Goldman Sachs said internal divergence in the S&P 500 has returned to levels seen during the dot-com bubble, but historical backtests do not show that weak breadth automatically leads to a market pullback. The gap has widened alongside rising U.S. Treasury yields, making the move look more like a rates shock than a simple case of AI concentration.
The S&P 500 is sitting less than 2% below its record high, yet the median stock in the index is about 16% below its own 52-week high. In a Sept. 28 note to clients, Goldman Sachs U.S. equity strategist Ben Snider said market breadth has fallen to its weakest level since the dot-com era: a small number of AI giants with strong earnings are holding up the index, while constituents that are more sensitive to the real economy and interest rates have lagged.
How breadth is being measured
Market breadth tracks how many stocks are participating in an advance. The report referenced three common ways to measure it:
- the share of index constituents trading above moving averages,
- the advance-decline line, which cumulates daily advancers minus decliners,
- and Goldman’s preferred gap between the median stock and the index, based on how far each is from its 52-week high.
Backtest author Dave Johnson, using Norgate Data, found that as of Sept. 25 only 27.4% of S&P 500 constituents were above their 50-day moving average. Given how close the index was to its high, that was a record low. The share above the 200-day moving average was 48.9%. The last time a similar combination appeared was in April 2000, when the dot-com bubble began to break.
Rates moved up as smaller stocks fell behind
From the Aug. 13 record high through Sept. 25, the S&P 500 fell 0.7%. The S&P 400 mid-cap index dropped 6.8%, and the S&P 600 small-cap index lost 7.5%. Over the same stretch, the U.S. 10-year Treasury yield rose from 4.64% to 5.18%.
On Sept. 29, the yield climbed to 5.24%, its highest level since June 2007. The S&P 500 fell 0.77% that day.
The note also said Goldman’s sentiment indicator tracking U.S. equity positioning had fallen to -0.9, matching the March low. The S&P 500 forward price-to-earnings ratio dropped from 22x to 19x, back near its 10-year average.
Snider wrote, 「These factors taken together suggest that if macroeconomic uncertainty declines, the market has the potential not only to move higher broadly, but also for recently lagging stocks to catch up.」
Weak breadth has not always led to losses
Backtest reviewed 13 cases from 1998 to 2024 in which the S&P 500 was near its highs while breadth was weak. The average returns over the following 1, 3, 6, and 12 months were +1.5%, +0.5%, +5.1%, and +10.9%, respectively.
Goldman’s similar warning from May 6 pointed to a different historical pattern. Since 1980, after sharp breadth contractions, the S&P 500 saw an average 10% drawdown within the following 12 months. In that comparison, drawdown refers to the gap between the high and low during the period, while the backtest above measures end-period returns.
Goldman says the timeline fits a rates shock better than an AI-only story
The timing does not fully support the idea that AI concentration alone explains the divergence. Data through Aug. 21 showed the equal-weight S&P 500 was still ahead of the market-cap-weighted version by about 3 percentage points for the year. The reversal was concentrated in the month or so after mid-August, in step with the rise in yields.
That, according to the report, makes the current split look more like a rates shock than a pure AI concentration trade.
What to watch next
The report said weakening breadth is a risk condition, but not necessarily a sell signal. Based on the data cited, near-term volatility risk looks elevated, while the medium-term picture is not definitively bearish. The key variable is still yields.
Goldman pointed to two sets of signals. A constructive turn would include the 10-year yield pulling back from 5.24%, the share of S&P 500 stocks above their 50-day moving average rebounding from 27.4%, and the equal-weight S&P 500 narrowing its gap with the market-cap-weighted version. A more negative turn would involve yields making fresh highs while the “Magnificent Seven” also lose momentum, weakening the final pillar supporting the index.
Snider added, 「We believe investors should focus on stocks where they hold differentiated views on long-term growth prospects, as a way to generate excess returns.」