Cyclical Trough Or Structural Impairment? Analysis and Proposed Solutions

FEATURE | MANAGED FUTURES

MANAGED FUTURES Cyclical Trough or Structural Impairment? Analysis and Proposed Solutions

By Ryan Davis, CAIA, and Barclay Leib

ystematic trend-followers, commonly Figure 1: 2008 Returns—HFRI Indexes referred to as managed futures Sfunds, long have been an important 30 component of institutional and high-net- 20 worth portfolios. These strategies’ unique ability to go both long and short in a vast 10 array of global financial markets histori- 0 cally has produced a return profile highly –10 uncorrelated to traditional markets. More Return (%) –20 recently, however, many investors have –30 come to question the asset class amid poor –40 performance in the post-global financial crisis period. In this article, we explore the recent struggles of the managed futures industry, discuss what the future may hold for such strategies, and offer suggestions for navigating this space. Source: FactSet Background The events of 2008 were a startling develop- On the heels of 2008, investors who previ- post-2008 as demand surged for tail-risk ment for many hedge fund investors; fol- ously shied away from these quantitative protection. Contributing to the growth in lowing a near-30-year period of few if any strategies poured into managed futures. managed futures assets since 2008, how- losses, double-digit declines in most hedge Assets in those strategies grew by 120 per- ever, was the fact that retail investors also fund strategies shattered an aura of abso- cent from 2007 to 2011 while the broader were given access to these funds for the first lute-return performance. When the dust hedge fund industry increased just 7.5 per- time, primarily through 1940 Act struc- settled on 2008, only two segments of the cent. As a consequence, the share of man- tures. Only one managed futures mutual industry generated substantial positive aged futures funds as a percentage of total fund existed before the financial crisis; that returns: short biased and managed futures hedge fund assets more than doubled to 9.3 number has swollen to 51 today. Total funds (see figure 1). More specifically, in a percent by the end of 2011, or $188 billion assets in the managed futures mutual fund calendar year in which the broad universe (see figure 2). Data aggregator BarclayHedge universe now stand at $8.6 billion—still a of hedge funds fell by 19 percent, the aver- estimates the current size of the managed fraction of its hedge fund peers, but not age managed futures fund rose 18 percent. futures market is even larger, at $260 billion insignificant. Many individual managed futures funds (although those figures are skewed by one were up 40 percent or more as they capital- extremely large hedge fund, Bridgewater Unfortunately for new entrants post-2008, ized on the sell-off in equities and extended Associates, which BarclayHedge classifies as the timing of their new allocations was trends in bonds, commodities, and curren- a commodity trading advisor or CTA). anything but optimal. Managed futures cies. This performance opened many inves- funds collectively posted negative absolute tors’ eyes to an asset class previously over- A spate of industry reports indicate that performance in three of the next four looked because of its intensely technical both high-net-worth and institutional cli- calendar years—the worst performance and quantitative nature. ents entered the managed futures space stretch in the recorded history of the asset

JULY / AUGUST 2014 37

(1996), for example, found that an equally Figure 2: Assets under Management weighted portfolio of either CTAs or commodity pools “increases the portfolio’s 200 Sharpe ratio by as much as 28 percent. The 180 HFRI Macro: Systematic Div. primary reason for this result is the low 160 correlation between managed futures 140 returns and the returns on other portfolio 120

illi on s ($) assets …” More recent studies such as one 100 by Ibbotson Associates (Chen et al. 2005) 80 of CTA data between 1980 and 2005 illus-

ss ets, B 60 A trated an asset class that generated superior 40 returns to U.S. equity markets with similar 20 volatility levels. Because of the uncorrelated 0 nature of this performance, however, the authors concluded that layering in an allocation to managed futures to a traditional Source: FactSet stock-bond portfolio improved the overall risk-adjusted return, i.e., improved the effi- cient frontier in a mean-variance Figure 3: Managed Futures Calendar-Year Returns framework.

