Post-Earnings Announcement Drift: The Anomaly That Refuses to Die

Introduction: The Most Stubborn Anomaly in Finance

In a field obsessed with market efficiency, one anomaly has survived over half a century of academic scrutiny, publication in top journals, and widespread adoption by quantitative funds. Post-earnings announcement drift (PEAD) is the empirically documented tendency for stocks that report positive earnings surprises to continue drifting upward for 60 to 90 days after the announcement, and for stocks that report negative surprises to continue drifting downward.

This is not a subtle effect buried in noise. The original academic studies documented annualized hedge portfolio returns of approximately 25%. Even after decades of publication, arbitrage activity, and structural market changes, PEAD remains statistically significant, particularly in smaller and less-followed stocks. It is one of the most replicated findings in all of empirical finance.

For quantitative traders, PEAD offers a rare combination: a well-understood economic mechanism, robust out-of-sample evidence, and practical implementability. This article will trace the anomaly from its discovery through its modern implementation, with precise citations and real numbers throughout.

The Discovery: Ball and Brown (1968)

The story of PEAD begins with one of the most influential papers in accounting research. In 1968, Ray Ball and Philip Brown published "An Empirical Evaluation of Accounting Income Numbers" in the Journal of Accounting Research (Vol. 6, No. 2, pp. 159-178). Their study examined the relationship between annual earnings announcements and stock price changes for 261 firms listed on the NYSE over the period 1946-1966.

Ball and Brown's key finding was that stock prices began moving in the direction of the earnings surprise well before the announcement date, but that a significant portion of the price adjustment occurred after the announcement. Specifically, they documented that stocks with positive earnings changes continued to earn abnormal returns for at least two months following the announcement.

This finding posed a direct challenge to the semi-strong form of the efficient market hypothesis (EMH), articulated by Eugene Fama in 1970, which holds that stock prices should fully and immediately incorporate all publicly available information. If earnings announcements are public information, prices should adjust instantaneously. The fact that the drift persisted for weeks after the announcement suggested either market inefficiency or an omitted risk factor.

Ball and Brown's paper has been cited over 8,000 times and is widely considered one of the foundational works of positive accounting research. But the systematic quantification of PEAD as a tradeable strategy came two decades later.

Bernard and Thomas (1989): The Definitive Study

Victor Bernard and Jacob Thomas published the definitive study of PEAD in 1989: "Post-Earnings-Announcement Drift: Delayed Price Response or Risk Premium?" in the Journal of Accounting Research (Vol. 27, Supplement, pp. 1-36). This paper transformed PEAD from an observed curiosity into a precisely measured anomaly with a clear economic mechanism.

Bernard and Thomas studied all NYSE and AMEX firms from 1974 to 1986 and sorted them into deciles based on Standardized Unexpected Earnings (SUE). Their central findings:

• Stocks in the top SUE decile (largest positive earnings surprises) outperformed stocks in the bottom SUE decile (largest negative surprises) by approximately 4.2% over the 60 trading days following the earnings announcement.
• The drift was approximately symmetric: positive-surprise stocks drifted up by about 2% and negative-surprise stocks drifted down by about 2% relative to the market.
• The hedge portfolio (long top decile, short bottom decile) generated approximately 25% annualized returns before transaction costs.
• The drift was not explained by risk factors such as beta, firm size, or the January effect.

How SUE Works: Measuring Earnings Surprise

The engine behind PEAD research is the Standardized Unexpected Earnings (SUE) metric, which quantifies the magnitude of an earnings surprise in a standardized, cross-sectionally comparable way.

Where:

• EPS_actual is the reported earnings per share for the quarter.
• EPS_expected is the expected earnings per share, typically derived from either (a) a seasonal random walk model (same quarter last year plus a trend), or (b) the consensus analyst forecast from I/B/E/S or a similar provider.
• σ(historical forecast errors) is the standard deviation of past forecast errors, which normalizes the surprise by the firm's typical variability. This ensures that a $0.05 surprise at a company with historically stable earnings is weighted more heavily than the same $0.05 surprise at a volatile company.

Bernard and Thomas used a seasonal random walk model for expected earnings: E[EPS_q] = EPS_{q-4} + drift, where the drift is the average quarter-over-quarter change from the trailing eight quarters. This approach has the advantage of requiring only historical earnings data and no analyst forecast inputs.

