TL;DR. A company can beat the consensus analyst estimate and still sell off the next morning. The reason is a quieter forecast called the whisper number, plus a long-documented pattern called post-earnings announcement drift (PEAD) — the tendency for a stock to keep moving in the direction of its earnings surprise for weeks after the print. Together, they explain a lot of the “beat-and-fade” behaviour traders see every earnings season.
Earnings season in one paragraph
Public companies in the U.S. report results once a quarter. Sell-side analysts publish estimates — revenue, earnings per share (EPS), sometimes free cash flow — and data providers average those into a consensus estimate. As FINRA puts it, “a company beats (exceeds) the estimate or misses (falls short of) the estimate based on how actual earnings or revenues compare to the consensus,” and the stock can “jump or tumble after investors compare the actual results to estimates” (FINRA, “What Is Earnings Season?”). What that summary leaves out is that the consensus is not the only number traders are watching.
What a whisper number actually is
A whisper number is the unofficial EPS or revenue figure that desks, buy-side analysts, and informed clients quietly converge on in the hours and days before a release. It is not in the FactSet or Refinitiv consensus — it lives in client notes, in options pricing, and in the tone of pre-print research.
The distinction matters because whisper numbers are usually closer to the truth. One long-running provider that aggregates whispers reports that its estimates “missed actual reported earnings by 21%” against an average miss of “44%” for the official published consensus, and that since August 1998 stocks beating the whisper number “closed higher by an average of 1.8%” on the print — while stocks that beat the consensus but missed the whisper “closed lower 55% of the time” and fell 0.3% on average (Earnings Whispers, “About Whisper Numbers”). That is the technical answer to “how can a beat be bad news.”
A worked example
Imagine Acme Semis reports Q2 EPS of $1.20. Consensus was $1.15, so headline coverage screams “Acme beats by five cents.” But the whisper had been creeping up all week to $1.25 — options were priced for it, hedge funds were positioned for it. The print is technically a 4% beat versus consensus and a 4% miss versus whisper. The stock opens down. Nothing about the company changed; the bar that mattered to the marginal trader was higher than the bar in the headline.
What is post-earnings announcement drift (PEAD)?
PEAD is one of the most stubborn anomalies in equity markets. After a positive earnings surprise, a stock tends to keep drifting up for weeks. After a negative surprise, it drifts down. The phenomenon was first documented by Ball and Brown in 1968 and named “post-earnings-announcement drift” by Bernard and Thomas (overview; original paper: Bernard & Thomas, “Post-Earnings-Announcement Drift: Delayed Price Response or Risk Premium?”, Journal of Accounting Research, 1989).
The formula: standardised unexpected earnings (SUE)
Researchers measure the size of an earnings surprise with a metric called standardised unexpected earnings (SUE):
SUE = (actual EPS − expected EPS) ÷ standard deviation of past surprises
Bernard and Thomas defined “expected EPS” with a seasonal random walk — same quarter, one year earlier — and used the standard deviation of unexpected earnings over the previous eight quarters as the denominator. The point of dividing by that standard deviation is to make the surprise comparable across companies: a 1-cent miss for a stable utility is enormous, while the same miss for a volatile small-cap is noise.
How big is the drift?
Bernard and Thomas (1989) sorted stocks into ten buckets (deciles) by SUE every quarter and tracked their abnormal returns for 60 trading days after the announcement. Stocks in the top SUE decile drifted up about 2% on average; stocks in the bottom decile drifted down about 2%. A “zero-investment” portfolio that went long the top decile and short the bottom decile earned roughly a 4% spread per quarter, or about a 25% annualised abnormal return before transaction costs — and the spread was positive in 41 of the 48 quarters in their 1974–1985 sample (Tan & Tas, “A review of the Post-Earnings-Announcement Drift,” 2021).
the PEAD literature.
The chart is a teaching diagram, not a back-test — modern spreads are smaller (see “Has PEAD shrunk?”).
The same pattern in table form
| SUE decile (earnings surprise) | Direction of drift | Approx. 60-day abnormal return |
|---|---|---|
| Top decile (largest positive surprise) | Up | +~2% |
| Deciles 7–9 (moderate positive) | Up (smaller) | +0.5% to +1.5% |
| Deciles 4–6 (near zero) | Mixed / flat | ~0% |
| Deciles 2–3 (moderate negative) | Down (smaller) | −0.5% to −1.5% |
| Bottom decile (largest negative surprise) | Down | −~2% |
| Top minus bottom (zero-investment portfolio) | Long top / short bottom | ~4% per quarter (≈25% annualised, pre-cost) |
Journal of Accounting Research, summarized in
Wikipedia
and the
Tan & Tas (2021) PEAD review.