30 In practice, the performance of popular HFRI Macro: Systematic Div. 25 indexes of managed futures funds supports 20 the academic research. Using a common 15 period since the inception of the Dow Jones UBS Commodity Index in February 1991, 10 the managed futures industry as repre-

Return Return (%) 5 sented by the HFRI Macro: Systematic 0 Diversified Index has generated the second highest Sharpe ratio of the major asset –5 classes, save for the broad category of hedge –10 funds. Its annualized return of 10.7 percent exceeds that of the S&P 500, and achieves it with approximately half the volatility. Source: FactSet Managed futures’ performance over this time period exceeds every major asset class class (see figure 3). Between 2008 and the Portfolios of Stocks and Bonds.” Lintner except that of hedge funds and real estate first quarter of 2013, managed futures funds concluded that “the combined portfolios of investment trusts (REITs). The average vol- generated an annualized performance of stocks (or stocks and bonds) after including atility of CTAs is also lower than every asset 0.4 percent, compared to 10.9 percent for … managed futures accounts (or funds) except hedge funds and bonds. the S&P 500 and 3.6 percent for the broad show substantially less risk at every possi- hedge fund composite. If one invested in ble level of expected return than portfolios The comprehensive risk management managed futures immediately following of stocks (or stocks and bonds) alone.” inherent to CTAs results in this mitigated 2008, the impact was not devastating on a Managed futures strategies proliferated in volatility profile, but its impact is most evi- stand-alone basis; but in the wake of strong the 1980s with the success of trend-follow- dent in the return distributions. Next to double-digit equity market returns, the ing investors such as John Henry and David fixed income, managed futures display the opportunity cost certainly has been high. Harding. most attractive drawdown characteristics of the major asset classes (see table 1). During Historical Review Since then, a range of academic studies the measured time period, only managed Managed futures investing received its first have validated the historical efficacy of futures and bonds experienced a maximum major academic backing in John Lintner’s managed futures investments and the bene- drawdown in the single digits. Although 1983 classic working paper, “The Potential fits of including them in diversified, tradi- hedge fund strategies generated a superior Role of Managed Commodity-Financial tional portfolios (McCarthy et al. 1996; return with less volatility than managed Futures Accounts (and/or Funds) in Edwards and Park 1996). Edwards and Park futures, they were not necessarily less risky,

38 INVESTMENTS&WEALTH MONITOR

Table 1: Risk/Return Statistics, February 1991–March 2013

Annualized Annualized Max Portfolio Return St. Dev. Sharpe Ratio Drawdown Skewness Kurtosis HFRI Macro: Systematic Div. 10.7 7.5 1.00 –9.3 0.15 –0.3 S&P 500 9.3 14.8 0.41 –50.9 –0.63 1.3 Russell 2000 10.4 19.2 0.38 –52.9 –0.53 1.2 MSCI EAFE 5.9 16.9 0.16 –56.4 –0.57 1.1 MSCI Emerging Markets 10.3 23.6 0.30 –61.4 –0.69 2.0 Barclays US Aggregate 6.7 3.6 0.96 –5.1 –0.28 1.0 HFRI Fund Wtd Composite 11.2 7.0 1.15 –21.4 –0.70 2.6 DJ UBS Commodity 4.9 14.9 0.12 –54.3 –0.58 2.7 Dow Jones US Select REIT 11.5 20.1 0.41 –70.5 –0.74 9.0 Source: FactSet as evidenced by a maximum drawdown of more than 20 percent. This is due to the Figure 4: Three-Year Annualized Rolling Returns negative skewness and positive kurtosis of 40 the hedge fund category’s return distribu- HFRI Macro: Systematic Div. tion, indicating there are fatter tails for 30 such strategies. Of the major categories, S&P 500 only managed futures strategies featured 20 positive skewness. This suggests that managed futures is one of the few asset classes 10 that mitigate left tail risk effectively. Return (%) 0