The SUE score is then used to rank stocks cross-sectionally. Stocks in the top SUE decile have the largest positive surprises; stocks in the bottom decile have the largest negative surprises. The PEAD strategy goes long the top decile and short the bottom decile.

Why SUE Magnitude Predicts Drift Magnitude

One of the most important findings from the PEAD literature is that the relationship between SUE and subsequent returns is monotonic: the larger the earnings surprise, the larger the drift. This is not a binary above/below distinction. A stock that beats earnings by 3 standard deviations drifts more than a stock that beats by 1 standard deviation.

Bernard and Thomas (1990), in a follow-up paper titled "Evidence That Stock Prices Do Not Fully Reflect the Implications of Current Earnings for Future Earnings" (Journal of Accounting and Economics, Vol. 13, pp. 305-340), provided additional evidence that the drift is driven by the market's failure to fully incorporate the implications of current earnings for future earnings. They showed that a disproportionate share of the drift concentrates around the subsequent quarterly earnings announcement, suggesting that investors only fully update their expectations when the next quarter's earnings confirm the trend.

Why Does PEAD Persist? Four Explanations

The persistence of PEAD across decades and markets has generated significant academic debate. Four main explanations have been proposed:

1. Investor Underreaction (Behavioral Finance)

The dominant explanation is that investors systematically underreact to earnings news. This is consistent with the broader behavioral finance literature on conservatism bias, documented by Edwards (1968) and applied to financial markets by Barberis, Shleifer, and Vishny (1998) in their model of investor sentiment.

The underreaction hypothesis holds that investors anchor too heavily on their prior beliefs about a company's earnings trajectory and update too slowly in response to new information. When a company reports a positive surprise, investors partially adjust their expectations upward but do not fully incorporate the information. As subsequent data (analyst revisions, management guidance, the next earnings report) confirms the new trajectory, prices gradually catch up.

2. Limits to Arbitrage

Even if sophisticated investors recognize the pattern, several frictions limit their ability to trade it away:

• Short-selling constraints: The negative-surprise side of the trade requires short selling, which is costly (borrowing fees), risky (short squeezes), and prohibited in many institutional mandates.
• Transaction costs: PEAD is strongest in small-cap stocks, which have wider bid-ask spreads and higher market impact costs. The same frictions that allow the anomaly to persist also make it expensive to exploit.
• Holding period: The drift unfolds over 60 to 90 days, during which the position is exposed to idiosyncratic risk that cannot be fully hedged.

Shleifer and Vishny (1997), in their influential paper "The Limits of Arbitrage" (Journal of Finance, Vol. 52, No. 1, pp. 35-55), formalized this argument: even when mispricings are known, risk-averse arbitrageurs with finite capital and short time horizons cannot always correct them.

3. Institutional Frictions

Many institutional investors are constrained by investment mandates, benchmark tracking, and quarterly performance evaluation. These constraints create predictable trading patterns around earnings announcements. Index funds must hold positions regardless of earnings news. Value managers may be slow to sell a stock that misses earnings if it still meets their valuation criteria. These frictions contribute to the delayed price response.

4. Risk-Based Explanation

Some researchers have argued that PEAD represents a rational risk premium rather than a market inefficiency. The idea is that stocks with large positive earnings surprises become riskier (perhaps because they are now priced for continued growth), and the drift is compensation for bearing that risk.

However, Bernard and Thomas (1989) tested and rejected this explanation. They showed that the drift cannot be explained by the CAPM beta, firm size, or the Fama-French factors. Subsequent research by Chordia et al. (2009) has also failed to find a risk-based explanation that fully accounts for the drift. The academic consensus today favors the behavioral explanation, with limits to arbitrage explaining why the anomaly is not quickly traded away.

The Numbers: Historical Returns

Let us look at the specific return magnitudes documented across multiple studies and time periods:

The trend is clear: PEAD returns have declined over time, but have not disappeared. Chordia, Goyal, Sadka, Shivakumar, and Subrahmanyam (2009), in "Liquidity and the Post-Earnings-Announcement Drift" (Journal of Financial Economics), showed that the decline in PEAD is concentrated in large, liquid stocks. For small and mid-cap stocks with low analyst coverage, the drift remained economically significant, with 60-day returns still exceeding 2% for extreme SUE deciles.

