How Hedge Funds Navigated the Start of COVID-19 Volatility
Blog post
May 8, 2020
- We used MSCI's Hedge Fund Intel data to examine the positioning of hedge funds leading up to and during the initial part of the COVID-19 crisis.
- We saw an increase in exposure to the information-technology and communication-services sectors, while exposures to other style factors remained stable.
- With one exception, we found that high-conviction long portfolios outperformed high-conviction short portfolios, even in the volatile markets of February and March.
Sourcing MSCI Hedge Fund Intel
We used MSCI Hedge Fund Intel data1 to analyze holdings of hedge funds on the MSCI HedgePlatform that have agreed to make their holdings available on an aggregated basis. The data contains metrics designed to identify high-conviction long and short positions: A high value on these metrics suggests high conviction in a stock by the hedge-fund community.2
Hedge-Fund Conviction Metrics
Metric | Description |
|---|---|
Metric Absolute Weight | Description Weight in the aggregate hedge-fund portfolio, computed separately for long and short positions |
Metric Active Weight | Description Measure of the most significant bets away from the market |
Metric Number of Owners | Description Number of hedge funds with either a long or short position in a given stock |
Metric Effective Number of Owners3 | Description Measure of the diversification of hedge-fund owners |
Metric Days to Liquidate | Description Number of days required to liquidate a position at average daily volume; the more days needed to liquidate a position, the higher-conviction it is |
Metric Black-Litterman Implied Active Return4 | Description Expected excess return on each stock, taking correlations into account; derived using reverse optimization |
Sector and Factor Exposures
To understand how hedge funds adjusted their positions in response to the initial sell-off, we began by aggregating hedge-fund holdings to obtain the net exposure at the stock level. Next, to account for change in portfolio exposures due to market movement, we adjusted the exposures at the end of January by market returns in February to calculate the hypothetical exposures at the end of February if there had been no portfolio rebalances. We then compared these exposures against the actual exposures at the end February to measure the change due to portfolio repositioning.
From a sector perspective, we saw a general increase in exposure across all sectors.5 However, the increase was not evenly distributed. Information technology and communication services were the two sectors with the largest increase in February. The net exposure to the energy sector at the end of February was very small. Interestingly, information technology and communication services were among the better-performing sectors in March, while energy was the worst-performing sector over the same period.
Net Sector Exposures of Hedge Funds
Using MSCI's FaCS® data to observe hedge-fund factor exposure, we did not observe any significant changes. On aggregate, hedge funds had positive exposure to growth, momentum, volatility and liquidity and negative exposure to yield, value, size and quality.
Net FaCS Exposures of Hedge Funds
Performance of High-Conviction Long and Short Portfolios
In previous research, we constructed a portfolio of high-conviction long positions and a similar portfolio for short positions for each of the six conviction metrics defined in the MSCI Hedge Fund Intel data (a total of 12 hypothetical portfolios). For a given metric, we formed the high-conviction long (short) portfolio by ranking the long (short) positions by the metric, selecting the top 10% of stocks and weighting them by market capitalization. In this previous work, we demonstrated that these high-conviction long portfolios outperformed the corresponding short portfolios between July 2014 and March 2018.6
When we looked at how these portfolios performed as the market came under significant stress during February and March, we saw that, while their returns were all negative, the high-conviction long portfolios outperformed for all metrics except "days to liquidate."7
Total Returns for Hypothetical High-Conviction Long and Short Portfolios
Data from February through March 2020.
Why Such Divergence Between Long and Short Positions with the Highest Black-Litterman Values?
The hypothetical high-conviction long portfolios formed by selecting stocks with the highest Black-Litterman implied active return outperformed the corresponding short portfolios by 17.3% over the period. To better understand the return differential, we used performance attribution to identify the top contributors to the active return of the high-conviction long and high-conviction short portfolios, which we will refer to as "BL Long" and "BL Short," respectively.
As we can see in the exhibit below, the top three positive contributors to the active return of BL Long were the China international country factor, dividend yield and leverage. The top three negative contributors to the active return of BL Short were the oil-and-gas exploration and production industry factor, dividend yield and the Brazil country factor.
Top 10 Positive Contributors to Active Return of the BL Long Portfolio
Data from February through March 2020.
Top 10 Negative Contributors to Active Return of the BL Short Portfolio
Data from February through March 2020.
Hedge funds play a large role in the global capital markets. However, there is limited information about their portfolio holdings. Using MSCI's Hedge Fund Intel data for January and February, we saw that hedge funds, in aggregate, increased their net exposure to the information-technology and communication-services sectors as the sell-off in the equity markets intensified. We used our conviction metrics to create hypothetical high-conviction long portfolios and short portfolios. With one exception, the high-conviction long portfolios outperformed the high-conviction short portfolios, even when the market was under significant stress in February and March.
The authors thank George Bonne and Roman Kouzmenko for their contributions to this blog post.
Further Reading
1The MSCI Hedge Fund Intel data has a reporting lag of 45 days. For that reason, the holding data for February became available in the middle of April.2For a complete description of the data and the conviction metrics, please see: Kouzmenko, R., Kumar, N., and Bonne, G. 2018. “Anatomy of Hedge Fund Portfolios.” MSCI Research Insight.3For a given stock, we calculate the effective number of owners as the inverse of the Herfindahl-Hirschman Index (HHI), 1/∑j (ωj)2, where ωj is hedge fund ’s fractional share of the combined holding in the stock by all hedge funds.4The Black-Litterman implied active return is calculated as ∑(ωnet – ωbmk) where ∑ is the stock covariance matrix from the MSCI Barra Global Equity Model, ωnet is the portfolio weight vector of a portfolio formed by aggregating all hedge-fund holdings and ωbmk is the portfolio weight vector of the MSCI ACWI IMI.5The sector exposures were calculated using net portfolio weights. The net weight of a stock was positive (negative) if hedge funds, on aggregate, had a net long (short) position in it. The sector exposures increased across the board because the proportion of net long positions in the aggregated hedge-fund portfolio increased slightly from January to February.6This report may contain analysis of historical data, which may include hypothetical, backtested or simulated performance results. There are frequently material differences between backtested or simulated performance results and actual results subsequently achieved by any investment strategy. The analysis and observations in this report are limited solely to the period of the relevant historical data, backtest or simulation. Past performance — whether actual, backtested or simulated — is no indication or guarantee of future performance. None of the information or analysis herein is intended to constitute investment advice or a recommendation to make (or refrain from making) any kind of investment decision or asset allocation and should not be relied on as such.7In the backtesting analysis, we formed the high-conviction portfolios at the end of January and February and observed their returns in the following month. We added a two-month lag to the conviction metrics for the backtesting to be realistic. For example, the high-conviction portfolios for the end of January were formed based on the conviction metrics for November 2019.
The content of this page is for informational purposes only and is intended for institutional professionals with the analytical resources and tools necessary to interpret any performance information. Nothing herein is intended to recommend any product, tool or service. For all references to laws, rules or regulations, please note that the information is provided “as is” and does not constitute legal advice or any binding interpretation. Any approach to comply with regulatory or policy initiatives should be discussed with your own legal counsel and/or the relevant competent authority, as needed.