The Beat Rate Is Lying to You, Just Not the Way You Think

Eighty-five percent. That's the share of S&P 500 companies that have topped estimates this earnings season, out of roughly 300 that have reported so far, and aggregate profit growth is tracking above 47%. Bulls have been quoting that pair of numbers all week like a incantation. RBC's got an 8,150 price target on the back of it. The framing writes itself: broad-based earnings power, soft landing intact, buy the dip.

Now go look at Friday's tape underneath the headline index move. The Nasdaq Composite fell 0.1% — fine, unremarkable — except it registered 52 new 52-week highs against 131 new lows. That is not a market grinding higher on broad participation. That is a market where a shrinking number of names are dragging the index while the median constituent quietly bleeds out. You can have an 85% beat rate and a deteriorating breadth chart in the same week, and the reason is uncomfortable: beating a Wall Street estimate has almost nothing to do with whether a business is actually gaining ground, and everything to do with whether a company's guidance was low enough in June for the number to be clearable in August.

The engineering underneath "broadening"

There's a competing narrative floating around trading desks right now — that leadership is broadening beyond the Magnificent Seven, that the equal-weight S&P is finally catching up to the cap-weighted index, that this cycle looks healthier than 2023 or 2024 because more companies are participating. Some of that is real. Some of it is Financials and Industrials riding a rate environment that hasn't actually eased, using EPS growth that's flattered by buybacks rather than volume or pricing power. The top three names in the index still account for roughly a fifth of its market cap. The top ten are closer to two-fifths. When XLK gains 5.5% in a session and XLI adds another point while XLC drops 2.7% and XLP falls 2.2%, that isn't breadth. That's rotation within a narrow band of acceptable trades, dressed up as breadth because six of eleven sectors happened to close green.

What margin debt does to a narrowing market

Layer onto this the leverage picture. FINRA margin debt sits north of $1.5 trillion, up roughly 49% year-over-year, one of the fastest leverage build-ups relative to index gains in the post-2010 data. Debt growing meaningfully faster than the market it's financing has preceded exactly three prior peaks worth naming — 2000, 2007, 2021 — and the point of citing that history isn't superstition, it's mechanical. Margin debt doesn't care about your stock-picking thesis. It cares about maintenance requirements. When breadth is already this thin and a decent chunk of the buying power behind the winners is borrowed, the market's ability to absorb a shock in any single mega-cap name gets worse, not better, the longer the concentration holds.

The Fed isn't cooperating with the story either

None of this is happening in a vacuum where the Fed is quietly easing the runway. Chair Warsh's committee just delivered three hawkish dissents, year-end rate projections sit at 3.6% to 4.1%, and the market is still pricing two more hikes before December on the argument that five straight years above target inflation and a June CPI improvement driven by gasoline rather than core services doesn't constitute the kind of disinflation that earns patience. That's the backdrop against which utilities fell 4.21% last week and REITs dropped 2.27% — the rate-sensitive complex getting repriced for a Fed that isn't pivoting, at the same moment the concentrated AI trade is getting bid on a ceasefire narrative Iran itself denies exists.

Put the two facts next to each other

An 85% beat rate against a hawkish Fed and cratering Nasdaq breadth is not a contradiction the market needs to resolve eventually. It's a description of what a late-stage, leverage-financed, concentration-dependent rally looks like while it's still working. The beat rate tells you management teams are good at setting expectations. The breadth chart tells you what's actually happening to the other 400-plus names in the index once you stop looking at the six or seven that are carrying it. Both are true simultaneously, and only one of them is a reliable guide to what happens when CoreWeave's Tuesday print, or the next payrolls read, or the next FOMC dot plot, gives the leveraged long a reason to sell first and ask questions later.

The bull case isn't wrong that earnings are good. It's wrong that good earnings, this narrowly distributed and this thinly test against 131 new lows, are the same thing as a healthy market.