One remarkable feature of managed futures –10 strategies is consistent performance over time. These strategies in isolation typically –20 run with volatility levels comparable to traditional equities—and experience year-to- year drawdowns in accordance with their Source: FactSet standard deviation—but the asset class exhibits steadier results when viewed on a some structural changes in financial mar- June 2012 illustrate this phenomenon: longer-term rolling basis. Observation of kets and the CTA industry could be Following several decades of consistently three-year rolling performance, for exam- grounds for a more persistent degradation low correlations among assets, correlations ple, illustrates the relative ability of CTA in alpha generation. Differentiating moved sharply higher in the fall of 2008 strategies to consistently generate positive between a cyclical trough in performance and have yet to recede (see figure 5). This returns (see figure 4). In fact, November and a secular trend toward lower returns is, provided an extremely difficult environ- 2012 was the only month in the history of of course, the key issue to examine. ment for systematic and discretionary trad- the HFRI index that the series posted a ers alike, but the result is particularly pro- negative three-year return. In contrast, pure Analysis of Recent Performance nounced for managed futures strategies. long-only equity markets and long-only The cause of recent underperformance in The inability to establish independent, commodity markets (not shown) experi- managed futures managers is a source of uncorrelated trades reduced the efficacy of enced prolonged periods of negative intense debate. Whether due to central such approaches, creating return profiles performance. bank interventions, increased assets in the where single-asset drawdowns caused by space, or simply a new regime in investor trend reversals tended to cluster. The ten- Despite the longer-term consistency, risk behavior, it is certain that asset class dency for investors to collectively trade in returns at present do have the appearance correlations spiked dramatically in the risk on/risk off fashion during the eco- of a slow overall trend downward. Some post-global financial crisis period. Data nomic recovery period—in which groups might argue that we are simply at a periodic from AQR Capital Management for three- of asset classes move together in rapid fash- performance trough, similar to the one wit- year average absolute pairwise correlation ion—occurred with such frequency that nessed in late 2002. On the other hand, across 59 markets during January 1903– alpha generation by CTAs eroded.

JULY / AUGUST 2014 39

We discuss a few of the main causes of CTA underperformance in more detail below. Figure 5: Average Pairwise Asset Correlations, 1903–2012

Central Bank Intervention and Market 0.60 Artificiality 0.50 Perhaps the most compelling theory for underperformance by managed futures 0.40 funds is the unprecedented level of central bank intervention that occurred in the 0.30 wake of the financial crisis. In 2012 alone, rr elation Co central banks changed interest-rate policy 0.20 158 times1 and implemented nonconven- 0.10 tional measures such as asset purchasing programs and direct currency interven- 0.00 tions, which often have far more disruptive 1903 1913 1923 1933 1943 1953 1963 1973 1983 1993 2003 2012 impact than rate policy alone. Source: AQR Capital Management Of course, central bank activity always was present in the market. The scope and pace Figure 6: Q2 2012 Cumulative Return of such events, however, increased dramatically as central bankers took on a more activist role. Compounding this phenome- 4.00 HFRX Macro: Systematic Div. non is the rise in influence of emerging 3.00 market central banks; global financial mar- 6/5/12—Draghi Comments (–1.7%) kets now are subject to the competing eco- 2.00 nomic interests of a much wider swathe of sovereigns compared to 20 years ago. 1.00 Return Return (%) 0.00 Such intervention presumably disrupted the natural trajectory of many markets. In –1.00 6/29/12—EU Summit Decision (–2.8%) instances where trends developed—and –2.00 where managed futures’ quantitative models start to initiate positions—central banks or policy makers intervened and the markets reversed. This created a start-stop Source: FactSet arena pitting political will versus true economic fundamentals and natural mar- daily HFRX Macro: Systematic Diversified These circumstances are anecdotal, but they ket trends. Index) lost 1.7 percent in a single day. underscore a broader phenomenon charac- terizing the economic recovery. Managed The events of June 2012 were particularly A similar event unfolded on the final day of futures funds are designed to capture large illustrative of this phenomenon. the second quarter of 2012. Systematic portions of a market move, not to identify trend-followers slowly crawled back to a inflection points. Increased frequency of Through June 5, 2012, managed futures positive 2.1 percent quarter-to-date return such inflections whipsawed intermediate strategies had generated a 3.6-percent through June 28. On June 29, though, a trend-followers and caused a protracted return quarter-to-date while the S&P 500 European Union summit yielded provisions period of underperformance. had lost approximately 8.4 percent. The that directly recapitalized banks in Spain next day, however, European Central Bank and Ireland and expanded the powers of Asset Growth President Mario Draghi announced at a the European Stability Mechanism. This Another popular theory for managed press conference that the central bank was catalyzed a sharp reversal in risk markets, futures funds’ struggles is the tremendous “ready to act” in response to deteriorating once again disrupting market trends (see asset growth in the industry. As previously market conditions. This sparked a sharp figure 6). The category lost 2.8 percent on mentioned, assets in systematic diversified rally in financial markets, working against June 29, and many long-tenured managed managers (as measured by Hedge Fund many positions held by managed futures futures managers experienced their worst Research) more than doubled between strategies. The group (as measured by the one-day loss ever. 2007 and 2011. Because the flood of assets

