Hedge funds are still up 10% year-to-date, but the crowded AI trade has not fully unwound after its sharp July reversal

Overview

Using 13-F filings, Goldman Sachs Prime Services data, and short-interest data, the report tracks the performance, leverage, AI exposure, sector allocation, and popular securities of U.S. equity hedge funds from the end of the second quarter of 2026 through mid-August. The central conclusion is that funds still recorded returns of approximately 10% year-to-date, but the July reversal in AI and momentum trades exposed exceptionally high crowding and volatility. Funds subsequently reduced leverage and expanded allocations to health care, financials, and energy, although multiple risk exposures remain significantly above historical norms.

Core views

The report analyzes the holdings of 991 hedge funds at the start of the third quarter of 2026, with $5.4 trillion in total equity positions, including $3.4 trillion in longs and $2.0 trillion in shorts. The holdings analysis is based on 13-F filings submitted through August 14 and primarily reflects positions as of June 30. Goldman Sachs Prime Services estimates indicate that U.S. equity long/short funds returned approximately 10% year-to-date through August 19. Returns were strong in the second quarter, particularly in June, but weakened in July as the AI trade and momentum factor pulled back. Performance rebounded in recent weeks in August, although volatility remained very high. The GSTHHVIP basket of the most popular long positions rose 12% year-to-date, compared with a 17% gain for the equal-weighted S&P 500; elsewhere, the report shows the standard S&P 500 gaining 13%. The most concentrated short basket, GSCBMSAL, rose 41% year-to-date, indicating substantial pressure on short positions. July was one of the most severe hedge fund gross-exposure reduction events of the past decade. GSTHHVIP underperformed the equal-weighted S&P 500 by 11 percentage points from the end of June to the end of July, its worst monthly relative performance in more than 20 years of history. The basket's annualized volatility over the past three months approached 30%, while the volatility of its excess returns relative to the equal-weighted S&P 500 rose to the highest level since the global financial crisis. Although fund gross and net leverage have declined significantly from extreme second-quarter highs, they remain elevated relative to history. Broad AI net exposure is also far below its year-to-date peak but remains above every level observed before 2026. Leveraged funds hold near-record net short positions in Nasdaq 100 index futures, with related futures shorts increasing 35% since mid-June. Although the median short-interest ratio among S&P 500 constituents declined slightly from the end of June, it remains near its highest level in more than 15 years. A screen for increases in short positions shows that IREN, SOLS, and OKLO were among the Russell 1000 stocks with market capitalizations above $5 billion that experienced the largest recent increases in shorts. Fund performance in recent months has been highly synchronized with the AI infrastructure trade. Funds entered the year with highly concentrated AI allocations, and the second-quarter AI rally drove popular-position performance and crowding to record levels. The correlation coefficient between GSTHHVIP's year-to-date excess returns and the Goldman Sachs AI basket reached 0.9. During the second-quarter rise from the March low to the June high, AI stocks accumulated by more funds generally performed better. These most popular winners then experienced roughly symmetrical drawdowns during the July deleveraging. In recent weeks, the AI segments that fell the most in July also rebounded the most and helped restore fund returns, although most AI infrastructure stocks have yet to return to their second-quarter highs. Fund participation in AI did not consist of broad-based additions. Total portfolio turnover rose to its highest level since 2021 in the second quarter, while information technology holdings turnover climbed to its highest level since 2011, reflecting rapid adjustments within the theme. Technology stocks accounted for 14 of the 20 Rising Stars. AEIS, VIAV, and DLR were representative AI-related stocks with the largest net increases in fund ownership, while SNOW, STX, and SPCX entered the VIP list. At the same time, funds reduced positions in most mega-cap technology stocks and numerous semiconductor stocks. AMZN and MSFT were exceptions that received net additions, NVDA holdings were broadly stable, and META entered the Falling Stars list. The report therefore concludes that although funds embraced the AI