Author Details

Howard Zhang

Howard Zhang

Howard Zhang was a vice president at MSCI

George Bonne

George Bonne

Executive Director, MSCI Research

Social Sharing

Extended Viewer

Is There a Hedge-Fund-Crowding Factor?

  • In 2020, hedge funds held heavy short positions in crowded, hard-to-liquidate stocks, including GameStop and AMC Entertainment, targeted by retail investors in early 2021.
  • We used MSCI Hedge Fund Crowding data to build a hedge-fund-crowding factor that showed statistical significance and had low correlation to traditional factors. 
  • Such a factor could help investors understand potential risks driven by crowded hedge-fund positions.

We have previously written about potential bubbles in the stock market and proposed a factor-based framework to assess the “bubbliness” or crowdedness of individual stocks. Here we approached the crowding problem from another angle — via the aggregated holdings of hedge funds. Using MSCI Hedge Fund Crowding data,1 we examined their most crowded positions and constructed a factor to measure crowding, driven specifically by hedge funds, as a way to identify potential risk.


What Were Hedge Funds’ Top Bets at the End of 2020?

We started with the asset weights in hedge funds’ aggregated long and short portfolios from MSCI Hedge Fund Crowding data. To account for an asset’s market capitalization, we normalized the holding by its weight in the MSCI ACWI Investable Market Index (IMI) and named the measure "weight multiple."2 Then we ranked stocks by this metric to find the top-10 companies on the long and short sides, as of Dec. 31, 2020, which are shown in the exhibit below.


Weight Multiples from Hedge Funds’ Aggregated Long Portfolio

Data as of Dec. 31, 2020.


Weight Multiples from Hedge Funds’ Aggregated Short Portfolios

Data as of Dec. 31, 2020.


We observed GameStop Corp. and AMC Entertainment Holdings Inc. were among the top-10 companies by weight multiple in the aggregated hedge-fund short portfolio. While other stocks had higher weight multiples, we selected these examples given the large effect these shorts had on the market, and vice versa, earlier this year.

Looking throughout 2020, we observed that GameStop’s weight multiple was around 30 in January 2020. Like many brick-and-mortar retailers, the company was hit hard by the pandemic, and by May hedge funds started significantly increasing their short positions on it. The weight multiple reached 60 in August and stayed high for the rest of the year. On the other hand, AMC’s weight multiple was low until August 2020 when it started to increase dramatically.


Full-Year-2020 Weight Multiples from Aggregated Hedge-Fund Short Portfolios


Evaluating Signs of a Potential Short Squeeze

Since hedge funds held heavy short positions in these stocks, we investigated how long it would take to liquidate their positions using the days-to-liquidate metric from MSCI Hedge Fund Crowding data.3 GameStop's days to liquidate was in the top decile for all of 2020, which suggests the potential for a large short squeeze if many of these funds had decided to exit their short positions. We also observed AMC’s days-to-liquidate decile dramatically increased starting in August when hedge funds began increasing their short positions..


Days-to-Liquidate Decile from Aggregated Hedge-Fund Short Portfolios in 2020


Factoring in Hedge-Fund Crowding

In the case of GameStop and AMC, the risk of a short squeeze became reality. So, we asked ourselves whether we could build a hedge-fund-crowding factor to measure this potential risk. To do so, we used metrics for hedge-fund conviction4 from long and short aggregated hedge-fund portfolios and defined our factor as an equal-weighted combination of them all.5 To evaluate the potential usefulness of the factor, we added it to the Barra US Total Market Equity Model for Long-Term Investors (USSLOW) and estimated factor returns at the end of each month via cross-sectional regression. Positive factor returns indicated that hedge funds’ bets performed as the managers expected — long positions outperformed and/or shorts underperformed. As shown in the exhibit below, the cumulative factor returns reached around 6% prior to the outbreak of COVID-19. Returns have declined by 3% since then, however, which may be driven by the strong performance of heavily shorted stocks.


Hedge-Fund-Crowding Factor’s Cumulative Returns

Data from September 2013 to January 2021.


Looking beyond return, the exhibit below summarizes the statistics of a hedge-fund-crowding factor and USSLOW style factors. The hedge-fund-crowding factor’s position on the list in terms of t-statistics and factor volatility suggests it could be a potential risk factor. The cross-validated R2 metric tells us the factor added additional explanatory power to the USSLOW model.


Statistics of Hedge-Fund-Crowding Factor and USSLOW Style Factors

Factor Percent of Absolute Value of T-Stat Greater than 2 Average of Absolute Value of T-Stat Annualized Factor Volatility (%) Cross-Validated R2 Gain (bps)
Beta 70.00 4.10 5.06 56.97
Momentum 56.67 3.25 3.82 42.78
Residual Volatility 48.89 2.47 3.84 17.35
Earnings Yield 45.56 2.11 2.49 12.56
Leverage 37.78 1.99 2.39 8.41
Mid Capitalization 35.56 1.84 2.40 10.54
Value 31.11 1.60 2.07 3.73
Long-Term Reversal 30.00 1.68 2.28 7.51
Liquidity 28.89 1.68 2.28 7.51
Management Quality 27.78 1.30 1.25 0.16
Hedge-Fund-Crowding 25.00 1.35 1.28 2.25
Earnings Quality 24.44 1.42 1.98 -0.15
Growth 24.44 1.41 1.38 -1.72
Dividend Yield 21.11 1.36 1.65 2.79
Prospect 20.00 1.30 1.11 -1.61
Profitability 18.89 1.29 1.59 -0.98

Factors in the table are ordered by their percent of observations with the absolute value of t-statistic greater than 2. Data from September 2013 to January 2021.

We also analyzed the cross-sectional exposure correlations between the hedge-fund-crowding factor and existing USSLOW style factors from September 2013 to January 2021. Except for momentum and dividend yield, we observed very low correlations with existing style factors. Our previous research has highlighted that our simulated aggregate hedge-fund portfolios have tended to tilt toward momentum and away from dividend yield as well.


Hedge-Fund-Crowding Factor’s Average Exposure Correlations with Existing USSLOW Style Factors


Using MSCI Hedge Fund Crowding data, we saw hedge funds held heavy short positions in crowded stocks before the massive short squeeze that occurred in January 2021, and our crowding conviction metrics indicated hedge funds might have had difficulty liquidating these positions. We also found the same Hedge Fund Crowding data could be used to create a risk factor that added explanatory power in the cross-sectional returns of stocks. Having such information may provide additional insights into potential short squeezes and other risks for individual stocks and portfolios.


The authors thank Roman Kouzmenko, Navneet Kumar, John Regino and Jay Yao for their contributions to this blog post.


1MSCI Hedge Fund Crowding data aggregates hedge funds’ long and short positions on a monthly basis, with a reporting lag of 45 days.

2The median value of weight multiples is 1. The higher weight multiples indicate stronger bets from hedge funds. A weight multiple of 60 means the asset weight in the aggregated hedge-fund portfolio is 60 times its weight in MSCI ACWI IMI.

3We define days to liquidate as the number of days required to liquidate a position at average daily volume.

4The metrics for hedge-fund conviction are absolute asset weight, active asset weight, total number of owners, effective number of owners, days to liquidation and implied active returns.

5We reversed the sign of conviction metrics from short portfolios before combining them with those from long portfolios.


Further Reading

Are (Stocks) Bubbles Rising?

Anatomy of Hedge Fund Portfolios

MSCI Integrated Factor Crowding Model

MSCI FactorLab

Peering into Peer Selection

Looking for a Better Hedge