Goldman Sachs said July in U.S. equities looked more like a liquidation of positions than a breakdown in headline indexes. The S&P 500 held its footing this week, traded within a 3.5% range for the month, and stayed less than 2% below its high. At the same time, equal-weight S&P 500, low-volatility S&P 500, and an ex-AI version of the index all hit record highs this week.
In a recent market note, Tony Pasquariello, head of hedge fund coverage at Goldman Sachs, wrote: “After high-velocity trading experienced a truly parabolic rise, over the past month, a heavy hammer smashed through consensus positions; I am inclined to think this frenzy has cooled.” His point was not that risk had disappeared, but that the most crowded, easiest-to-lever, and most reflexively chased trades had already been forced to cool off.
Calm indexes masked violent moves underneath
Goldman said the easiest mistake in July was to look only at the S&P 500. The index itself did not flash panic. On the surface, it was a contained move. Beneath that, active managers were dealing with something very different as popular momentum trades, the AI chain, Korean equities, and Asian long-short strategies all went through leverage compression.
The split was visible in the numbers. The S&P 500 saw average daily moves of less than 1% this week, while Goldman’s flagship momentum basket recorded average daily volatility close to 10%. On June 22, Goldman’s TMT momentum basket was still up 145% for the year. It then suffered the worst drawdown on record before posting a 17% rebound in a single day. Asian fundamental long-short funds, after a record first half, were hit with their biggest monthly drawdown in the past decade, while South Korea’s KOSPI jumped 18% overnight.
Goldman’s framing was that the issue was not how much the market fell on any one day. The problem was that some of the most profitable trades from earlier in the year suddenly lost liquidity. Investors positioned in the S&P 500 saw stability. Investors concentrated in high-momentum technology names saw swings that were close to disorderly. That was the key divide in July: index-level waters stayed fairly calm, while a batch of position-level trades had already capsized.
Deleveraging went beyond ordinary rebalancing
Goldman said several data points showed that this was more than a routine portfolio adjustment. Global technology exposure went through the biggest selling wave in more than five years. Assets under management in Korea equity leveraged ETFs stood at $53 billion at the June peak and have now dropped to $15 billion. Goldman’s prime brokerage business also recorded the largest reduction in total exposure since late 2022.
More granular positioning data pointed the same way. Fundamental long-short clients’ leveraged exposure to the momentum factor fell to the 28th percentile of the past year’s range. In Goldman’s description, crowded trades had shifted from a situation where nearly everyone was still on board to one where a meaningful share of investors had already stepped off, or had been forced off.
The bank did not say painful reversals were gone for good. It said that compared with early July, the impulse to chase upside had clearly faded, while cash balances and discipline had increased.
AI trades are now being judged on returns, not just narrative
In the second half of July, Goldman said the pressure on AI trades was no longer just about profit-taking. A more basic question moved to the center: can the huge AI capital expenditures from hyperscale cloud companies generate returns that are clear enough and durable enough?
Market skepticism around that question increased last week. This week’s answers were mixed, but better than the most pessimistic version. Meta did not show that material AI returns were already in hand. Microsoft, by contrast, gave a clearer signal that capital spending was converting into revenue and AI products, and doing so at scale. Amazon then reported re-accelerating AWS growth and expanding cloud margins. Credit spreads on hyperscaler bonds tightened at the same time.
Goldman said that shift mattered. If the AI trade is reduced to massive spending with distant returns, valuations come under pressure. If some companies can show that investment is already starting to turn into revenue, the market is less likely to treat the entire AI chain as one undifferentiated trade. Dispersion has already appeared. The phase in which an AI label alone could lift valuation multiples has become much harder after this washout.
Long-end rates are back as an equity problem
After the FOMC meeting, equity traders did not get much relief. Moves in the long end of the U.S. Treasury curve briefly spilled into stocks. Goldman said the tougher issue was not only rates themselves, but also the change in the Federal Reserve’s communication style.
Markets had grown used to a higher level of transparency. Now the setup looks more restrained, with less explicit guidance. That leaves traders trying to infer policy direction from fewer clues, which in itself adds friction.
Goldman argued that policy direction matters more than parsing every phrase. Even so, long-end rates cannot be ignored for equities, especially duration-heavy names. Valuations in AI, technology, and growth stocks are more sensitive to far-end discount rates. If pressure from the long end of global bond markets continues, a stable index backdrop does not mean the ride will be comfortable.
U.S. equities still have support, but less upside elasticity
On the broader view, Goldman did not turn bearish. The bank said U.S. stocks still have support from a healthy economy, strong earnings growth, the prospect of more constructive fund flows, and nearly $1 trillion in AI capital expenditure still moving through the system.
That helps explain why the S&P 500 was able to hold steady even as deleveraging intensified beneath the surface. The index is not free of risk. Goldman’s point was that it still has enough support factors underneath it.
But the bank did not read that as a reason for aggressive bullishness. U.S. equities still lean favorable in direction, it said, while risk-reward sits in the middle of the range and the upside elasticity for global equities is weaker than it was in the previous phase. More volatility may lie ahead in the near term. Summer liquidity is not well suited to shifting risk, and once a trade becomes crowded, illiquid, and structurally complex, moves can be amplified. At the portfolio level, Goldman said the better adjustment is to raise liquidity and reduce complexity rather than keep chasing the steepest trades.
The Nasdaq 100’s message: the bull market remains, but the path is harder
Goldman used the Nasdaq 100 to summarize the market’s current condition. The index is down 8% from its June high but still up 12% for the year. Over the past nine months, it has fallen in six of them, yet is still up 9% on a point-to-point basis. Its price-to-earnings ratio has also retreated to the low end of its range from the past few years.
Those figures capture the setup clearly. The trend has not broken, but the path has become much harder to sit through. For traders, the destination and the route are not the same thing. Goldman’s conclusion was that the Nasdaq 100’s primary bull trend remains in place, but if the market continues to follow a pattern of rally, position washout, and repair, making money will be harder than simply calling the direction right. July, in its view, already delivered that warning: the market does not reward crowding, and it does not forgive leverage.