Standardized Unexpected Earnings: Formula, Example, and Limits

The standardized unexpected earnings formula is the actual reported earnings per share minus the consensus analyst EPS forecast, divided by the standard deviation of those analyst forecasts. The result is a Z-score: a SUE of +2.0 means the company beat the consensus by two standard deviations of analyst disagreement, which is a far more useful signal than the raw dollar surprise on its own.

Written out:

SUE = (Actual EPS โˆ’ Consensus Forecast EPS) รท Standard Deviation of Analyst Forecasts

The Three Numbers You Need

Every SUE calculation runs on three inputs, all available at the moment of the earnings release.

The first is actual EPS, the number the company officially reports for the quarter. Some analysts track GAAP earnings, others track an adjusted non-GAAP figure. The choice matters less than consistency: use whichever definition the analyst forecasts were targeting.

The second is the consensus forecast EPS, typically the mean (sometimes the median) of individual analyst estimates compiled before the announcement date. This is the market’s collective expectation going in.

The third is the standard deviation of those individual analyst forecasts. This captures how much the analysts disagreed. A tight cluster means the market was confident; a wide spread means there was real uncertainty about the outcome.

The first two give you the raw earnings surprise. The third is what makes SUE worth calculating, because it scales the surprise against how predictable the company’s earnings were in the first place. A five-cent beat when every analyst was within a penny of each other is a genuinely shocking result. The same five-cent beat when estimates ranged over fifty cents barely registers.

A Worked Example

Suppose a company reports EPS of $1.55 for the quarter. The mean analyst forecast was $1.50, and the standard deviation of those individual estimates was $0.025.

The numerator is $1.55 โˆ’ $1.50, a raw surprise of $0.05. Divide $0.05 by $0.025 and you get a SUE of +2.0.

That +2.0 is a Z-score, the same standardized measure used throughout statistics. The reported earnings landed two standard deviations above what the analyst community expected. Because the output is standardized, you can compare it directly against the SUE score of any other company, regardless of share price, market capitalization, or industry. A SUE of +2.0 for a regional bank carries the same statistical meaning as a SUE of +2.0 for a large-cap semiconductor firm.

Reading the Score

The sign tells you direction. A positive SUE means the company beat expectations; a negative SUE means it fell short. A score roughly between โˆ’0.5 and +0.5 suggests the report landed close enough to the consensus that the market received little new information.

The magnitude tells you how surprising the result was, and the interpretation follows the same logic as standard deviations in any normal distribution. A SUE of +1.0 is a mild beat. A SUE of +2.5 means the actual result was so far from the consensus that it would have been considered highly unlikely given the spread of analyst estimates going in. The farther the score sits from zero in either direction, the more new information the market has to absorb.

This is why SUE replaces raw dollar surprises for cross-company analysis. A one-cent surprise at a high-confidence stock is far more meaningful than a ten-cent surprise at a company nobody could predict. SUE strips away that noise and lets you compare surprises on equal footing.

The Time-Series Alternative

The version described above is the analyst-forecast model, and it is the most common approach in practice. But academic research has long used a second method that bypasses analyst estimates entirely, which matters if you are looking at a company without broad coverage.

The time-series model builds the expected value from the company’s own earnings history. The most common version is a seasonal random walk: the best forecast for a given quarter is what the company earned in the same quarter one year earlier, plus a small trend adjustment. The scaling factor becomes the standard deviation of the company’s historical forecast errors rather than analyst disagreement.1Brandeis University. A New Measure of Earnings Surprises and Post-Earnings-Announcement Drift

This version is especially useful for small or thinly covered companies where only one or two analysts publish estimates. With so few forecasts, the standard deviation of analyst estimates becomes unreliable, and the company’s own earnings history is the better anchor. All you need is four or more years of quarterly earnings, which is freely available through SEC filings.

Where the Formula Breaks

SUE is a reliable metric in most cases, but a few situations will produce misleading scores.

Thin analyst coverage is the most common problem. When only two or three analysts cover a company, the standard deviation of their estimates is statistically fragile. A single outlier can inflate or collapse the denominator, producing SUE values that look dramatic but reflect nothing about the actual earnings surprise. Most researchers require a minimum of three to five analyst estimates before treating the analyst-forecast SUE as meaningful.

A related issue is the near-zero denominator. When analysts are in almost perfect agreement, the standard deviation approaches zero, and dividing by a number close to zero produces an enormous SUE score. A company that beats by a single penny against a standard deviation of $0.001 would show a SUE of +10, which sounds extraordinary but really just means the analysts all said the same thing and were slightly wrong. Screening for a minimum standard deviation threshold avoids this distortion.

The choice between GAAP and non-GAAP earnings adds another layer of risk. If the company reports on a non-GAAP basis but the analyst estimates were built on GAAP, the mismatch contaminates the calculation. Consistency between actual and expected figures is essential, and not every data source makes it easy to verify which basis a given estimate used.

Where the Data Comes From

The analyst-forecast version requires individual analyst estimates, not just the headline consensus number, because you need the full distribution to compute the standard deviation. The standard professional source is I/B/E/S Estimates, now owned by the London Stock Exchange Group, which compiles individual analyst detail, consensus figures, and actual reported earnings across thousands of public companies.2LSEG. I/B/E/S Estimates – Company Data Bloomberg Terminal and FactSet offer similar databases with consensus breakdowns and standard deviations.

Several free financial data sites publish the number of analyst estimates, the consensus mean, and the high-low range. The standard deviation is not always published directly, but you can approximate it from the range and count if you need to run the calculation without institutional data.

Why the Calculation Matters

SUE is not just an academic exercise. The reason it has stayed in use for decades is a documented market pattern called post-earnings announcement drift, or PEAD. First documented by Ball and Brown in 1968, PEAD is the observation that stock prices keep moving in the direction of the earnings surprise for weeks or months after the announcement itself.3Ivey Business School. When Two Anomalies Meet: Post-Earnings-Announcement Drift and Value-Glamour Anomaly A company that beats expectations tends to continue outperforming for at least 60 days. One that misses tends to keep sliding for just as long.

SUE is the preferred metric for studying this effect because raw dollar surprises confuse firm size with information content. By standardizing the surprise against forecast uncertainty, SUE isolates the actual new information the market received. That is the practical payoff of the formula: a score that lets you rank surprises across companies, quarters, and industries on a single scale.