Post-Publication Decay: McLean and Pontiff (2016)

A natural question about any published anomaly is whether it survives publication. Once academics document a pattern and practitioners implement it, the resulting arbitrage activity should compress the returns.

R. David McLean and Jeffrey Pontiff addressed this question comprehensively in their 2016 paper "Does Academic Research Destroy Stock Return Predictability?" (Journal of Finance, Vol. 71, No. 1, pp. 5-32). They examined 97 anomalies documented in academic literature and found that, on average, anomaly returns decline by approximately 32% post-publication.

However, the decay is not uniform across all anomalies. Some anomalies disappear entirely after publication, while others persist with reduced magnitude. PEAD falls into the latter category. McLean and Pontiff's analysis showed that earnings-based anomalies in general, and PEAD in particular, have been remarkably persistent compared to other anomalies such as asset growth, accruals, or investment-to-assets.

The persistence of PEAD is attributed to the limits-to-arbitrage factors described above. Even if quantitative funds aggressively trade PEAD, their combined capital is insufficient to fully arbitrage the effect across the thousands of earnings announcements that occur each quarter. The short-selling constraint on the negative-surprise side is particularly binding: it is much harder to push a stock down than to push one up.

Practical Implementation: Trading PEAD

Translating the academic evidence into a practical trading strategy requires several design decisions. Here is a framework based on the empirical literature and quantitative practice:

1. Measuring the Surprise

There are two main approaches to quantifying the earnings surprise:

• SUE (analyst-based): Use the consensus analyst EPS estimate as the expected value and compute SUE = (Actual - Consensus) / historical standard deviation. This is the cleanest approach but requires a consensus estimate feed (Refinitiv I/B/E/S, Bloomberg, or FactSet).
• Gap and volume proxy: When analyst data is unavailable (for microcaps or international stocks), the earnings-day price gap and volume spike serve as proxies for surprise magnitude. A stock that gaps up 8% on 5x average volume is exhibiting price discovery consistent with a large positive surprise.

2. Analyst Coverage as a Signal Amplifier

PEAD is strongest for stocks with low analyst coverage. This finding has been documented by Zhang (2006) in "Information Uncertainty and Stock Returns" (Journal of Finance, Vol. 61, No. 1, pp. 105-137), who showed that information uncertainty amplifies the drift. Stocks followed by fewer than 5 analysts exhibit significantly larger post-earnings drift than stocks followed by 20+ analysts.

The logic is straightforward: low-coverage stocks have less pre-announcement information incorporated into prices, so the earnings announcement itself carries more informational weight. The slower diffusion of information among a smaller investor base also means the underreaction is more pronounced and corrects more slowly.

3. Market Capitalization Effects

Small-cap stocks show stronger PEAD than large caps. Bernard and Thomas (1989) documented this in their original study, and it has been confirmed in every subsequent analysis. The practical implication is that a PEAD strategy tilted toward small and mid-cap stocks will capture more of the effect, but will also face higher transaction costs and lower liquidity. There is a trade-off between signal strength and implementation cost.

4. Regime Sensitivity

PEAD is not equally strong in all market environments. During high-volatility defensive regimes (VIX above 25), the signal-to-noise ratio of individual earnings surprises deteriorates because macro factors dominate stock-specific returns. Effective implementation requires larger surprise thresholds during these periods. A practical approach: require a minimum absolute SUE score of 2.0 in normal regimes and 3.0 or higher in defensive regimes.

5. Holding Period

The academic literature consistently shows that the drift unfolds over 30 to 60 trading days (approximately 6 to 12 calendar weeks). Bernard and Thomas (1990) showed that a disproportionate share of the drift concentrates around the subsequent quarterly earnings announcement (approximately 60 trading days later).

Practical holding period guidelines:

• Minimum hold: 20 trading days (4 weeks). Exiting earlier leaves significant drift on the table.
• Optimal hold: 40-60 trading days (8-12 weeks). This captures the bulk of the documented drift.
• Maximum hold: Exit before or at the next earnings announcement. The next quarter's report either confirms or reverses the signal. Holding through introduces binary event risk.