Numbers approximate magnitudes from the original Bernard & Thomas study (1974–1985 sample); see “Has PEAD shrunk?” below for recent estimates.
Why the drift happens (the leading explanations)
- Investor underreaction. The classic behavioural story: most investors fail to fully appreciate what a current surprise implies for the next quarter. Bernard and Thomas’s follow-up work showed that roughly a quarter of the total drift bunches into the three-day windows around the next earnings release, even though those days are only ~5% of trading days — consistent with the market repeatedly being surprised in the same direction (PEAD overview).
- Slow information processing and limited attention. Smaller stocks with less analyst coverage, lower institutional ownership, and thinner liquidity show bigger drifts — consistent with information taking longer to be absorbed.
- Limits to arbitrage. Even when traders see the drift, the cost and risk of putting on the trade (shorting illiquid losers, financing thousands of small positions) eats into the return.
- Risk premium, partly. A portion of the drift may compensate for genuine risks — expected earnings growth volatility, liquidity shocks — rather than being a pure mispricing.
Why “beats” sometimes drop — the whisper plus the guide
Three things move the tape on report day, in roughly this order:
- Result vs. whisper, not vs. consensus. If the buy-side’s informal bar was higher, a consensus beat is a whisper miss.
- Forward guidance. Companies often issue a forecast for the next quarter or full year. A beat on the just-reported quarter paired with a soft guide for next quarter routinely sells off — investors are paying for the future, not the past. FINRA explicitly notes that “simply beating or missing estimates isn’t always what moves a company’s stock price” and points to guidance and sentiment as drivers (FINRA).
- Quality of the beat. A beat driven by a lower tax rate, share buybacks, or a one-time gain is treated very differently from a beat driven by accelerating revenue and expanding margins. PEAD studies typically find the drift is concentrated in the high-quality, persistent earnings surprises, not the noisy ones.
Has PEAD shrunk? Yes — but it’s not zero
The PEAD spread has decayed since Bernard and Thomas first measured it. Reviews of the academic literature put the quarterly top-minus-bottom spread at roughly 5% in the 1974–1985 sample, around 3% or lower through the late 1990s and 2010s, and only about 1 percentage point in some recent samples — with several papers arguing the drift has effectively disappeared once realistic transaction costs are netted out (Martineau & Zoican, “Why Has PEAD Declined Over Time?”, Columbia Business School working paper).
overview;
Tan & Tas, 2021;
Martineau & Zoican, Columbia Business School). Recent estimates vary; some papers find the drift has effectively disappeared once transaction costs are netted out.
Likely culprits for the decline: more hedge-fund capital arbitraging the anomaly, deeper liquidity and tighter spreads after decimalisation, faster information processing (algorithmic trading, machine-readable filings), and disclosure reforms such as Sarbanes-Oxley that pulled more information into the announcement window itself.
Common mistakes when reading earnings
- Treating the headline beat as the whole story. The number that matters is the print versus the bar the marginal trader was carrying — which is often the whisper, not the consensus.
- Ignoring guidance. For most growth stocks, next-quarter and full-year guides move the stock more than the printed quarter does.
- Assuming the drift means easy money. The classic 25%-annualised number is gross of transaction costs, short-financing fees, and the very real risk that the next quarter’s print snaps the trade in the opposite direction. After-cost returns in modern samples are far smaller.
- Using EPS-only surprises. Revenue beats, margin beats, and segment-level beats all matter; a clean EPS beat driven by tax-rate quirks can mask weakness underneath.
- Forgetting day-zero noise. The first 30–60 minutes of post-print trading can be dominated by short-covering and dealer hedging in options — not by considered repricing. The “drift” is measured over weeks, not minutes.
What to learn next
If PEAD made sense, the natural next stops are: how analyst institutional positioning shifts around earnings; how option Greeks (especially vega and gamma) reprice around the announcement; and how zero-day-to-expiry options distort the very first minutes of price action.
Sources
- FINRA — What Is Earnings Season?
- Wikipedia — Post–earnings-announcement drift (overview & literature summary)
- Bernard & Thomas (1989), “Post-Earnings-Announcement Drift: Delayed Price Response or Risk Premium?”, Journal of Accounting Research
- Tan & Tas (2021), “A review of the Post-Earnings-Announcement Drift,” International Review of Financial Analysis
- Martineau & Zoican, “Why Has PEAD Declined Over Time? The Role of Earnings News Persistence” (Columbia Business School working paper)
- Earnings Whispers — About Whisper Numbers
- Earnings Whispers — The Earnings Whisper Grade
Disclosure: This article was produced with AI assistance and reviewed before publication. It is for informational purposes only and is not investment advice.