Goldman Sachs trading desk: The sell-off of US momentum stocks is fiercer than anything seen since 2020! However, there is no sign of "panic" yet, and retail investors remain the biggest support.
U.S. high-beta momentum stocks experienced their fiercest sell-off in this round, but the market has not yet fallen into panic-driven liquidation.
Goldman Sachs trading desk described the July 7 trading as an "ugly day" in their latest recap report. Due to Samsung's revenue being slightly below buy-side expectations and news of DeepSeek developing its own AI chips, the global AI sector suffered a chain reaction sell-off. The high-beta momentum factor (GSPRHIMO) plunged about 6% in a single day, and its 5-day cumulative drop has exceeded 20%, surpassing Goldman's previous expectations of a "brief summer correction."
Goldman Sachs trader Guillaume Soria said that the speed and magnitude of this sell-off are unprecedented since the COVID-19 "stay-at-home narrative" shifted to "reopening" in 2020, but at that time there were stronger fundamental catalysts, which are lacking in this round.
Despite the ferocity of the sell-off, Goldman Sachs trading desk explicitly stated "there is no panic mode seen on the desk yet." The current de-risking remains orderly, mainly driven by systematic/factor flows, rather than panic-driven liquidation. Meanwhile, retail investors have become the market's most important support force—during the session they turned net buyers, with net inflows at the close reaching the 90th percentile of the past three years.
Notably, UBS cash trading desk data also show excess selling flow remains moderate and does not constitute a panic signal, though it warns that continued sell-off could shift the situation from orderly to panic-driven liquidation.
AI sector suffers chain reaction, momentum factor down over 20% in five days
The catalyst for this round of sell-off comes from the global AI industry chain.
On July 7, Samsung released preliminary earnings, recording a record operating profit, but revenue was slightly below buy-side expectations, and its stock fell 9% in one day in Korea. Meanwhile, DeepSeek announced it will start developing its own AI chips, further intensifying market concerns about the AI hardware landscape and triggering spillover sell-offs across the global AI complex.
Trading volume surged in semiconductor ETFs and leveraged ETFs fields. According to Goldman Sachs data, SOXL trading volume was 35% higher than its 30-day average, and this product has fallen 50% from its historical high 10 sessions ago. DRAM-related trading volume was 60% above average; SMH volume was up 30%.
Goldman Sachs TMT expert Peter Bartlett pointed out that although the overall fundamental view of the technology sector remains positive, especially among professional tech investors, buying and defensive operations have become distinctly "selective" in recent weeks, a clear cooling compared to the frenzy in May and early June.
Goldman Sachs trading desk indicated SK Hynix's upgraded guidance and META cloud business dynamics were considered triggers, but "fundamental views remain positive," which is one key reason the current situation has not evolved into panic.
Orderly de-risking, not panic liquidation, but risks remain
Despite the fierce sell-off, trading desk data from Goldman and UBS point to the same conclusion: The core of the current sell-off is deleveraging, not increasing bearish positions.
UBS cash trading desk observed that selling pressure mainly comes from long-term accounts, while hedge funds are mainly doing tactical operations—mostly short covering and selective buying, which supports the judgment that "deleverage-driven" rather than "bear-driven."
However, risks should not be ignored. Technically, high-beta momentum (GSPRHIMO) and the broad AI index (GSTMTAIP) are not yet in oversold territory at current declines, year-to-date gains are still as high as 23%; from a five-year perspective, relevant positions remain extremely crowded.
Goldman Sachs trading desk bluntly stated: In the absence of upward catalysts, there remains more room for future selling.
Additionally, July has historically been one of the worst months for momentum factor performance, and this month is on track to be the worst July on record for high-beta momentum. Goldman historical data shows that the factor's largest historical drawdown is about twice the current decline, meaning that if market leadership shifts, this round's decline could expand significantly.
As institutions retreat in droves, retail investors’ contrarian buying has become the most notable structural feature of the current market. Goldman Sachs data show retail net buying during the session continued to increase, with net inflows at the close reaching the 90th percentile in a three-year lookback, making them one of the day's most important support forces.
Meanwhile, the market's internal structure has not completely collapsed. 283 S&P 500 components closed up on the day, showing breadth support, echoing Morgan Stanley strategist Michael Wilson's earlier "market broadening" theme.
Facing the current situation, Goldman Sachs trader Guillaume Soria takes a cautious stance, arguing it is too far-fetched to call a 20% five-day drop a "healthy correction," but notes that after a strong first-half 57% performance, momentum factor positions have been reset—from historic highs a month ago down to the 60th percentile on a one-year lookback, overall performance remains strong.
Soria believes the market may currently be in the "late stage" of this adjustment, but if negative news widens the cracks, or market leadership shifts, a deeper correction may unfold.
Risk Warning and DisclaimerThe market involves risk, and investors should remain cautious. This article does not constitute personal investment advice, nor does it take into account individual users' specific investment objectives, financial situations, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investments made based on this article are at your own risk.