A Record Quarter and a Falling Stock — What "Beating Estimates" Actually Measures

On October 8, Samsung Electronics reported the best quarter any Korean company has ever reported. Preliminary operating profit for July through September came to 107.4 trillion won, up 782.5% from a year earlier — the first time a company in Korea has cleared 100 trillion won of operating profit in a single quarter. Revenue was 195 trillion won. Both numbers topped the record the company had set only three months before.

The profit figure also came in slightly above the consensus estimate — the average of analysts' forecasts, which FnGuide put at about 106 trillion won.

That afternoon, Samsung's shares closed down 2.42%, at 268,500 won.

A record, a beat, and a falling price, all on the same day. It looks like a contradiction. It is not, and the reason is worth understanding this week in particular, because the American earnings season is starting, and the same arithmetic is about to be applied to hundreds of companies at once.

Here is the judgment I keep coming back to: an earnings "beat" is not a grade on the company. It is a grade on the forecast. And the forecast is a number someone else set — usually in a place where it can be beaten.

What Happened on October 8

Korean coverage that evening pointed to the same explanation. Expectations for memory chips had already climbed so high that a result slightly above the published estimate was not enough to push the price further. Lee Kyung-min, an analyst at Daishin Securities, described it as a "sell-on" — selling that arrives because good news has already been partly reflected in the price. Aju Business Daily noted that Micron had also reported results above consensus at the end of September, with little reaction in its stock.

In other words, the market did not grade Samsung against last year, or against its own record, or against the published estimate. It graded the result against a number that existed only in the price — what buyers had already paid for in advance.

I wrote here in August that a purchase price is a summary of expectations — that a thing can become more expensive without becoming more useful, because beliefs about the future can change while the present stays put. October 8 was the same sentence running in reverse. The present got better than it had ever been, and the price fell, because the belief had already been paid for.

The Anchor Built to Be Beaten

To see why a beat means less than it sounds like, look at how often companies beat.

FactSet, which tracks analysts' estimates for every company in the S&P 500, published the figures on October 9. Over the past ten years, 77% of S&P 500 companies reported earnings above the average estimate in a typical quarter, and actual earnings came in 8.0% above estimates on average. Over the past four quarters, the beat rate was 82%.

The most striking line is this one: the actual earnings growth rate for the index has exceeded the estimated growth rate at the end of the quarter in 37 of the past 40 quarters. The three exceptions were the first quarter of 2020 and the third and fourth quarters of 2022.

Thirty-seven out of forty is not a scoreboard. It is a pattern in how the starting line gets drawn. Whatever the reasons the estimates end up where they do, they end up somewhere the companies, as a group, clear almost every time. So when a headline says a company "beat," it is telling you the most common outcome there is. The surprise would be a miss.

That is what makes the published estimate a strange thing to anchor on. It is a reference point that most investors treat as neutral, when the record says it leans low. An earnings beat, by that measure, is roughly the same as showing up.

This season, the anchor has an unusual height. FactSet's current estimate for third-quarter earnings growth in the S&P 500 is 29.6%, which would be the third consecutive quarter above 25%. Using its own historical averages, FactSet calculated that the final figure would likely land above 35%.

The arithmetic behind that forecast is worth seeing once, because it shows how a beat changes the headline number without anything changing in the economy. FactSet's own example: a company expected to earn $1.05 a share, against $1.00 a year earlier, is projected to grow 5%. If it reports $1.10 instead, its growth rate for the quarter becomes 10%. Multiply that across hundreds of companies that beat in a typical season, and the index's growth rate climbs as the reports come in.

How far it climbs depends on which history you use. Over the past ten years, the growth rate has risen by 6.9 points on average between the end of a quarter and the end of its earnings season. Over the past five years, by 6.4 points. Over the past four quarters, by 14.1 points. Apply the most conservative of those to the estimate on September 30 — 29.2% — and you get roughly 35.6%. Apply the most recent, and you get 43.3%.

The early reports fit the pattern. Of the 19 S&P 500 companies that had reported third-quarter results by October 9, 84% beat the average estimate, by 4.9% in aggregate.

Notice what this does to the word "surprise." By the time the season ends, the market will likely have been "surprised" by a growth rate several points above the one it was told to expect — as it has been in 37 of the last 40 quarters. A surprise that arrives on schedule is not much of a surprise. It is a second anchor, sitting above the published one, and prices may already be assuming it.

So the number the market will actually react to is not 29.6%. It is whatever buyers have already built into prices on top of 29.6% — and, just as with Samsung, that number is never published.

Why the Last Few Quarters Set the Bar

There is a second force underneath this, and it is the one investors share with analysts.