40 INVESTMENTS&WEALTH MONITOR

from investors coincided with poor performance, many inferred a causal relationship. Figure 7: Global Futures and Options Volume by Region As a general rule, the potential for alpha 30,000 generation erodes in the face of increased assets and strained capacity, holding all other factors constant. 25,000

It is true that CTA assets swelled rapidly, but 20,000 this coincided with an increase in both the number and volume of futures contracts. Between 2003 and 2012, global volume of 15,000 futures and options contracts increased 161 percent, from 8.1 trillion to 21.2 trillion con- 10,000 tracts (see figure 7), according to the Futures In Billions of Contracts Industry Association. Estimates of futures and options volume since 1998 suggest a 5,000 more dramatic expansion (in the range of 10x). And thanks to the rise in electronic 0 trading, futures markets remain some of the 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 most liquid markets in the world; average bid-ask spreads steadily declined since 2000 North America Europe Asia-Pacific Latin America Other because of the automation of transactions. Source: Futures Industry Association

At first glance, managed futures assets as a Figure 8: Three-Month Treasury Calendar-Year Returns share of global futures markets do appear to be quite trivial. Hurst et al. (2012) estimates that CTA assets currently make up 5 percent 9 of commodities markets, 0.2 percent of 8 BofA ML US T-Bills 3M equity markets, 2.6 percent of bond markets, 7 and 0.2 percent of currency markets. Those 6 figures are not inconsequential but generally 5 reveal an industry that is not dominating the futures markets. 4 Return Return (%) 3 These numbers, however, may not fully illus- 2 trate the scope of CTA influence. Managed 1 futures assets constitute a small percentage 0 of overall market volume, but on certain key trading days CTA flows can dominate and impact market pricing. Execution skid and Source: FactSet piling-on effects by brokers who see these flows certainly creates short-term serial Unfortunately, it is difficult to generate a managed futures funds’ absolute returns. autocorrelation of pricing, as one manager’s cohesive conclusion on this topic; the dis- Because futures contracts require only a model trips another and a wide variety of tinction between correlation and causation fraction of the overall notional value of the CTAs seek liquidity in the same market of recent poor performance is blurred. contract to be posted (known as margin), direction at approximately the same time. In What does appear more certain is that, at typically 75–85 percent of a managed futures ways, this issue is similar (although of a the very least, CTAs had to become more fund’s assets sit in short-term government much smaller scale) to what happened with sophisticated in their execution to mini- bonds or money market securities. Between equity quantitative models in August 2007— mize market impacts and to avoid detri- 1990 and 2008, the average annual return on where the downsizing of leverage by one mental front running by other parties. three-month Treasury bills was 4.4 percent; quantitative model tripped other models since 2008, however, that return dropped to into downsizing and eventually created Collateral just 0.1 percent per annum (see figure 8). short-term irrational movement in various An often overlooked factor on CTA perfor- The current interest-rate regime imposed by equity pairings. mance is the impact of low interest rates on the Federal Reserve eliminated this comple-

