A Paradox: Avoid Correlation by Following the Trends

A common Wall Street adage runs, “the trend is your friend.” Skeptics sometimes add to that the words, “…until it isn’t.” Commonfund has issued a new white paper that illuminates both sentiments, in the course of contending that futures funds managed by Commodity Trading Advisors (CTAs) can play a valuable part within institutional policy portfolios.

In particular, the authors, from Commonfund’s Hedge Fund Strategies Group, seek to defend managed futures against the charge that they are invidiously opaque, a “black box.” CTAs got their start as vehicles for following trends (“taking positions and letting profits run while cutting losses short” as the white paper puts it) and this still remains their most common model type.  This subset of CTAs in particular, then, is much more transparent in their operations then its “discretionary, thematic counterparts.” CTAs are data driven, and the data by which they are driven is available to the rest of the world.

The authors quote Rishi Narang, author of a highly regarded book on quantitative trading, Inside the Black Box (2009). Narang says that trend-following works from the premise that trends represent a process of consensus building. As a new idea takes hold, “the earliest adopters of this idea place their trades in accordance with it” and they get to watch in satisfaction as “a growing mass of market participants adopts the same thesis.” The early adopters, for this purpose, don’t have to be the absolute first adopters, because a price may “take a considerable amount of time” to get to the equilibrium level where the consensus has been fully discounted. Early adopters can surf the wave to the beach.

A Paradox

This, in the eyes of the Commonfund, presents something of a paradox. For managed futures are not strongly correlated to other asset classes, that is one of the bases of their appeal. If they work by trend following, and that works by consensus, how is it that they aren’t correlated?

In the table above, the DJCS Managed Futures index is used as a proxy for managed futures as an asset class. Equities are likewise represented by a proxy, specifically by the S&P 500 Index (SPX). As you can see, the correlation of those two classes [lines a and e, respectively] is actually negative: -.08. Correlation of DJCS with fixed income, represented by the Merrill Lynch Treasury Master Index [c], is also negative and even lower: -.12.

So how does one avoid correlation by adopting a consensus? The apparent paradox disappears when one considers the different time frames involved.  The table above charts correlations over a very long term, from 1994 to 2011. In shorter time frames, correlations with major asset classes are often high, and they of necessity frequently shift in magnitude, sign, and significance.

CTAs produced negative returns in the third quarter of 2008, the quarter that ended with that scary September. They were “caught unfavorably exposed to the direction of the equity market.” But managers did manage to get on the short side as the trend picked up speed into the fourth quarter. So, in that fourth quarter, the DJCS index recorded one of the most successful quarters in its history. This illustrates the general truth, that though “abrupt reversals are often missed by the CTA strategy … prolonged moves tend to be captured,” and CTA strategies work when and if they profit more from the latter period than they lose during the former.

Lookback Mountain

Commonfund also wants to lay to rest, or at least put into context, the notion that what CTAs do could also be done through the right option portfolio. The authors cite a paper by William Fung and David Hsieh in 2001 that concerned resemblance of CTA results to those of a portfolio of lookback straddles.  A lookback straddle (lookback puts combined with lookback calls) by definition pays a premium for a payoff that equals the market’s high during the lookback period, minus the market’s low during the same period. The premium is such that the payoff only rarely covers it, but on occasion covers it with plenty to spare. In short, a portfolio of straddles will lose a little bit often, and will sporadically win big. Fung and Hsieh concluded that a portfolio of such options could “capture the general characteristics of the entire family of trend-following strategies.”

Commonfund acknowledges that Fung and Hsieh had a point, but believes that they missed something, too. The risks of the two portfolios are different. The cheap risk of a trend-following strategy, as September 2008 illustrated, is a very sharp reversal of the established trend. The chief risk of straddle positions, on the other hand, is that the underlying assets might stay within a narrow range. Without a sharp move in either direction, the premium cannot be recovered.

Commonfund’s conclusion is that CTA returns “have demonstrated substantial long-term diversification properties in the context of a broad, multi-asset class policy portfolio.”

Be Sociable, Share!


Leave A Reply

← Looking for Abnormal Market Activity From Ring Knockouts to Court Deliberations: How Markets Predict →