When growth has been extraordinary for three quarters in a row, extraordinary becomes the baseline. The most recent results are the easiest to remember and the most vivid to extrapolate, so they quietly become the definition of normal. That is recency bias, and it works on everyone who looks at the same charts.

The consequence is a moving bar. A year earlier, the same quarter had earned Samsung about 12 trillion won. By October, 107.4 trillion was being measured not against that, but against the quarter just before it and the price just before it — and against those, a 20% jump in profit from the previous quarter was simply the next step.

Nothing about the business got worse on October 8. The yardstick got longer.

The Anchors We Bring Ourselves

Analysts set one anchor. We set the others ourselves, usually without noticing.

Psychologists call the habit anchoring bias: a number we happen to have in mind pulls every later judgment toward it, whether or not it has anything to do with the question. The most common anchor is the price we paid. If you have ever held a stock and caught yourself waiting for it to "get back to what I paid," you have used the oldest anchor in investing. The market does not know your entry price and has no obligation to return to it. But the number shapes the decision anyway — when to sell, when to add, what counts as a loss.

The second is the peak. Last week I wrote about Korean investors who borrowed a record amount near the top. One analyst noted in August that individual buying had been concentrated between roughly 7,000 and 8,500 on the KOSPI, and expected steady selling from people looking to shrink their losses and leave once the index climbed back into the 7,000 to 8,000 range. That is the anchor at work across a whole market: a price level matters because many people remember buying there, not because of anything about the companies.

The third is the recent past. A long run of good results starts to feel like the definition of normal — the same trap as reading an all-time high as a verdict on a strategy. A year of strong returns makes the next year's ordinary return feel like a disappointment. A year of losses makes an ordinary rebound feel like a trap — the mood I tried to describe in a guide to the year your portfolio is down 30%. Neither feeling has any information about what comes next. Both come from the last number we saw.

What these anchors have in common is that none of them describes the asset. They describe our own history with it.

Where This Stops

None of this is a view on Samsung's stock, on memory chips, or on where the KOSPI goes next — an index in which Samsung is one of two companies carrying close to half the weight. Samsung's conference call on October 29 will add segment detail and next year's outlook, and the stock will react to that too. I have no prediction worth publishing about which way.

And beats are not meaningless. A large miss can be real information, and so can the quality of a beat — whether it came from sales, from cost cuts, or from a one-time item. The narrower point is that the headline word "beat" carries far less information than its tone suggests, because the bar it refers to is cleared 37 times out of 40.

Reading This Earnings Season Without the Borrowed Anchor

A few questions that help, none of which require predicting anything.

Compared with what? Before reacting to "beat" or "miss," look at the result against the same quarter last year. That comparison uses no one's forecast. It is the business against itself.

Is the market grading the result or the guidance? For many companies, the stock moves on what management says about the next quarter, not on the quarter just reported. Samsung's analysts were already pointing to next year's estimates and the October 29 call.

Which anchor are you using? If the number in your head is the price you paid, the price at the peak, or last year's return, it is worth noticing that the market is not using any of them.

Did anything about the business change today? On many earnings days, the honest answer is no — the expectations changed. Those are different events, and only one of them is about what you own.

Lines to Watch

Not predictions — the figures that would tell you how this season is being graded.

FactSet's blended earnings growth rate for the S&P 500, updated weekly through the season, against the 29.6% starting estimate.
The share of companies beating estimates, against the average of 77% over the past ten years and 78% over the past five.
How stocks react to beats. When a high beat rate meets falling prices, the anchor that matters is the one in the price, not the one in the estimate.
Samsung Electronics' conference call on October 29, where full results and next year's outlook replace the preliminary figures.

The Yardstick Moves

There is a small, uncomfortable freedom in all of this.

Most of the numbers that make an earnings day feel like a verdict are set by someone else — an analyst's average, a crowd's expectations, a price paid in advance. One more is set by us: the price we paid and the peak we remember.

Samsung earned more in three months than any company in its country ever had, and the stock fell. Not because the result was bad, but because the yardstick had already been moved past it.

The business is the thing you own. The yardstick is only what everyone agreed to measure it with this week.

Data as of October 13, 2026. Sources: Samsung Electronics, preliminary third-quarter 2026 results (Samsung Newsroom Korea, October 8, 2026); Aju Business Daily (October 8, 2026), on Samsung's closing price, the FnGuide consensus, and analyst comments; FactSet, John Butters, "S&P 500 Likely to Report Earnings Growth Above 35% for Q3" (October 9, 2026); Maeil Business Newspaper (August 4, 2026), on the concentration of individual buying between 7,000 and 8,500.

Related Posts:

Visuals on this post are AI-generated. The author works with AI as a research and drafting assistant; topics, judgments, and final edits are the author's own. This post is observation, not investment advice. See full Disclaimer for details.

Search This Blog

IJin Insight