In the past month, the US stock market hit 14 new highs, with momentum stocks surging. Goldman Sachs has uncovered a pattern from the past 40 years: similar market trends usually lead to a decline one month later.

The AI wave is pushing the US stock market towards a highly concentrated one-sided market. The S&P 500 index has set 14 consecutive historical highs in the past month, with a cumulative increase of 10% so far this year, but this rally is almost entirely driven by technology and AI-related stocks, and market breadth has narrowed to one of the lowest levels in decades. According to Wind Data, Goldman Sachs' weekly US stock strategy report released on May 15 stated that the TMT sector, centered on technology, media, and Amazon and Tesla, contributed 85% of the S&P 500's gains this year, with the index rising only 3% after excluding TMT. Nvidia, a company, accounts for about 9% of the S&P 500's market capitalization but contributed 20% of the index's total returns this year. At the same time, Goldman Sachs' momentum factor (GSMEFMOM) has surged 25% in the past three months, making it one of the strongest rallies on record, with total leveraged and net exposure of hedge funds to the momentum factor nearing five-year highs. Goldman Sachs warns that similar momentum spikes have occurred 11 times since 1980, and after such a spike, the momentum factor typically peaks and falls back about a month later. When the S&P 500 is near a high, such momentum spikes often indicate lower-than-average stock market returns in the coming months. Goldman Sachs maintains an end-of-year target of 7600 points for the S&P 500, with only about 1% upside from current levels. Market concentration reaches extremes, "One Big Trade" dominates The structural characteristics of the current US stock market have attracted widespread attention. According to the Goldman Sachs report, while the S&P 500 has set 14 consecutive historical highs in the past month, the proportion of index constituents trading above their respective 200-day moving averages has been declining. Currently, the median stock in the S&P 500 is about 13% lower than its 52-week high, and market breadth has shrunk to one of the narrowest levels in decades. The core of this pattern is AI trading. The information technology sector has contributed about 659 basis points to the S&P 500 this year, accounting for 66% of the total return; the communication services sector has contributed 132 basis points, accounting for 13%. The top ten contributing stocks have contributed 84% of the index's gains this year, with Nvidia, Google, Micron Tech, and Broadcom leading the way. Several fund managers have reported to Goldman Sachs that in the current market environment, it is extremely difficult to find investment opportunities unrelated to the AI theme. Goldman Sachs characterizes this phenomenon as "One Big Trade" - the market is no longer a collection of stocks, but a highly homogenized directional trade centered around AI betting. Historical patterns after a rapid rise in momentum factors: short-term continuation, medium-term pressure Goldman Sachs' systematic study of momentum factor trends since 1980 shows that the current situation is highly similar to 11 comparable scenarios in history. In these scenarios, after a momentum factor has risen by 20% or more in three months, it usually continues for about a month, with an average additional gain of 6%, and then turns downward in the following two to three months. For the overall S&P 500, when a rapid rise in momentum occurs while the index is near a high, subsequent returns tend to be significantly weaker. Goldman Sachs data shows that in five instances of a rapid rise in momentum when the S&P 500 is within 5% of its historical high, the median returns for the index in the following one and three months are 0% and 0%, respectively, with a positive return probability of only 20% to 40%. By contrast, in six scenarios where a rapid rise in momentum occurred while the index was at a low, median returns for the subsequent three to six months exceeded 8%. The closest historical examples to the current scenario include mid-1998, late 1999, mid-2015, and the end of 2021. Goldman Sachs believes that macroeconomic trends and the outlook for AI investments will be crucial variables in determining the trajectory of the momentum factor and the broader market. A reversal in AI investment expectations, or a sudden deterioration in the macroeconomic environment, could trigger a "downward retracement" of the momentum factor; conversely, an unexpected improvement in the macroeconomic outlook could trigger an "upward retracement" of laggard stocks. Earnings outlook upgrades support the rally, but structural differentiations are clear Unlike the speculative markets of the late 1990s or 2021, Goldman Sachs points out that this rally has some fundamental earnings support. Since the beginning of the year, the bottom-up consensus estimates for 2026 and 2027 earnings per share (EPS) for the S&P 500 have been raised by 8%. The main sources of these upgrades are highly concentrated: the 2027 EPS estimates for stocks related to AI infrastructure have been raised by approximately 32% year-to-date, while the energy sector has seen a 19% increase, and excluding these two types of companies, the 2027 EPS estimates for the S&P 500 have remained almost flat year-to-date. However, Goldman Sachs notes that this sideways trend has been better than the usual downward revisions, and over the past month, the breadth of EPS revisions for all sectors of the S&P 500 has been positive, with more upgrades than downgrades. At the industry level, recent stock price performance generally aligns with the direction of EPS revisions, but there are significant deviations in magnitude. The 2027 EPS estimates for the energy sector have been raised by about 26% since before the war, but the sector's year-to-date increase has been only about 4%; on the other hand, the semiconductor sector has seen stock price gains surpass earnings revision levels, reflecting factors such as inflows of leveraged ETF funds and market expectations for long-term profit growth exceeding analysts' recent forecasts. How investors can respond: Diversified portfolio and hedging with low momentum stocks Faced with a highly concentrated AI momentum market, Goldman Sachs provides two strategies for investors. First, hold some low momentum stocks as a hedge. Goldman Sachs' research on the most extreme momentum reversals in the past 100 years shows that in these scenarios, previously lagging stocks (i.e., low momentum stocks) not only outperformed relatively but also achieved absolute positive returns. Goldman Sachs has identified 25 S&P 500 constituents that are currently at the short end of the momentum factor but have received recent upgrades in EPS estimates, for investors to consider. Second, construct an "Insensitive Portfolio." Goldman Sachs has screened stocks from the Russell 1000 index that have the lowest sensitivity to AI trading and US economic growth pricing, while also having positive EPS revisions. This portfolio accounted for only 13% of the median stock daily return variation by AI and economic growth factors over the past year, far below the 30% for the median stock in the Russell 1000. The screening results include multiple stocks from sectors such as energy, consumer staples, and healthcare, with a median market capitalization of about $25 billion and a median forward P/E ratio of about 17x. On a sector allocation level, Goldman Sachs points out that consumer staples have the lowest correlation with AI and momentum factors, while healthcare and real estate investment trusts have only a mildly negative correlation with both. Considering the potential slowdown in the economy in the coming quarters, holding defensive sectors with limited sensitivity to AI in a diversified stock portfolio has some appeal.