The S&P 500 continues to hit new highs, while Goldman Sachs trading desk is secretly reducing positions.

The S&P 500 just hit a new all-time closing high, but Wall Street is filled with a strange atmosphere—it’s not celebration, but caution. Last Friday’s close saw the S&P 500 reach a new high, but 324 of its components declined that day, with a net breadth reading of -148, the second-worst breadth performance at a new high in history. In other words, the index is at a new high, but most stocks are falling.

This “index rising, individual stocks falling” disconnect makes me think of the rebound after the March 2020 crash—back then, only a few tech stocks carried the index upward, and it wasn’t long before a sharp correction followed. And now, Goldman Sachs’ trading desk has already sounded the alarm.

Hedge funds’ largest deleveraging in seven months

Goldman Sachs commodity brokerage data shows that last week, the nominal leverage reduction in U.S. stocks hit a seven-month high, mainly driven by risk liquidation. The reduction in leverage was most aggressive in consumer discretionary and technology sectors, marking the third-largest single-week leverage reduction in the past five years.

What does this mean? Simply put, hedge funds are collectively “reducing positions to hedge risks.” I saw a similar scene in April 2020, when, during the rebound after the pandemic shock, hedge funds also suddenly deleveraged heavily, and the market then experienced about a 10% pullback.

Goldman Sachs trader Brian Garrett wrote in a weekend memo that hedge fund net exposure “remained relatively restrained within a +53% to -53% range throughout the year,” considering this a cautious risk management approach in a market environment full of “unknown unknowns.” In plain language: even the smartest funds are buying insurance—should retail investors also consider whether they are being too optimistic?

$25 billion passive sell-off imminent

The second warning comes from pension rebalancing. Goldman Sachs estimates that pension rebalancing at the end of April will generate about $25 billion in U.S. stock sell orders. How big is this number? It ranks among the top 15 all-time estimated sell-offs since 2000. Excluding quarterly expiration factors, this could be the largest single-month sell estimate ever.

Pension rebalancing is “passive selling,” unaffected by market sentiment—selling as much as needed. This means regardless of how the market moves next week, this $25 billion in sell orders will hit the market. I remember a similar rebalancing in October 2022, which was followed by about a 3% decline in the S&P 500 over the next two weeks.

The biggest buyers are already “fully invested”

The third signal comes from trend-following strategies (CTA). Since April, the CTA crowd has been the most important source of funds driving global stock market gains, with about $53 billion net bought into global equities this month alone, including roughly $32 billion into the S&P 500. However, Goldman Sachs futures trading desk data shows this buying momentum has already ended.

In plain terms, the “buy high, sell low” machines of the CTA group have bought enough now, and they are no longer net buyers—in fact, they are slightly leaning toward selling when markets are stable. This means the market has lost an important “automatic stabilizer.” If the market declines, CTA selling could further amplify the downturn.

Semiconductor sector’s extreme move reminiscent of 2000

The fourth signal comes from the extreme movement in the semiconductor sector. The Philadelphia Semiconductor Index (SOX) has risen for 18 consecutive trading days, setting the longest streak ever, closing Friday about 50% above its 200-day moving average. This is the most extreme deviation from the 200-day moving average since the bubble peak in 2000.

I remember in March 2000, the Nasdaq was similarly at a high—index hitting new highs but breadth extremely poor, and semiconductors soaring. What happened next? Over the following two years, Nasdaq fell 78%. Of course, the fundamentals are completely different now—AI-driven semiconductor demand is real—but the rule that “too much rise leads to fall” has never changed.

Sentiment indicators enter “stretched” territory

The fifth signal comes from Goldman Sachs’ U.S. stock sentiment indicator: investor positioning already shows signs of being “stretched.” From the derivatives market, the S&P 500’s gamma positioning is in a rare zone, with market makers exhibiting an extremely net short gamma stance on the spot. This means that if a directional move occurs, volatility could be significantly amplified.

Almost no professional investors currently hold outright bullish positions; implied volatility for call options in July was only around 12. This “upside-only” space remains a “lonely trade”—a very interesting phrase, indicating that the smartest money in the market isn’t optimistic about short-term moves.

Is a correction a buying opportunity?

Although these five warning signals point to a short-term pullback, Goldman Sachs still believes the S&P 500 will close significantly higher by 2026, and that any correction should be viewed as a structural buying opportunity. Historical data shows that since the financial crisis, whenever the S&P 500 experienced a decline of over 10% and then retouched previous highs, the average returns over the next 1 week, 1 month, and 3 months were 1.5%, 5.2%, and 8.6%, respectively.

My view: be cautious in the short term, optimistic in the long term. This week will be the busiest of the year, with both the Federal Reserve and the Bank of Japan announcing rate decisions, and about 44% of the S&P 500’s market cap scheduled to report earnings, including tech giants like Google, Microsoft, Amazon, Meta, and Apple. These events, combined with the five signals above, make short-term volatility inevitable.

But if you ask me, I’d say: a correction is an opportunity. Just don’t rush in at the first sign of a dip—wait until the market digests these risks. Investment decisions should be based on your own situation; the market always carries uncertainties.