I wanted to see if the "84% of companies beat earnings" stat was actually rigged. So I scraped five years of every S&P 500 report.
The short answer is: kind of. The slightly longer answer is what the rest of this page is about.
Why almost everyone beats.
When I started pulling this data I expected the beat rate to wobble. Some quarters companies do well, some they don't. That's what "expectations" are for, right?
It doesn't wobble. Eighty-four percent of S&P 500 earnings reports beat consensus, and the figure has barely moved in twenty years. If you've worked in finance you know why; if you haven't, the answer is that the "consensus" isn't really set by the analysts. Six weeks before each report, the investor-relations team at the company quietly walks the analysts down. The conversation is unrecorded. The analysts revise their models. The number the press quotes on results day is, in part, the number the company helped author.
I can't prove that in a single chart, but the distribution itself is suspicious. If beats were genuinely hard, the surprise % should be a bell centered on zero. It isn't. It's a wall, with a spike between +1% and +5%. That's exactly where you'd aim if you wanted to clear the bar without overshooting and resetting expectations higher next quarter.
Where the 1,419 recent surprises actually landed.
The cluster on the right is the manageable beat zone. Companies could overshoot if they wanted to. They don't, because overshooting would reset the bar higher next quarter. So they aim for +1% to +5% and clear it.
Every company, every quarter.
Same data the histogram is summarizing, but laid out as one big grid so you can drill into any specific name. Hover for the headline numbers. Click for the full report card. Sort or filter to find the pattern you're looking for.
A beat doesn't mean the stock goes up.
This is the part that I kept finding surprising even though I should have known better.
If you plot every recent earnings report on two axes, surprise on the horizontal, next-day stock move on the vertical, you'd expect a diagonal: bigger beats, bigger jumps. That's not what's in the data. Beats are scattered above and below zero. So are misses. There's a loose tendency but no real line.
I think what's going on is that the market doesn't actually trade the beat. It trades the guidance for next quarter, the tone on the call, the three sentences from the CFO that I would never have read if I weren't writing this. By the time the headline number prints, most of it is already priced. The expensive lesson for anyone day-trading earnings is on this chart: a cloud computing company beats by twelve percent, guides down by half a percent, loses a fifth of its market cap before lunch.
Did beating predict the next-day move?
If the answer were "yes, mostly," dots would cluster along a rising diagonal. The cloud you're looking at instead is the story. I labelled the most useful outliers (beats that fell, misses that rose) so you can poke at the specific names.
Some sectors always beat. Some always struggle.
Once you group by GICS sector it's pretty obvious that the eleven groups aren't playing the same game.
Tech, healthcare, and financials beat consensus around nine times out of ten. Real estate beats six. Communication services and materials sit in between. I think the gap mostly comes down to how predictable each underlying business is. Utilities and staples are slow and regulated, so analysts learn to set tight ranges they'll always clear. Energy is the opposite: commodity-pegged, and managements get wrong-footed by oil more often than they like to admit.
Click any sector to filter the grid above and see the actual companies inside it.
The earnings prints I kept coming back to.
Most quarters vanish into the tape. These eight, for me, did something the others didn't. Each one reset a company, a sector, or the way an entire trade was valued.