2. Guessing next quarter's earnings gives no edge

Level 1: The price is a forecast

Imagine somebody whispers in your ear the exact profit a company will report three weeks from now. It looks like the ultimate edge. It is not, and understanding why is the step that separates whoever watches results from whoever watches expectations. The figure, on its own, is worth nothing. What is worth something is the difference between that figure and what the price already took for granted.

The consensus is not information: it is the starting line

When an S&P 500 company is about to report, there is a public number summarizing what the analysts covering it expect. That number is not hidden: it circulates on every terminal and in the financial press. And here is the point many people never quite accept: if it is public, it is already in the price.

A price forms when somebody buys and somebody sells knowing the same information. If the whole market knows that earnings of €2 per share are expected, nobody is going to pay today as though they expected 1.50. The consensus is not a prediction you can exploit; it is the floor from which everything else is measured.

Worse still: the published consensus and the expectation built into the price are not quite the same thing. The consensus is the average of a handful of formal estimates, often out of date and often talked down by the company itself in the preceding weeks. The expectation built into the price is what the people actually putting money down believe. When the two come apart, you get the reactions that look absurd from the outside.

The arithmetic of a quarter: it moves far less than it looks

This calculation ought to be compulsory before trading an earnings release. Take a company that earns 100 a year and that the market values at 2,000, that is, twenty times earnings. One quarter contributes about 25.

Suppose that quarter beats the consensus by 10%: that is 2.5 of extra profit, and extraordinary by definition, because nobody has said it will be repeated. That excess comes to 0.125% of the value of the company. And yet the share rises 15% that day, that is, 300.

Compare them: 2.5 of surprise against 300 of reaction, one hundred and twenty times as much. No sensible reading holds that the market is paying for that quarter. What it has done is revise the entire list of assumptions the previous module described, because something in the presentation has made it think that sales, margin or the duration of growth will be different from what it believed. The quarter is not the news; it is the clue.

Beating forecasts is the statistical norm, not a feat

One figure is worth keeping in mind: around three in four S&P 500 companies beat the earnings consensus every quarter, and that proportion holds quarter after quarter for decades. It is not that American companies are systematically better than forecast. It is that the bar is set where it can be cleared.

The mechanism is well understood and has several parts:

• The company guides the analysts in the preceding weeks, with caveats and remarks that nudge the estimates toward a comfortable level.
• Accounting has legitimate slack in revenue recognition, in provisions and in the timing of certain expenses.
• Analysts are paid to be right, and coming in below the final figure costs less professionally than coming in above it.

The practical consequence is devastating for anyone trading on headlines: if beating is what is expected, beating is not news. The information is in by how much the bar was cleared, in how it was cleared and, above all, in what is said about the quarters to come.

What really moves when a share falls 20%

When you see a 20% collapse after a good set of results, do not look for the mistake in the published figure. Look in the list of assumptions. Almost always, one of these has changed:

• The growth rate the company admits to for the coming quarters.
• The margin it says it can sustain, or the cost it concedes it will take to hold on to it.
• The capital it will have to invest, larger than was thought.
• The length of the period over which it can go on growing above its cost of capital.

Any of those four affects every future year at once, which is why each one moves a multiple of what a quarter moves.

Real case: Netflix, January 2022

On January 20, 2022 Netflix reported its results for the fourth quarter of 2021. It comfortably beat the earnings-per-share forecast, which stood at around $0.80, coming in at roughly 1.33. Revenue arrived in line with expectations and the subscribers added over the quarter landed very close to forecast, at around 8.3 million.

Any automated summary that night would have led with the company beating expectations. The next day the share fell by around 22%, its largest one-day drop in almost a decade.

The reason was not in the quarter reported, but in the guidance for the next one: the company projected some 2.5 million new subscribers for the first quarter of 2022, against the close to 7 million the market expected. That figure did not change one quarter; it changed the whole subscriber growth curve that was holding the price up, and with it the complete list of assumptions.

This is not an isolated case, nor one confined to companies in trouble. In August 2024, Nvidia reported sales and profit above consensus, with its business booming, and its share price fell in the following session. When the price discounts excellence, delivering excellence is not enough.

Take away the line that sums up this module: the money is not paid for the figure, it is paid for the revision.