Smart money, crowded trades?

Approaches to stock selection vary widely across the hedge fund universe, even among managers practicing the same strategy. Valuation, potential catalysts, time horizons and technical factors based on price history all influence decisions to own a stock, both in terms of names and position sizing and hedging. A name that is attractive to one manager might leave another unmoved, or escape consideration altogether.

Such variation is a natural result of underlying diversity in core competencies and sources of “edge,” as well as manager DNA. For example, some might be strict adherents to value in the tradition of Warren Buffett or Benjamin Graham, while others target attractive growth stories. Other managers might have grown up on bank merger desks and focus on stocks with the potential for corporate actions, such as acquisitions or spinoffs. Still others may specialize in sectors, use quantitative screening methods or seek out activist opportunities, holding large concentrated positions, often over multiyear horizons.

For investors constructing multimanager portfolios, there is a strategic benefit to taking advantage of this diversity. Narrow areas of specialization can be combined in the same portfolio to generate aggregated underlying positions that reflect differentiated sources of alpha.

And yet, in a multimanager portfolio, duplicate positions are not uncommon and are, perhaps, inevitable. The sources of such overlap are far from clear. Is it random, or are there gravitational forces pulling managers into the same names? How extensive is overlapping, and is it necessarily bad? How does an investor measure and manage this “risk”?

Investors might be inclined philosophically to respond to overlap in a few ways. They might, for instance, infer that if more than one manager is drawn to the same stock it only strengthens the investment thesis and, perhaps, makes that name more attractive. A reasonable conclusion, but one must also imagine the potential downside. It introduces an additional form of risk — one that relates to being widely held and, in some small way, connects a manager's performance and buy-and-sell decisions to hundreds of other hedge funds.

Goldman Sachs Group Inc. publishes a “VIP” index of the 50 U.S. equities most widely held by hedge funds based on brokerage and 13F filings. The firm also publishes a counterpart “VISP” index of the most widely held short positions. Based on the premise that the performance of commonly held positions does, in fact, influence a portfolio, the GS VIP index can then be applied to estimate the effects.

The VIP index has a number of interesting properties. First, it has outperformed both the broad Standard & Poor's 500 stock index and the VISP over its history. At least superficially, the outperformance might seem to corroborate the idea that the VIP represents “smart money,” i.e., the aggregated picks of elite managers. Insofar as it does, an investor might even welcome the exposure.

But the VIP can also underperform, and these periods have often been particularly unfavorable for hedge fund investors. Gaps in the rolling relative performance as shown in Exhibit 2 include the so-called credit crunch of 2002, the credit crisis of 2008 and, perhaps most significantly, an extended stretch in 2011, a period that many hedge fund investors associate with protracted underperformance relative to broad markets. (The chart somewhat masks the severity of these periods of underperformance as it displays rolling 12-month returns.)

VIP-like effects can also be magnified in relatively benign markets. Hedge fund investors might still have fresh memories of the so-called “spring 2014 rotation,” the period in March and April of this year that was unusual for hedge funds' relative performance. Many funds experienced disproportionately negative returns, even as equity markets generally finished flat to positive. Amid a brief but severe two-month downdraft, with widespread position shifting and rumors of emergency selling, exposure to popular names stood out as a key predictor of poor performance. And, the VIP demonstrated one of its worst two-period sequences relative to both the S&P 500 and the VISP.

For these reasons, it might make sense for hedge fund investors to evaluate VIP-like exposure as a beta risk — that is, as a source of systematic market exposure, as opposed to an unmeasured, idiosyncratic source of alpha.

One statistical approach is to consider the relative performance of the held-long VIP vs. the held-short VISP on the premise that the spread of the two indexes is sensitive to hedge-fund-specific market conditions. Put another way, in periods of accelerated market activity, names that make up these two indexes are especially vulnerable to accelerated selling on the long side and covering on the short side. While over the long term the VIP should outperform the VISP, technical pressure may create the opposite effect and, for that reason, the spread may make an interesting beta factor.

Looking more closely, the factor (VIP-VISP) is a statistically significant predictor of returns of the HFRI Equity Hedge index. Exhibit 3 represents the return of the index in terms of exposure to five common risk factors. The results suggest that for the equity hedge strategy, VIP exposure can be considered separate and distinct from the other factors that are accounted for — in this case, including small cap, value and momentum. In other words, it would not be quite correct to suggest VIP-VISP is simply a proxy for growth vs. value or small cap vs. large.

The VIP-VISP beta itself might, therefore, represent not just the idiosyncratic risks of 50 stocks, but the risks of hedge funds' accelerated buying or selling. For this reason, it might be helpful to consider VIP-VISP exposure as “popular” in benign conditions but with the potential to become “crowded” in adverse conditions, and to consider it separately from those other forms of hedge fund risk.

Such an interpretation might be supported by evidence from broader HFRI hedge fund strategy universes.

Consider the census of 4,000-plus reporting managers tracked in Exhibit 4. By universe, the chart estimates statistical effects of the exposure for each of several thousand managers after accounting for other headline risk factors specific to each strategy. On the chart, if a point is aligned with 15% on the y-axis, and 86% on the x-axis, it indicates that 15% of a manager's return can be explained by VIP-VISP exposure, and by that amount, it would be larger than 86% of other managers within that universe.

In other words, it gives a sense of distribution and scale: Where does the exposure tend to occur and how large is it?

Not surprisingly, exposure is significant for “equity hedge,” particularly for U.S. and global U.S.-based managers (but not global ex-U.S. managers). All strategies, however, reflect some degree of exposure. That includes strategies such as macro, which are not characterized by single-name stock selection. For investors constructing multimanager portfolios, however, it is meaningful that the fund-of-funds strategy demonstrates the most widespread and significant exposure, including relative to the equity hedge index itself. In that the beta risk represents exposure to other hedge funds, it makes sense that the fund-of-funds strategy would be the most sensitive to this risk. It also underlines the significance of this form of exposure for investors constructing multimanager portfolios.

This carries several practical implications. First, it is worth repeating — emphatically — that the VIP index outperforms broad equities over time, which puts a sort of asterisk on efforts to manage the factor risk it represents; as far as systematic risks go, it is, in many ways, relatively attractive. It is equally notable, however, that so-called conventional risk factors, such as small cap and value, also tend to perform positively over time (based on Fama-French SMB and HML factors). All of this suggests that this risk, too, is worth measuring and monitoring, and can improve and validate hedge fund portfolio diversification.

For investors in multimanager hedge fund portfolios, a few conclusions suggested by the evidence are:

• Some position overlap might be inevitable in a multimanager portfolio, but the degree of exposure can vary greatly among strategies and managers.
• Such exposure might be considered favorable over time, as suggested by the long-term outperformance of the VIP vs. the S&P 500.
• However, “popular” in benign conditions can become “crowded” in times of stress.
• It makes sense for investors in multimanager portfolios to measure and manage “popular position” factors as a form of beta exposure.

Kristofer Kwait is managing director and head of research, and John Delano is director-hedge fund strategies group, with the Commonfund in Wilton, Conn.