rally during the second quarter, they also began diversifying both within AI and beyond the theme. The VIP list comprises the 50 stocks appearing most frequently among the top 10 holdings of fundamentally driven funds. This edition specifically uses 733 funds holding between 10 and 200 distinct U.S. equity positions. AMZN ranked first for the 11th consecutive quarter. Nine of the top 10 names were mega-cap technology companies, with V the sole exception. Eleven new stocks entered the list in the second quarter, including COMP, CPNG, FTAI, FWONK, SE, SNOW, SPCX, STX, TDG, and TMO. Since 2001, the basket has outperformed the S&P 500 in 58% of quarters, generating an average quarterly excess return of 50 basis points, albeit at the cost of higher volatility. Its allocation is not sector-neutral, covering only 8 of 11 sectors, with consumer staples, energy, and materials absent. Information technology has the highest weight at 32%. Sector allocation indicates that funds are expanding exposure beyond AI. Health care was the largest net sector weight at the start of the third quarter, accounting for 19% of total net exposure and representing a 962-basis-point overweight relative to the Russell 3000, with the sector tilt near a 10-year high. Funds primarily added biotechnology exposure in the second quarter, but most health care subsectors received additions, with managed care the exception. Information technology accounted for 25% of long portfolios but only 15% of net exposure, creating a substantial underweight relative to its 33% weight in the Russell 3000. The report notes that this record relative underweight mainly reflects technology's exceptionally large benchmark weight. The net financials tilt rose to its highest level since before the global financial crisis, while the net energy tilt climbed to its highest level since 2015, with additions in both sectors broadly distributed across multiple subsectors. Specialist fund holdings further illustrate preferences within sectors. The report defines a specialist fund as one that allocated more than 65% of its long equity assets to a particular sector in at least three of the past five years and held between 10 and 200 stocks. This edition identifies 23 financials specialist funds and 27 health care specialist funds. Popular holdings among financials specialists are concentrated in regional banks. The median stock on their list is held by 22% of specialist funds, compared with only 1% of generalist funds. COF is a popular holding among financials specialists and is also included in the VIP list, while CME and ICE were among the stocks with the largest net increases in the number of fund holders during the second quarter. Popular holdings among health care specialists primarily come from biopharma, with BSX and UNH among the few exceptions. AXSM entered the Rising Stars list, while TMO entered the VIP list. Changes in the number of holders and ownership concentration provide another stock-selection signal in the report. Historically, Rising Stars with the largest increases in the number of fund holders generally outperformed peers for several quarters after gaining popularity, while Falling Stars underperformed peers on average. INTC retained its Rising Star status in this edition, MU was included for a second consecutive quarter, and NRG moved from Rising Star in the prior quarter to Falling Star. When concentration is measured by aggregate hedge fund ownership value as a share of a company's market capitalization, a strategy buying the 20 most concentrated stocks in the S&P 500 has outperformed the index in 59% of quarters since 2001, with an average quarterly excess return of 144 basis points. The basket outperformed the S&P 500 by 12 percentage points year-to-date through August 19. COO, FOXA, HSIC, IEX, IT, PODD, RDDT, and TAP were newly added this quarter, but the basket is not sector-neutral, and its stocks are generally in the lower-to-middle range of the S&P 500 market-cap distribution. ETF usage is also rising. ETFs' share of hedge fund long portfolios increased to 5.6% in the second quarter, the highest level since the global financial crisis. Funds held approximately $310 billion in ETF shorts, accounting for 62% of total ETF exposure, indicating that ETFs are used more as hedging instruments than as directional investments. By comparison, individual-stock shorts accounted for only 35% of total individual-stock positions. The report also uses the GSTHVISP basket, consisting of 50 S&P 500 constituents, as a short hedge for the VIP long portfolio. The equal-weighted basket selects stocks that are not included in the VIP list and have the largest dollar short positions.