6. Portfolio Construction

A well-constructed PEAD portfolio typically holds 20-40 positions at any given time (long only or long-short), entered throughout the earnings season as reports come in. Key construction principles:

• Stagger entries: Earnings announcements are clustered in the 3-4 weeks following quarter-end. Enter positions as surprises are identified rather than waiting for the full season to complete.
• Position sizing: Weight positions by SUE rank. Larger surprises get larger allocations. A Kelly-scaled approach, capped at a fractional Kelly fraction (e.g., 0.25), provides a disciplined sizing framework.
• Sector diversification: Avoid concentration in any single sector. Earnings surprises can cluster by industry (e.g., all banks beating on net interest income), and sector concentration transforms an alpha strategy into an unintended sector bet.

Combining PEAD with Insider Signals

One of the most powerful refinements to a PEAD strategy is combining it with insider trading data from SEC Form 4 filings. The intuition is compelling: if a stock reports a large positive earnings surprise and insiders were buying in the weeks before the announcement, the convergence of these two signals is far more informative than either alone.

Lakonishok and Lee (2001) documented that insider purchases predict future returns even after controlling for earnings surprises. The combination works because the two signals capture different dimensions of information: PEAD captures the market's underreaction to public information (the earnings report), while insider buying captures private information about the company's trajectory that may not yet be reflected in the earnings numbers.

Practical convergence criteria:

• Stock reports positive earnings surprise (SUE > 1.5 or gap > 3%)
• One or more insiders filed Form 4 purchases (code P) in the 30 days before or 10 days after the earnings announcement
• Insider purchases are not 10b5-1 plan trades
• Combined dollar conviction exceeds a meaningful threshold relative to market cap

When these conditions are met simultaneously, the historical evidence suggests a significantly amplified drift. This is the approach that Alpha Suite's quantitative engine implements: every signal is evaluated not just on its earnings surprise magnitude, but also on the pattern of insider activity surrounding the announcement.

Critiques and Limitations

No honest treatment of PEAD would be complete without acknowledging its limitations:

• Transaction costs: The strongest PEAD signals come from small-cap stocks where implementation costs are highest. Ng, Rusticus, and Verdi (2008), in "Implications of Transaction Costs for the Post-Earnings Announcement Drift" (Journal of Accounting Research, Vol. 46, No. 3, pp. 661-696), showed that transaction costs significantly reduce but do not eliminate PEAD profits.
• Crowding: As more quantitative funds implement PEAD strategies, the initial reaction to earnings surprises has become faster and more efficient. The "easy" returns (large-cap, high-liquidity names) have largely been arbitraged away.
• Data requirements: Computing SUE accurately requires clean historical earnings data, analyst estimates, and careful handling of stock splits, special items, and accounting changes. Data errors can generate false signals.
• Short-side risk: The short side of the PEAD trade (shorting negative-surprise stocks) is both costlier and riskier than the long side. Short squeezes, borrow costs, and recall risk can all erode returns.
• Earnings quality: Not all earnings surprises are created equal. Surprises driven by revenue growth are more persistent than those driven by cost-cutting or one-time items. Cao and Narayanamoorthy (2012) showed that decomposing the surprise into revenue and expense components improves PEAD predictability.

The Bottom Line: PEAD Is Real, Reduced, but Not Dead

Post-earnings announcement drift is one of the most well-documented phenomena in financial economics. First observed by Ball and Brown in 1968, rigorously quantified by Bernard and Thomas in 1989, and confirmed in dozens of subsequent studies across multiple markets and time periods, PEAD represents a genuine and persistent deviation from market efficiency.

The magnitudes have declined from the 4%+ quarterly hedge returns documented in the original studies to something closer to 1.5-3% in recent decades, concentrated in smaller and less-followed stocks. But for systematic traders who can implement efficiently in the small/mid-cap space, PEAD remains a meaningful source of edge.

The key to modern PEAD implementation is combining the core signal (earnings surprise magnitude) with amplifying factors: low analyst coverage, small market cap, insider buying confirmation, and regime-aware position sizing. This multi-factor approach captures the economic mechanism that drives the drift while managing the implementation challenges that limit its exploitation.

Alpha Suite's PEAD scanner implements exactly this framework. Every earnings announcement is scored using both formal SUE computation (when analyst estimates are available) and gap-volume proxies (for broader coverage). The results are cross-referenced with the insider signal pipeline to identify the highest-conviction opportunities, where the company's own executives are confirming with their wallets what the earnings numbers are saying.