JULY / AUGUST 2014 41

mentary source of return, which often cush- ioned managed futures funds through peri- Figure 9: Risk On/Risk Off Index odic stretches of underperformance. 0.50 0.45 One final point on recent performance is that although the scope of managed futures 0.40 strategies’ recent problems appears unprece- 0.35 dented, new historical research suggests 0.30 otherwise. Hurst et al. (2012) extends the 0.25 common research period of managed 0.20 futures from the 1980s onward back to 0.15 1903. The authors reconstructed a time 0.10 series of common trend-following models 1990 1992 1994 1996 1998 2000 2004 2006 2008 2010 2012 based on 59 markets across four major asset HSBC RORO Index classes (commodities, equities, bonds, and currencies). Although such a theoretical Source: HSBC construction should be viewed with some level of skepticism, the model does provide the European debt crisis erupted in early markets are loosening up as a result, lead- a proxy for examining longer-term perfor- 2010. Any decline in this phenomenon ing to more dispersion within asset mance of simple trend-following systems. would be a welcome development for tradi- classes—particularly in currencies and tional CTA trend-followers because they commodities. These developments are posi- Hurst et al. (2012) determined that the rely on independent, uncorrelated assets to tive for both traditional asset managers as drawdown of 13.5 percent (as measured by generate successful portfolio positions. well as systematic trend-followers. their model) between February 2009 and July 2011 is the sixth worst peak-to-trough There is reason to believe that a reduction in Downside Performance decline in the past 110 years. In the 10 risk-on/risk-off trading is indeed occurring. From an historical perspective, arguably worst instances during the measurement Analysis by HSBC, which utilizes principal the most important property of managed period, the average drawdown was 14.8 component analysis on 34 major assets to futures is its negative correlation to risk percent and the average peak-to-recovery create a measure of market correlation, indi- assets during periods of drawdowns or risk length was 25 months. This places the cates the tendency for a broad group of aversion (although this profile could evolve recent move right around the mid-point of assets to be influenced by a single primary for reasons we discuss in the next section). those historical events. factor is abating. The firm’s aptly named Risk Schneeweis et al. (1996), for example, On/Risk Off Index neared a 0.50 level in late examined the downside performance char- The Case for Managed Futures 2012 but has since fallen to under 0.35 (see acteristics of managed futures strategies The critical question for investors at this figure 9). Although still elevated compared and concluded that CTA strategies do, juncture is whether managed futures strate- to the levels observed between 1990 and indeed, offer uniquely positive performance gies still make sense for their portfolios. 2006, conditions are substantially improved. in such environments. An analysis of nega- Performance during the past four years tive months for the S&P 500 since February shook the confidence of many investors in This phenomenon is supported by a num- 1991 provides additional context to this this space, with some going so far as to pro- ber of additional sources. Anecdotally, a assertion: managed futures were the only claim the CTA model broken. A number of broad range of the asset managers that major asset class to exhibit negative correla- items are worth discussing that influence Fortigent Research interacts with pointed to tion to equity markets (see table 2). the answer to that question. lower correlations between asset classes and between individual securities. This assertion This performance profile is critical for Risk-On/Risk-Off Trading is backed up by hard data indicating that investors to remember. The past four years The first, and most important, issue at hand average pair-wise correlations among the of performance were disappointing, but is the potential decline in risk-on/risk-off predominant asset classes shrunk by more this period is marked by a near constant trading dominating financial markets since than 16 percent since 2011.2 Many financial climb in equity markets. A decline in equity

Table 2: Downside Correlation to S&P 500, February 1991–March 2013 HFRI Macro: Barclays US MSCI Russell DJ UBS Dow Jones US HFRI Fund Weighted Systematic Diversified Aggregate EAFE 2000 Commodity Index Select REIT Composite –0.16 0.00 0.72 0.68 0.31 0.52 0.62 Source: FactSet

