The earnings beat is mostly noise

October 14, 2026·8 min read

Every quarter, the financial press treats earnings "beats" and "misses" as verdicts on a company. They are closer to weather reports. A beat measures one noisy quarter against a forecast that was probably wrong in the first place. Quarterly surprises barely move long-run returns. This guide explains why, and it gives you two questions that separate a real signal from noise. Read it once, and the next earnings season will look different. The headlines will still shout. You will know when to listen.

The Beat/Miss Sideshow

A beat usually tells you more about the forecast than about the company. Earnings are a company's profit for a period, and analysts are people who study companies and forecast that profit. The published estimate is usually an average of many analysts' forecasts, called the consensus estimate. A beat means profit came in above that average, and a miss means it came in below.

Those forecasts are not neutral. Analysts move their forecasts slowly, a habit called sticky estimates, and they cluster near each other, a habit called herding. They also start each year too high.

Think of ten people guessing the weight of the same suitcase. Most will land near each other. Most will adjust slowly when new facts arrive. Analysts work the same way.

These habits make the forecast a moving target. The number you see in a headline is a snapshot of an average that shifted all quarter. A beat against that moving target says less than a beat against a fixed bar.

You open your phone and see the green headline. BEAT. Your first thought is that the company is on fire. Pause there. Ask what the headline actually measured.

Companies then lower their own forecasts, which is called guidance, so the bar is easy to clear. The result is choreography: more than 70% of S&P 500 companies beat estimates in a typical quarter. The S&P 500 is an index of about 500 large American companies.

Seven out of ten clear the bar. If a test is built so nearly everyone passes, the pass tells you nothing. That is the whole trick, and it repeats every three months.

Suppose a company first forecasts $1.00 of profit for each share, then lowers that forecast to $0.90 a few months later. It then earns $0.95. Headlines call that a beat, but the business did not improve. The bar moved, not the company.

The market knows this game. Stocks often fall after a beat and rise after a miss. Prices react to the surprise measured against the unofficial expectation traders actually use, called the whisper number, not the headline forecast.

Suppose the published forecast is $1.00, but traders quietly expect $1.10 after a strong month. The company earns $1.05. Headlines call it a beat. Traders call it a miss, and the stock falls. The headline and the market measured against different bars.

Now flip it. Suppose the published forecast is $1.00, but traders quietly expect $0.90 after a weak month. The company earns $0.95. Headlines call it a miss. Traders call it a beat, and the stock rises. Same company. Same quarter. Different bar.

Why One Quarter Says So Little

One quarter is one draw from a wide range of outcomes. Quarterly profit is noisy, and three forces add much of the noise. The first is unusual gains or costs that do not repeat, called one-time items. The second is sales or costs landing in a different quarter than expected, called timing shifts.

Three forces. Three ways for a quarter to mislead you. None of them show up in the headline.

The third force is error in the estimate itself, called estimation error. Profit itself is estimated, not counted like cash in a drawer. Companies must judge the value of inventory, unpaid bills, and future costs. Those judgments move the reported number even when the business did not change.

A company that sells on credit must guess how many bills will go unpaid. A company that holds goods must guess what those goods are worth. Both guesses land in the profit number, and both can move it without any change in the business.

Take a builder halfway through a housing project. How much of the profit counts this quarter? The answer depends on a judgment about how far along the work is. Two honest accountants could give you two different numbers. The houses did not change. The judgment did.

Suppose a company earns $2.00 per share in a quarter, but $0.50 came from selling land it no longer needed. The core business earned $1.50. The headline says $2.00, but the trend did not change. A reader who stops at the headline overstates the quarter by a third.

Timing shifts work the same way. Suppose a large order slips from March into April, so the first quarter looks weak and the second quarter looks strong. The business did not change. The calendar did.

Or picture a store whose biggest shipment lands one day late, on April 1 instead of March 31. The first quarter looks soft. The second quarter looks strong. The store sold the same goods. The truck was late.

That is why reacting to one quarter is like judging a hitter by one at-bat. One swing tells you almost nothing about the hitter. One quarter tells you almost nothing about the company. This wobble has a name: sampling variation, the spread you get from looking at one small slice of data.

One at-bat. One quarter. Same problem.

A Significance Mindset

Treat each earnings report as one small clue, not a verdict. A single data point barely moves your beliefs. That is the habit of updating your beliefs a little with each new fact, called Bayesian thinking. One quarter earns one small update, not a new conclusion.

One spoonful of soup tells you the pot is salty. It does not tell you the whole recipe. Earnings work the same way. One report seasons your view. It does not rewrite it.

What should move your beliefs is a pattern. Look for sales, also called revenue, rising across four quarters. Look for profit as a share of sales, called the profit margin, widening while the company raises its forecast. Or look for the opposite. A pattern is a story, and one quarter is a headline.

The opposite pattern is just as clear. Sales falling across four quarters, profit as a share of sales narrowing, and a forecast cut twice tell you the business is shrinking. One weak quarter says little. Four weak quarters say much more.

Suppose a company grows sales for four straight quarters and raises its forecast twice. That pattern says more than any single beat. Now suppose a second company posts one strong quarter after three flat ones. The pattern says little, even if the headline shouts. Patterns carry the weight, and single quarters do not.

Here is how that looks in practice. Quarter one: sales up a little. Quarter two: sales up again, and the margin widens. Quarter three: same again, and the company raises its forecast. Quarter four: same again. No single quarter made the case. Four of them did.

Four quarters give you four clues instead of one, and each clue is still noisy. A trend that survives four quarters is harder to fake than a trend that lasts one.

Two Questions to Ask

When you see BEAT in all caps, ask two questions before you react. First: is this surprise big compared with how far forecasts usually miss? Second: does it change the multi-quarter story, or only this quarter's headline? If both answers are no, you watched noise get a parade.

The two questions work as a filter. The first question asks whether the surprise is big enough to notice. The second asks whether it matters. A report must pass both filters to earn more of your time.

The first question sets a bar. Every forecast carries some error, so a small beat is routine. A surprise counts only when it clears the usual size of that error. Suppose forecasts for a company usually miss by about $0.10. A beat of $0.05 is routine. A beat of $0.40 is a different matter.

The second question checks the story. A beat that lifts the multi-quarter trend matters. A beat that leaves the trend unchanged does not. Suppose a company beats by $0.40 and raises its forecast for the year. That changes the story. Suppose a second company beats by $0.40 on a one-time gain and holds its forecast. That changes nothing but the headline.

Try the pair on a headline tonight. You see a company beat by a wide margin. First question: is the margin wide compared with this company's usual error? If yes, keep going. Second question: did the company raise its forecast for the year, or did the beat come from a one-time gain? A raised forecast changes the story. A one-time gain does not. Two questions. Ten seconds. Most headlines fail the first one.

Use the two questions in order. If the surprise is small, stop there; if the surprise is large, check the story. A beat that passes both tests is worth your attention. The rest is noise.

That is the whole method. Not a verdict. A filter. Noise gets a parade every quarter. You do not have to attend.