Analysis framework

The report first reconstructs funds' U.S. equity long holdings as of June 30, 2026, using 13-F filings submitted through August 14 as the data cutoff. It then combines these data with Goldman Sachs Prime Services' anonymized equity long/short fund portfolios to estimate fund returns through August 19 and leverage levels through August 18. The study compares the returns of popular-long, concentrated-short, and AI baskets and uses excess-return correlations, changes in the number of holders, ownership concentration, and historical quarterly performance to test the relationship between crowded trades and returns. At the sector level, it calculates net weights and relative tilts against the Russell 3000 while screening financials and health care specialist funds based on their long-term sector asset allocations. Finally, the report constructs VIP, Rising Stars, Falling Stars, high-concentration, and hedge baskets and combines them with July 31 short-interest data and ETF positions to assess risk allocation.

Methodology notes

• Event-Driven Strategy and Behavioral FinanceFund Flow/Positioning Analysis

  Analysis of 13-F Holding Changes and Number of Fund Holders

  The report compares the number of funds initiating or increasing positions in a stock with the number reducing or exiting positions, uses the net change in the number of holders to measure stock popularity, and identifies Rising Stars and Falling Stars.

• Quantitative/Factor/Portfolio Theory

  Popular-Holdings Baskets and Historical Quarterly Backtests

  The report constructs rules-based baskets using factors such as the frequency with which stocks appear among funds' top 10 holdings, ownership concentration, or short-position size, and then calculates their historical hit rates, quarterly excess returns, and year-to-date performance relative to the S&P 500.

• Quantitative/Factor/Portfolio Theory

  Excess-Return Correlation Analysis

  The report compares the daily excess returns of the VIP and AI baskets relative to the equal-weighted S&P 500. The 0.9 year-to-date correlation coefficient demonstrates that popular fund positions and the AI trade were highly synchronized.

• Sector/Industry Analysis Framework

  Net Sector Tilts and Specialist Fund Screening

  The report compares funds' net sector weights with the Russell 3000 and defines as specialist funds those that allocated more than 65% of long assets to the same sector in at least three of the past five years, thereby identifying preferences within sectors.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Hedge Fund VIP Basket (GSTHHVIP)

  Represents the 50 popular long positions appearing most frequently among the top 10 holdings of fundamentally driven hedge funds.

  Strengths

  Outperformed the S&P 500 in 58% of quarters since 2001, with an average quarterly excess return of 50 basis points.

  Weaknesses

  Rose 12% year-to-date but lagged the S&P 500 benchmarks shown in the report; annualized volatility over the past three months approached 30%.

  Comparison

  Underperformed the equal-weighted S&P 500 by 11 percentage points from the end of June to the end of July, its worst monthly relative performance in more than 20 years.

  Risks

  The basket is not sector-neutral, has a 32% information technology weight, and the report explicitly notes that its historically high returns have been accompanied by high volatility.

• Concentrated Short Basket (GSCBMSAL)

  Represents hedge funds' most concentrated short positions.

  Weaknesses

  The basket rose 41% year-to-date, indicating substantial pressure on short portfolios.

  Comparison

  Its year-to-date gain was significantly higher than the VIP popular-long basket's 12% return.

• High-Concentration Basket (GSTHHFHI)

  Comprises the 20 S&P 500 stocks with the highest aggregate hedge fund ownership as a share of company market capitalization.

  Strengths

  Outperformed the S&P 500 in 59% of quarters since 2001, with an average quarterly excess return of 144 basis points; outperformed by 12 percentage points year-to-date through August 19.

  Weaknesses

  Constituents are generally in the lower-to-middle range of the S&P 500 market-cap distribution.

  Comparison

  COO, FOXA, HSIC, IEX, IT, PODD, RDDT, and TAP were newly added this quarter.

  Risks

  The basket is not sector-neutral.

• AMZN, MSFT

  When funds reduced positions in most mega-cap technology stocks, these two were exceptions that received net hedge fund additions during the second quarter.

  Strengths

  AMZN ranked first on the VIP list for the 11th consecutive quarter.

  Comparison

  Compared with most mega-cap technology stocks, AMZN and MSFT showed stronger fund positioning trends.

  Risks

  They remain part of a holdings group highly correlated with popular technology and AI trades.

• AEIS, VIAV, DLR

  Among the AI-related stocks with the largest net increases in hedge fund popularity during the second quarter.

  Strengths

  Ranked among the AI stocks with the largest increases in the number of fund holders or allocations.

  Comparison

  AI stocks receiving larger additions generally performed better during the second-quarter rally but also broadly underperformed during the July drawdown.

  Risks

  Their performance is highly correlated with the AI theme and hedge fund deleveraging.

• Health Care, Financials, and Energy Sectors

  The primary sector allocation areas that funds expanded beyond AI.

  Strengths

  The health care tilt was near a 10-year high, financials reached the highest level since before the global financial crisis, and energy reached the highest level since 2015, with additions broadly distributed.

  Weaknesses

  Managed care was the exception within health care that did not receive broad additions.

  Comparison

  Health care net exposure accounted for 19% of total net exposure and was overweight the Russell 3000 by 962 basis points.

• ETF Positions

  Funds primarily use ETFs as portfolio hedging instruments rather than purely directional investments.

  Strengths

  ETFs provide a concentrated channel for portfolio hedging.

  Weaknesses

  ETFs' share of long portfolios rose to 5.6%, indicating that their use in fund assets reached the highest level since the global financial crisis.

  Comparison

  ETF shorts accounted for 62% of total ETF exposure, while individual-stock shorts accounted for only 35% of total individual-stock positions.

