The Stock Market Has Looked Like This Three Times Since 1927

Research note: S&P 500 member breadth through September 25, 2026; CRSP industry data through August 31, 2026. Frozen September 26, 2026. How we checked it · How we use AI.

• It is new on the modern record. Since 2012 the same breadth readings appeared in 14 earlier episodes, and in every one the index was already at least 5.6% below its high.
• It is rare on the long record. Three episodes in 99 years, on a survivorship-free industry version of the same measure.
• Most tops look nothing like it. 17 of 19 bear-market peaks since 1929 had broad participation in their final three months, including 1929 and 2007.
• The two narrow tops were the famous ones. January 1973 (−48.2%) and March 2000 (−36.8%).

A stock index can rise while most of its members fall, as long as the largest companies carry it. That is what the S&P 500 has been doing. At the September 25, 2026 close it was within a percent of its record, and 74% of its members were below their 50-day average, with 54% below their 200-day.

My read is that this is a genuine warning about the structure of the market and a weak one about its timing. The setup is rare, and the two times it appeared before 2023 were the Nifty Fifty and dot-com tops. But most bear markets began from broad markets, and the most recent narrow one, in 2023, resolved with a strong year. It moves me from comfortable to watchful, without turning into a reason to sell.

Here is what that looks like. When SPY set its record on August 13, 2026, 69% of S&P 500 members were above their 50-day average. By September 25, 2026, SPY was 0.8% below that record and the share had fallen to 26%. The index barely moved while most of its members slipped below their trend.

That is the setup. The chart below marks every time the same combination appeared on the long industry record.

What the S&P 500 looks like now

Our S&P 500 breadth series starts in 2010 and uses today’s members. Since 2012, fewer than 27% of members above their 50-day and fewer than half above the 100- and 200-day happened in 14 earlier episodes. Every one came during a correction, with SPY at least 5.6% below its high, and the median year afterward was +12.3%. Those were washouts. This one arrived at a high. (Every episode is in the study download.)

Three times since 1927

To see further back, I used Ken French’s daily CRSP industry portfolios, which include every listed company of their day. The share of the 49 industries above their averages has moved with our S&P 500 member readings almost exactly since 2012, with a correlation of 0.96. On that record, the setup appeared three times.

Each of the three looked the same on the day. Two years later the market was 42% lower after September 1972 and 21% lower after March 2000, and 47% higher after May 2023, against a typical two-year gain of 25%. It is worth walking through them one at a time.

Most tops were broad

The more useful question runs the other way: what did breadth look like before each bear market began? Of the 19 S&P 500 bear-market peaks since 1929 that came near a 52-week high, 17 had broad participation in their last three months. The broad ones were the peaks of 1929, 1933, 1934, 1937, 1938, 1939, 1946, 1948, 1956, 1961, 1966, 1968, 1980, 1987, 2007, 2020, 2022: in each, most industries stayed above their long averages into the top. Only two looked like today.

So narrow breadth is neither necessary for a top nor common before one. When it did appear near a high, it came with the two great narrow-leadership markets of the past century. That is the uncomfortable part of today’s reading, and the reason I would not dismiss it.

How sturdy is the finding?

The thresholds are choices, so I reran the scan on 80 nearby settings, moving the 50-day cutoff from 20% to 35%, the longer cutoffs from 40% to 55% and the distance from the high from 2% to 5%. 1972 appears in 95% of them, 2000 in 56% and 2023 in 51%. 10 years appear in at least a quarter of the settings, and the market was higher a year later in 7 of them. On a second survivorship-free universe, the 50 largest size and value portfolios, the strict setup finds 1972 and 2023, without 2000.

The honest summary is that the strict version of the setup is rare and has a poor record, while looser versions are common and ordinary. Today’s reading is the strict version.

The decision is probably still ahead

One thing all three episodes share is that the setup did not mark the high. The index kept climbing each time: 9.6% over 106 days after September 1972, 11.8% over 16 days after March 2000, 9.8% over 61 days after May 2023. The bearish cases and the bullish one looked the same for weeks to months after the signal.

Each of those rallies then ended in a 10% decline, in April 1973, April 2000, October 2023. What separated the endings came after that first drop. In 1973 and 2000 it was the start of a bear market that took the index down by almost half. In 2023 the decline stopped near 10%, the rest of the market joined the recovery, and the index went on to new highs.

So I read today’s setup as a reason to pay attention, and I would not act on it alone. On this record the market has had room to run after the signal, and the real test has come at the first meaningful pullback: whether breadth repairs itself as the index falls, or the index follows its members down.

What would change my read

Toward the 2023 ending: breadth recovering, with more than half of S&P 500 members back above their 50-day average while the index holds near its high. That is what broadening looks like, and it is how 2023 resolved.

Toward the 1972 and 2000 endings: the index falling 5% or more while breadth stays weak. That would mean the largest companies have stopped carrying the market and nothing has replaced them.

How we checked it

• The setup. Fewer than 27% of stocks above their 50-day average, fewer than 50% above their 100- and 200-day averages, with the index within 3% of its 52-week high. It is the reading the S&P 500 member series showed in September 2026. A new episode starts after 21 sessions without a match on the member series and after six months on the industry series.
• S&P 500 members. The share of today’s members with enough history above each simple moving average, daily since 2010, with SPY price returns. Carrying today’s members back biases older readings upward.
• Industry proxy. Ken French’s 49 value-weighted industry portfolios (CRSP release 202608), turned into total-return indexes; an industry counts once it has enough history for each average. The market is CRSP’s value-weighted return with dividends. Its 52-week high uses the same series.
• Robustness. 80 threshold sets on the industry series, and the strict setup on the 50 portfolios in the five largest size deciles of French’s 10×10 size and book-to-market set.
• Tops. S&P 500 daily closes since 1928, with a bear market after a 20% decline and a new bull after a 20% rise. Peaks that never came within 3% of a 52-week high are excluded. The industry data runs through August 31, 2026, so today’s reading appears only on the member series.
• Frozen evidence. The tables and chart use the snapshot of September 26, 2026. Only the latest-reading box updates.