The Index That's Lying by Omission: What Market Breadth Is Actually Saying

The S&P 500 fell 0.1% on September 16th when the Federal Reserve raised rates. The equal-weight version fell 1.3%, only 32% of stocks finished positive, and new 52-week lows outnumbered new highs by more than three to one. When the headline number and the median stock beneath it stop agreeing, the index has stopped being a summary and started being a highlight reel of the names carrying five hundred on their backs.

On September 16th, the Federal Reserve raised rates a quarter point to a range of 3.75 to 4.00 percent, a unanimous decision, with guidance pointing to at least one more hike before year end. The S&P 500 that day fell 0.1%. If that single number were the whole story, the logical conclusion would be that markets barely noticed. Almost every other measure that day says otherwise.

One Index, Two Completely Different Days on September 16

The equal-weight S&P 500 fell 1.3% that session. The Dow dropped 1.7%. Small caps fell roughly 2%. Meanwhile the Nasdaq Composite rose 0.7%, and the headline S&P 500 slipped only a tenth of a point, held up almost entirely by Alphabet, Meta, and Nvidia. Underneath the index, breadth was ugly: only 32% of S&P 500 constituents finished the day positive, and new 52-week lows outnumbered new highs by more than three to one, over 350 stocks hitting fresh lows against roughly 100 hitting new highs. A widely tracked measure of how many S&P 500 members are trading above their own 50-day average fell to 39.7%, its weakest reading since April.

None of that shows up if you only check where the index closed.

When the Headline Number and the Actual Market Stop Agreeing

There is a simple way to see this divergence directly: compare the S&P 500's standard, capitalization-weighted version, where the largest companies dominate the return, against its equal-weight version, where all 500 members count the same regardless of size. When the two move together, gains are broad-based. When the cap-weighted version starts pulling ahead, it means a shrinking group of the largest names is doing an increasing share of the work.

Over the month leading into the Fed decision, the cap-weighted index outran the equal-weight version by roughly 1.8 percentage points. That is not an extreme reading on its own, but it has been widening, and September 16th showed exactly what that widening looks like in practice on a bad day: the mega-caps absorbed the shock, everything else did not.

Why a Rate Decision Does Not Hit All Five Hundred Stocks the Same Way

The unevenness is not random. Fed Chair Kevin Warsh described policy as "not yet restrictive," which shifted market attention from whether rates would rise again to how many more increases are still coming. Rising oil prices compounded the concern, adding a second inflation input on top of policy itself.

Small caps, regional and interest-rate-sensitive names, and retailers absorb that kind of repricing hardest. They tend to carry more floating-rate or frequently refinanced debt, rely more on domestic credit conditions, and have less pricing power to pass through higher input costs. A handful of AI-linked mega-caps, by contrast, are currently being valued substantially on a separate growth narrative, one that has so far proven fairly insulated from the immediate rate conversation. So a hawkish Fed surprise does not hit the market evenly. It hits the parts of the market most exposed to financing costs, while capital keeps concentrating in the handful of names investors believe do not need to care as much.

The result is a market that can post a flat headline number while the median stock in it is having a genuinely bad day.

What Market Breadth Divergence Has Historically Described

This matters for anyone using "the S&P is near a record" as shorthand for broad economic or market health, because that shorthand is becoming less reliable in direct proportion to how narrow leadership gets. A market where 100 stocks are making new highs while 350 are making new lows is not the same market as one where those numbers are reversed, even if the headline index closes in the same place in both scenarios.

This exact pattern, headline strength built on a narrowing base, has shown up before several of the market's worse stretches. Going into the 1973 to 74 bear market, roughly fifty large-cap growth names (the so-called Nifty Fifty) carried the index higher for months while the average stock had already begun rolling over; when the unwind came, those same names lost 45% or more. Breadth deteriorated through the second half of 1999 well before the Nasdaq's eventual March 2000 peak, with individual technology stocks topping months ahead of the index itself. Advance-decline measures were diverging lower into the 2007 market peak too, even as the S&P kept notching marginal new highs.

None of that makes narrow breadth a reliable timing tool on its own. In each case, the divergence ran for months, sometimes the better part of a year, before it resolved into anything, which is exactly why treating it as a short-term sell signal has burned people as often as it has helped them. What it has reliably done, across all of these episodes, is describe the market's actual structure more honestly than the index level did at the time. Right now it is describing a market where a shrinking handful of names is absorbing what the rest of the market cannot.

When an Index Becomes a Highlight Reel

An index level is a single number standing in for five hundred different stories, and lately most of those stories do not agree with each other. When the group carrying the average shrinks this much, the average stops being a summary and starts being a highlight reel: technically accurate, systematically incomplete, and increasingly a description of ten companies wearing a costume with five hundred names stitched onto it.