Tag Archives: models

An academic article entitled “Passive Hedge Funds” has recently attracted quite a lot of comment in the Financial Times, Bloomberg, and on a variety of websites. Those whose ambition in life seems to be to discredit hedge funds and their managers at every turn have, of course, latched onto it. But the paper’s title is tendentious, its argument familiar and in some places flawed, and its conclusions really quite anodyne. Investors seeking hedge fund-like exposure through liquid alternatives will find that some products are similar to those described in that article; they should examine them very carefully before investing. The purported humor of math jokes often depends on the technical use of a term that has other, more familiar meanings. Thus, my college roommate’s knee-slapper about how every integer is interesting relied on a definition of ‘interesting’ as ‘having a unique property.’ The joke took the form of a mathematical induction: 1 is the multiplicative identity, 2 is the only even prime, 3 is the lowest true prime, 4 is the lowest perfect square… so if there is an uninteresting integer, it is interesting, because it is the lowest one. Maybe you had to be there. I am reminded of this moment of boundless mirth by a paper entitled ” Passive Hedge Funds ,” by Mikhail Tupitsyn and Paul Lajbcygier. This title has, inevitably, attracted comment, including headlines such as “Study: Hedge Funds Don’t Do S**t, Suck” (gawker.com) or, with less sophistication and élan, “New Study Argues Hedge Funds are an Even Worse Scam than We Thought” (vox.com) and even more prosaically, “The Case Against Hedge Fund Managers” (ai-cio.com). With the apparent exception of the latter, these commentators were so enamored by their deeply considered wisdom that they clearly felt no need to read the paper. Because its authors are quite explicit about their idiosyncratic use of the term ‘passive.’ They even put scare-quotes around it. The commentators just missed the punchline. It is hard to dispute Humpty Dumpty: “When I use a word, it means just what I choose it to mean ─ neither more nor less.” Since they take pains to explain what they mean by it, I have no argument with the authors’ use of ‘passive.’ They might have used ‘hippopotamus,’ which is more euphonious, but lacking poetic souls, they chose ‘passive,’ and missed the opportunity for a great title. The sense in which the authors use ‘passive’ to describe hedge fund return patterns is that they have linear correlation to hedge fund β. The crux of their argument is that “A manager with genuine investment skill should not only have “passive” linear risk exposures to alternative risk factors ( i.e ., alternative beta) but should also produce enhanced returns through nonlinear ‘active risk exposures.'” This is contentious, as will be seen below, but it is simply posited as a truth rather than justified. Was their choice of ‘passive’ tendentious and self-promoting? Of course: how else would a postdoc and an associate prof at Melbourne’s #2 university get noticed in the Financial Times or Bloomberg, let alone a temple to the Muses such as gawker.com? Was it helpful? Our commentators’ complete failure to understand the authors’ intent makes it rather obvious that it was not. The Tupitsyn and Lajbcygier article is, as their review of the literature makes clear, one of a long line of academic studies that propose models for hedge fund returns. Even critics more competent than our commentators tend to latch onto these studies as “proof” that hedge funds offer little value-added. But anything can be modeled ─ conventional mutual funds, sunspot frequencies, even (allegedly) the earth’s climate. Problems arise when, as Emanuel Derman and others have noted, the models are mistaken for reality. And hedge fund β ─ against which the authors argue hedge fund managers fail to add value ─ is, at best, a very peculiar concept, and arguably a spurious one. On consideration, the authors’ argument begins to look strangely circular: hedge funds fail to add value relative to metrics that derive from their own returns. This is something like arguing that I am a lousy swimmer because I am unable to swim faster than myself. I may well be a lousy swimmer, but comparison with my own performance will not establish that. A good portion of Tupitsyn’s and Lajbcygier’s analysis is devoted to returns on hedge fund indices. In choosing these as a database, they, like many before them, commit the fallacy of composition. The fact that you can calculate a mean return from a pile of reports does not indicate that there is such a thing as an average hedge fund: it is not only possible, but likely that none of the funds analyzed exhibited the mean return. Further, there is no reason to expect continuity from one time period to another: a fund whose return was close to the center of the distribution in one period may be an outlier in the next. Hedge fund returns are widely dispersed both synchronically and over time, so that the value of hedge fund indices is pretty much restricted to service as performance metrics for specific time periods. The standard error of the mean = s/√n, where ‘s’ is the σ of the population and ‘n’ is its size. Obviously, the error is significantly higher and thus the epistemic value of the mean significantly less, the more dispersed the population is. Given the wide dispersion of hedge fund returns, the value of their average is largely restricted to the bragging rights it gives to marketers fortunate enough to work for funds that have outperformed it. The authors are aware of these limitations, and devote some analysis to the returns of individual, real world funds. They find that most funds have strong linear exposures to familiar factor influences on investment returns. They conclude that “The nonlinear risk is more pronounced in arbitrage styles and styles following multiple strategies, and it is weaker in directional styles.” This should hardly be surprising ─ arbitrage is inherently non-linear ─ and it is not at all clear why the presence of linear risk in other sorts of strategies should somehow suggest dereliction of duty on the part of their managers. If, for example, a dedicated short fund carried no (negative) equity exposure, its investors would certainly have reason to object! Admittedly, fewer long/short funds make use of their ability to add value by adjusting their net exposure than might be expected, and with relatively stable long/short ratios, their exposure to equity risk factors would, of course, be linear. The same would be true of any long-only equity fund, and would certainly not attract criticism. In fact, long/short funds have increasingly tended to pursue a trading-oriented (“risk on/risk off”) response to changes in their risk perceptions in place of making changes to their short positions. As a group, hedge funds provide us with ample reasons to criticize them. Despite declining over the last few years, fees are in most cases still too high for the service provided. Lack of transparency inhibits rational analysis and portfolio construction, while providing a breeding ground for a wide range of abuses and sharp practice. The artificial mystique that this opacity fosters is repulsively reminiscent of Ozma of Oz. However, neither an adolescent potty-mouth nor accusations of fraud are not needed to make these points forcefully and to draw the appropriate conclusions for investors. Nor are “discoveries” that hedge fund α is not a matter of otherworldly powers to bend the laws of economics to the manager’s will ─ that their skills might be very similar in both nature and quantity to the skills that conventional portfolio managers exhibit. Tupitsyn and Lajbcygier have made a small contribution to the growing literature on hedge fund replication ─ nothing less, but certainly nothing more. Theirs is only one approach to hedge fund replication, and to my mind a less than satisfactory one. Factor replication is an inherently backward-looking approach to modeling, and when applied to the return streams from hedge funds, likely to result in some rather peculiar portfolios. A technique that I suspect has much more promise is the creation of robo-managers ─ algorithmic trading techniques that mimic the trading strategies hedge funds are known to pursue. Many hedge funds, particularly CTAs, are already effectively automated. While it is illegal to steal their code, it is possible to imitate it based on an analysis of their returns. In considering an investment in liquid alternative funds, many of which are “quantitatively-driven” in ways that are rarely specified explicitly and require research to understand, the nature of the security selection technique should be given careful consideration. Approaches similar to that of Tupitsyn and Lajbcygier are worth a look, but may not deliver all that they promise; the source of the factor exposures they purport to imitate must be investigated. Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. (More…) I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.