A market-breadth rabbit hole · Sep 2026
A chart I came across makes a sharp point: at every major S&P 500 peak since 1973, new highs were at least 6% of index market cap — until August 13, when the all-time high came with just 3.73%. The thinnest ever, the author says, and an ominous sign. It’s a genuinely good thing to watch, so I rebuilt the whole series from point-in-time data and pushed on it three ways.
First, credit where it’s due — this is a real observation about a real thing. Bull markets do tend to narrow before they end: fewer and fewer stocks carry the index to its final high while the average name has already rolled over. Watching participation instead of just the index level is exactly the right instinct, and compiling it back to 1973 is genuine work. The raw new-high counts in the table line up almost exactly with what I pull from Norgate at each peak, so we’re looking at the same market. I just wanted to poke at the conclusion.
Three questions fell out of it. Does “thinnest ever” hold up under a different, equally fair way of counting? Has a thin-breadth high actually been a warning in the past? And if you traded on it, would you have come out ahead? The answers, in order: it depends how you weight it, not really, and no — a plain 200-day moving average does the job better.
Aug 13 · new highs by market cap
The author’s figure — and yes, the lowest in their table of major peaks going back to 1973.
Aug 13 · new highs by share of stocks
Weight every stock equally and 2000 (3.4%) and 2022 (3.6%) were thinner than today. The ranking flips with the weighting.
One year after a thin-breadth high
Average SPY total return after a new high with breadth under 6% — a touch above the +12.3% all-days baseline.
The table measures new highs as a share of market cap. That’s a reasonable choice, but it isn’t the only one — you can also just count what fraction of the 500 stocks are making new highs. At most peaks the two agree. At a few they part ways sharply, and that’s where the “thinnest ever” claim lives.
March 2000 is the tell. Only sixteen stocks were at new highs — a paltry 3.4% by count — but those sixteen were the giant tech names, so weighting by size lifted the reading to a comfortable-looking 12.45%. Today’s few new highs are not the mega-caps, so the same math runs the other way and pulls the cap-weighted number down to 3.73%. By simple stock count, 2000 and 2022 were both thinner than now. “Thinnest ever” is true only in the one construction that, in a top-heavy index, mostly tracks whether a handful of giants happened to be at a high that day.
share of market cap vs. share of stocks · 1973–2026
Here’s the part that surprised me. If narrow breadth at a new high were ominous, you’d expect weak returns to follow. They don’t. Take every day the S&P made a fresh three-month high since 1993 and split them by how broad the move was. New highs on thin breadth (under 6% of stocks) were followed by returns at or slightly above the everyday baseline — and a hair better than broad highs — at every horizon out to a year.
It’s not luck; it’s structural. Chronically narrow breadth is the signature of a mega-cap-led bull — the late 1990s, 2020–21, 2023–25 — where the index keeps grinding higher for years while the average stock lags. A signal that’s been lit through most of the best runs in the sample isn’t much of a top-timer.
1993–2026 · average return, dividends included
The real test of an “ominous” signal is whether acting on it helps. So I gated SPY on it: hold when breadth is broad, step aside when it isn’t, and compare against buy-and-hold and against a plain 200-day moving-average trend filter. The breadth gate is the worst of the three by a wide margin. It parks you in cash during exactly the narrow-breadth mega-cap advances that produce most of the return, so it compounds at essentially zero — and its drawdown protection is worse than the moving average anyway. Layer it on top of the 200-day line and it only subtracts.
| Gate | CAGR | Max DD | MAR | Sharpe | In market |
|---|---|---|---|---|---|
| Buy & hold | +10.9% | −55% | 0.20 | 0.65 | 100% |
| 200-day moving average | +8.2% | −26% | 0.32 | 0.72 | 75% |
| Breadth > 6% | −0.3% | −35% | −0.01 | −0.01 | 32% |
| 200-day and breadth>6% | −0.2% | −34% | −0.01 | 0.00 | 31% |
| 200-day or breadth>6% | +8.1% | −26% | 0.31 | 0.71 | 76% |
Narrow participation is worth watching — the table is right that it thins out near tops. But by the tradeable version of the question, the tell was never the breadth reading. It was the trend line the breadth reading is standing in front of.
Same idea, one rabbit hole deeper.Respect the observation, skip the alarm. Breadth this narrow is real, and it does say the current advance is fragile — fewer names are holding it up, and when leadership like that cracks it can crack fast. That’s worth knowing. What the history won’t support is using the reading as a timer: thin-breadth highs have paid off normally, and a strategy built on stepping aside for them would have missed years of gains. If you want a rule that has actually protected capital, it’s the boring one — watch whether the index is above or below its 200-day line. Treat narrow breadth as context that raises your attention, not as a sell button. Today it’s a warning light; it has been glowing through most of the best bull markets on record.
#SPX52WHI and siblings), point-in-time, back to 1970. Equal-weight breadth = new highs ÷ (advances + declines + unchanged).None of this makes the original point wrong, and that’s the good part. It grabbed the right variable — participation, not just price — and did the work to line it up across history. Chasing that same variable a step further just adds a coda: how thin “thinnest” is depends on your ruler, thin highs have paid off fine, and the tradeable edge sits in the trend line, not the breadth number. The market really is narrow right now. It’s just not a countdown.