Goldman Sachs top quantitative analyst: Buying the dip in chip and momentum stocks, with lower leverage and lighter positions now

``` Goldman Sachs’ top quantitative strategist believes that now is the time to go long on semiconductors and momentum stocks. Shawn Tuteja, Head of ETF and Custom Basket Volatility Trading for Goldman Sachs Global Banking and Markets, noted in his latest report that the recent nearly 20% drawdown in the semiconductor and AI sectors’ momentum factor is mainly driven by technical and structural reasons. He believes that with system leverage decreasing and institutional positions becoming more balanced, investors now have tactical opportunities to “testingly build positions” in semiconductor and momentum stocks on the long side. The current market consensus—largely uncontested—is that the S&P 500 year-end target above 8,000 points remains unchanged. Goldman Sachs forecasts 2027 earnings per share at $385, compared to the market average of $398. Taking the midpoint and applying a 20.5x P/E multiple still yields over 8,000 points for the index, which is within less than 7% of its current closing price. Technical Drivers for Momentum Factor Drawdown Tuteja believes that this over 20% drawdown in the momentum factor fundamentally reflects improvements in market pricing efficiency, not a collapse in AI trading logic. (Volatility-adjusted drawdown of Goldman Sachs High Beta Momentum Stock Basket) He points out that, historically, momentum strategies tend to underperform in the second half of July. As the market became increasingly aware of this pattern, “rotation trades” began earlier, starting around July 1 instead of waiting until July 18. From a volatility-adjusted perspective, this drawdown path closely matches previous sell-offs in similar realized volatility environments, just with a more rapid arrival this time. Client feedback from Goldman Sachs’ cash trading desk further supports this judgment. According to Goldman’s TMT trading desk, current client sentiment is around 7.5–8 out of 10, down from 9.5–10 a month ago. However, most clients agree that technical factors are the primary drivers of this correction. Several clients note that the semiconductor index SOX rose by about 100% between April and May, with only two corrections of roughly 5%. Thus, a period of consolidation is expected. They also note, if the market truly believed that AI trading has peaked, there would be a broader sell-off in AI stocks. Yet names like Dell and CRDO remain at high levels, and large-cap SaaS and IT service stocks have not seen obvious rebounds, which doesn’t align with a typical “top” signal. Leverage Recedes, Structural Pressures Digesting The total assets of US leveraged semiconductor ETFs dropped sharply from around $157 billion at their peak in mid-June to approximately $104 billion as of July 8, a fall of about $53 billion. This deleveraging not only represents outflows but also directly alters the market’s microstructure. (Total Assets Under Management of US Leveraged Semiconductor ETFs) At the peak, daily short Gamma exposure from just semiconductor leveraged ETFs was about $2.8 billion, meaning that a 3% single-day sector rally would force these ETFs to buy approximately $8.5 billion of semiconductor stocks to rebalance. Currently, that daily Gamma exposure has shrunk to around $1.9 billion. Notably, if we only accounted for the shrinkage implied by spot price declines, the complex’s AUM would have fallen to about $89 billion. The actual $104 billion AUM means that various investors bought an extra ~$15 billion via leveraged semiconductor ETFs during the downturn, showing “buying the dip” remains active. Meanwhile, hedge fund positioning data also points to intensified deleveraging. Goldman Sachs Prime Services data shows that in recent weeks, fundamental long-short funds have significantly cut AI and momentum exposure. Their current total leverage ratio is now at its lowest decile for nearly a year. Since June 22, these managers are down 2.2%, but are still up about 15.5% year-to-date. Additionally, institutional financing costs have risen noticeably. Recently, the one-month financing rate for popular names like SK Hynix and Samsung reached as high as the federal funds rate +12%, further reducing the appeal of holding positions during sideways markets. Single-Stock Volatility Premiums Rare, Compression Window Opening In terms of volatility, the current premium of single-stock implied volatility over index implied volatility is at its highest level in the past 20 years. Goldman data shows that the weighted average implied volatility (three-month) of the S&P 500’s 50 largest constituents is about 26 volatility points higher than the index’s implied volatility. This means the index’s implied correlation is very low; when semiconductors fall, healthcare, financials, and other sectors must be rising. (Three-month weighted average of implied single-stock volatility vs. three-month index implied volatility) Tuteja points out that since AI trading began, one-month straddle strategies (rolled weekly, delta hedged daily) on semiconductor ETFs have persistently yielded positive returns, while equivalent S&P 500 straddle strategies have persistently lost money—a sharp contrast. (Rolling 20-day realized volatility for semiconductor ETFs since the start of AI trading) He believes that as system leverage falls, daily short Gamma exposure compresses, and positioning stabilizes, implied and realized volatility in semiconductors should begin to compress as well. Currently, the skew of single-stock call options relative to puts remains attractive—an opportunity for investors who want to hold long exposure while locking in upside and mitigating downside to consider collar or put-spread collar strategies. Upside Converging, Fundamentals Shift Focus to Earnings Season Compared to a month ago when some AI stocks doubled or even tripled in short order, today’s market expectations for upside have become more rational. Goldman’s research team had previously promoted buying collar strategies (“sell out-of-the-money calls to pay for put protection”)—interest was strong in April and May, but very few trades were completed, as investors were generally afraid of missing out on further gains by selling calls. Tuteja points out that the expected 2026 capex of hyperscale cloud providers was revised up by about $100 billion in Q2, serving as a core catalyst for the April to May semiconductor rally. However, Goldman’s research team expects no similar capex upward revisions this coming earnings season. As we enter the July earnings season, clients’ focus will shift toward AI return on investment, the development of open-source models, and token usage trends. Tuteja adds that the biggest variable that may shape the market’s medium-term outlook is the path of the Fed’s potential new rate hike cycle. Mainstream equity accounts expect zero hikes, the market is pricing in two, while several macro accounts are betting on four or more—these three scenarios would have fundamentally different impacts on market leadership structures. Risk Warning and Disclaimer The market carries risk; investments must be made cautiously. This article does not constitute personal investment advice, nor does it consider an individual user’s particular investment goals, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article suit their particular circumstances. Investing based on this content is at your own risk. ```