Key data

• Number of Funds in Sample991Holdings analysis sample at the start of the third quarter of 2026
• Total Equity Positions$5.4 trillionIncludes $3.4 trillion in longs and $2.0 trillion in shorts
• Year-to-Date Return of U.S. Equity Long/Short Funds+10%Goldman Sachs Prime Services estimate through August 19, 2026
• GSTHHVIP Year-to-Date Return+12%The report also shows the equal-weighted S&P 500 at +17% and the standard S&P 500 at +13%
• Concentrated Short Basket Year-to-Date Return+41%GSCBMSAL through August 19, 2026
• VIP One-Month Relative Drawdown-11 percentage pointsRelative to the equal-weighted S&P 500 from the end of June to the end of July, the worst in more than 20 years
• VIP Annualized Volatility Over the Past Three MonthsNearly 30%Volatility of popular hedge fund longs increased significantly
• VIP and AI Basket Correlation Coefficient0.9Correlation of year-to-date excess returns relative to the equal-weighted S&P 500
• Change in Nasdaq 100 Futures Shorts+35%Since mid-June 2026
• Health Care Net Sector Weight19%Share of total fund net exposure and the largest net sector weight at the start of the third quarter
• Health Care Tilt Relative to Russell 3000+962 basis pointsNear a 10-year high
• Information Technology Long and Net Weights25% / 15%Information technology has a 33% weight in the Russell 3000
• ETF Share of Long Portfolios5.6%Highest since the global financial crisis
• ETF Short Positions$310 billionAccounts for 62% of funds' total ETF exposure
• Historical Share of Quarters in Which VIP Outperformed58%Since 2001, with an average quarterly excess return of 50 basis points
• Historical Performance of High-Concentration StrategyOutperformed in 59% of quartersAverage quarterly excess return of 144 basis points since 2001
• High-Concentration Basket Year-to-Date Relative Performance+12 percentage pointsRelative to the S&P 500 through August 19, 2026
• Specialist Fund Screening Results23 financials funds and 27 health care fundsRequires a sector asset share of at least 65% in at least three of the past five years

Impact & implications

The report shows that positive year-to-date hedge fund returns do not mean crowding risk has disappeared. Popular longs, leverage, and AI exposure remain above long-term norms, and July's performance demonstrated that reversals in these variables can amplify drawdowns simultaneously. Broad additions to health care, financials, and energy indicate that portfolios are moving beyond a single AI trade, but information technology still dominates popular longs. Meanwhile, ETF shorts, Nasdaq 100 futures shorts, and elevated individual-stock short ratios show that funds are also maintaining substantial hedges. Historical data for the Rising Stars, VIP, and high-concentration baskets support using changes in institutional holdings as signals for subsequent performance, although the report also emphasizes that these baskets exhibit sector biases and high volatility.

Risks

• 13-F data do not include complete international holdings and therefore cannot represent funds' global equity portfolios.
• 13-F filings do not fully disclose short positions, requiring the report to supplement its analysis with Prime Services data, futures positions, and published short-interest data.
• Holdings data are time-lagged: core 13-F positions are as of June 30, the filing cutoff is August 14, and short-interest data reflect positions as of July 31.
• Goldman Sachs Prime Services data are derived from aggregated client positions and should not be regarded as a comprehensive description of the entire market.
• The VIP and high-concentration baskets are not sector-neutral and may be significantly influenced by technology weights or mid-cap stock characteristics.
• The VIP basket's historically higher returns have been accompanied by high volatility, and recent excess-return volatility has risen to its highest level since the global financial crisis.

What to watch

• Track whether fund gross leverage, net leverage, and net AI exposure continue declining from second-quarter highs or rise again during a market rebound.
• Monitor the synchronization between GSTHHVIP and the AI basket and whether popular AI stocks can return to their second-quarter highs after the July drawdown.
• Watch whether near-record net shorts in Nasdaq 100 futures and elevated short-interest ratios among S&P 500 stocks continue to increase.
• Track whether elevated net tilts toward health care, financials, and energy persist and whether the record information technology underweight relative to the Russell 3000 narrows.
• Monitor the next round of changes to Rising Stars, Falling Stars, and VIP constituents, particularly stocks showing changes in the number of fund holders in the same direction for multiple consecutive quarters.
• Watch ETFs' share of long portfolios and the size of ETF short positions to assess whether funds' reliance on portfolio hedging instruments increases further.