Companies Beat Earnings 75.9% of the Time. Analysts Predicted 43.9%. Betting Against Them Would Have Won
Wall Street analysts set earnings bars they expect companies to clear, yet the numbers they publish tell a different story. Understanding the gap between their predictions and reality reveals a system built on incentives that have little to do with…
This post may contain links from our sponsors and affiliates, and Flywheel Publishing may receive compensation for actions taken through them.
In the sample studied by researchers Daniel Rabetti, Jiaqi Shao, and Che Zhang, companies cleared Wall Street’s earnings bar 76% of the time. The analyst consensus implied they would do so only 44% of the time. That gap described a forecast tilted against the outcome it was trying to predict, and a bettor who assumed companies would beat quarter after quarter would have looked prescient.
The finding is descriptive, backward-looking, and drawn from a narrow window. It is a measurement of how miscalibrated the consensus was in the sample, not a trading recipe, and the authors do not present it as one.
Where the Expectations Number Actually Comes From
When a headline says a company “beat by a penny,” the penny is measured against a consensus estimate. That number is an average of forecasts published by sell-side equity analysts, the researchers employed by brokerage firms and investment banks who cover a stock. Data vendors like Refinitiv, FactSet (NYSE:FDS | FDS Price Prediction), and Bloomberg collect those individual estimates and publish the mean or median as the consensus figure that shows up in earnings coverage.
Most readers treat the consensus as a neutral prior, the way a weather forecast is neutral about whether it rains. It emerges instead from an industry that sells research, maintains relationships with the companies it covers, and competes for underwriting business. Those incentives leave fingerprints.
How the Bar Gets Quietly Lowered
The central mechanism the study points to is the “walk-down.” A company issues guidance, holds calls with analysts, and steers estimates toward a number it is confident it can clear. By the time the quarter is reported, the bar has drifted low enough that beating it is the expected outcome. The beat is manufactured through expectations management as much as through operating performance.
Two other pressures compound the effect. Analysts who publish sharply bearish numbers risk losing access to management, which is the raw material of their job. They also face career risk for straying from the herd, because being wrong alone is punished more than being wrong with everyone else. Both forces compress estimates toward a beatable middle.
Why the Pattern Points to Incentives Over Forecasting Difficulty
The study offers three patterns that would be strange if analysts were simply doing their best against an unpredictable world.
- For companies riding long streaks of beating estimates, the prediction market’s accuracy improved while analyst accuracy got worse. If forecasting were the shared problem, both should struggle together. The divergence suggests analysts were pulled off the truth by something other than uncertainty.
- The bias was more pronounced in years when companies were issuing equity, meaning selling new shares to the public. Those are the years when brokerage firms compete for underwriting fees, and optimistic research is worth more to the corporate client than a candid estimate.
- On heavily shorted stocks, meaning stocks with a large share of the float borrowed and sold by investors betting the price will fall, the prediction market’s edge over analysts shrank. Short sellers publish research to defend their positions, and their pressure appears to drag analyst numbers closer to accurate.
What This Changes for a Reader of Earnings Coverage
A beat, by itself, carries almost no information. It is the base case. The pieces of an earnings report that actually moved stocks in the study’s window were the size of the surprise (how far the reported number sat above or below expectations), the forward guidance the company issued, and whether estimates had been quietly trimmed in the weeks before the release.
The concrete habit worth adopting is checking where the estimate sat sixty and thirty days before the report, not only the morning of. If the number drifted down and the company then cleared the lowered bar, the “beat” describes expectations management more than business momentum. Free tools on Yahoo Finance and most brokerage platforms show that estimate history.
Two caveats belong inside the argument rather than tucked into a footnote. As Larry Swedroe reported in Financial Advisor Magazine, the sample covers only six months and is limited to firms with active prediction markets, which skews toward larger and more liquid names. The mechanisms are inferred from observed patterns rather than directly measured, and the authors call the evidence preliminary and in need of replication over longer horizons.
The practical lesson survives the caveats. Treat a beat as the default and read the guidance and the revisions instead. When coverage tells you a company topped expectations, the useful question is whose expectations, and when they were set.
Contact [email protected] for any questions or corrections.
