run. An amendment written after the first run added split cleaning and the equal-weighted benchmark; the first run, without them and against the S&P 500, came out negative.
| Period | Announcements | Against the equal-weighted universe | Against the S&P 500 |
|---|---|---|---|
| Selection, 2017–2021 | 2,931 | +0.94% (t 1.1) | −0.12% (t −0.1) |
| Validation, 2022–2025 | 2,736 | +2.33% (t 3.0) | −1.72% (t −1.8) |
| 2026, eight months | 430 | +7.28% (t 4.7) | +2.41% (t 1.3) |
The second results column matters as much as the first. Against the equal-weighted universe the rule held up in validation, even at double costs (+1.89%, t 2.4). Against the S&P 500, an index dominated by large firms, it lost over the same period. So the result depends on the benchmark.
The limits are serious. The universe contains only firms listed today, so survivors. Prices exclude dividends. The threshold catches many firms with a consensus near zero, which is exactly the small-base trap. In a separate test, simply buying every announcement the day before gave +1.83% (t 4.9) in 2022–2025. The entry differs, so the figures cannot be subtracted, but it suggests part of the gain reflects the period and the announcement premium described by Frazzini and Lamont.
We also tested whether a large surprise can be anticipated, on 63,678 announcements. The model separated cases well (AUC 0.81, where 0.5 is a coin toss), but much of that comes from the small base: without it, AUC falls to 0.71. Trading volume before the announcement predicted almost nothing (AUC 0.49 and 0.51).
How to read a claim about PEAD
| The question | Why it matters | Where to check |
|---|---|---|
| Which firms are included? | Microcaps alone can produce the effect. | Martineau, UCLA Anderson |
| How is the surprise defined? | Percentages on a small base inflate the result. | The study's methodology |
| What is the benchmark? | The same signal can beat one benchmark and lose to another. | The table above |
| Are costs included? | A small drift can vanish after costs. | The study's methodology |
| Is the result from before or after publication? | Published effects tend to shrink. | McLean and Pontiff |
This text explains a concept. It is not a recommendation to buy or sell any share.
Sources
- Ball & Brown (1968), Journal of Accounting Research 6(2), 159–178
- Ball & Brown (2019), „Ball and Brown (1968) After Five Decades” / presentation, Wharton
- Bernard & Thomas (1989), Journal of Accounting Research 27, 1–36
- Bernard & Thomas (1990), Journal of Accounting and Economics 13(4), 305–340
- Martineau (2022), Rest in Peace Post-Earnings Announcement Drift, Critical Finance Review 11(3–4), 613–646
- McLean & Pontiff (2016), Journal of Finance 71(1), 5–32
- UCLA Anderson Review (2026), Is Post-Earnings Announcement Drift a Thing Again?
- Frazzini & Lamont (2007), NBER Working Paper 13090