42 INVESTMENTS&WEALTH MONITOR

markets—particularly a sustained decline— to managed futures to this mix resulted in a Many other market participants with their would provide an environment in which modest reduction of Sharpe ratio to 1.20 own objectives and preferences provide managed futures strategies potentially during the four-plus-year time period. opportunity for skilled investors to profit. could differentiate themselves from other Although return is modestly reduced, a Research by Keynes (1930) famously pre- asset classes. comparable decline in volatility occurred dicted that a risk premium exists in futures due to the uncorrelated nature of systematic markets because of the presence of natural Portfolio Diversification trend-following. Also noteworthy is the hedgers (both producers and consumers) The most critical evaluation of the managed substantial improvement in maximum who trade with non-economic motivations futures industry surrounds its diversification drawdown during that period. The simple (i.e., to control risk). Managed futures properties. Much of the academic literature exercise suggests that, in a mean-variance funds’ roles as liquidity providers in the regarding the space makes particular note of framework, an investor was only slightly market allows them to exploit this pre- these strategies’ lack of correlation to tradi- worse off despite enduring the worst perfor- mium, providing a fundamental basis for tional assets and the improved risk-return mance stretch in the history of the managed why these types of strategies can generate profile of portfolios that allocate to managed futures industry. One could argue this was positive absolute returns over time. futures. The evidence suggests that in the not a substantial opportunity cost to bear Hedgers certainly have not abandoned the recent years marking managed futures diffi- for the continued downside benefits that futures marketplace, leaving this important culties, their lack of correlation to other asset managed futures offer. theoretical source of CTA return intact. classes remains intact (see table 3). Futures Market Risk Premium The Big Caveat: A Rising-Rate Because this noncorrelation persisted, the Finally, an important, fundamental ques- Environment portfolio diversification benefits of includ- tion when evaluating CTAs is: What is the The counterpoint to the above discussion on ing managed futures did not suffer a mate- source of their return and has that source risk reduction and diversification is how rial degradation (see table 4). For example, a dissipated? Many academic studies have managed futures strategies will behave in a traditional 60/40 portfolio of the S&P 500 investigated the persistence of alpha gener- rising-rate market environment. Much of and Barclays Aggregate Bond indexes since ation of CTAs within futures markets. One the historical research on rising interest rates the beginning of 2009 generated 12.7 per- common and important take away from indicates that both traditional equities and cent annualized return with a 9.9 percent that work is the fact that CTAs are not fixed-income securities exhibit difficulty in annualized standard deviation, or a Sharpe solely trading against one another within such regimes. Those results are well-docu- ratio of 1.27. Adding a 20-percent allocation the futures markets (i.e., a zero-sum game). mented, but less research exists on the

Table 3: Correlation Statistics, 2009–March 2013 HFRI HFRI Macro: Fund DJ US Correlation Syst. S&P Russell MSCI MSCI BarCap Wtd DJ UBS Select 2009–3/2013 Div. 500 2000 EAFE EM Agg Comp. Commodity REIT HFRI Macro: Systematic Div. 1.00 0.12 0.02 0.16 0.11 0.19 0.31 0.33 0.09 S&P 500 0.12 1.00 0.95 0.91 0.83 –0.07 0.83 0.69 0.81 Russell 2000 0.02 0.95 1.00 0.84 0.80 –0.18 0.80 0.60 0.83 MSCI EAFE 0.16 0.91 0.84 1.00 0.89 –0.04 0.89 0.74 0.76 MSCI Emerging Markets 0.11 0.83 0.80 0.89 1.00 0.00 0.90 0.75 0.67 BarCap Agg 0.19 –0.07 –0.18 –0.04 0.00 1.00 –0.15 –0.03 0.11 HFRI Fund Wtd Composite 0.31 0.83 0.80 0.89 0.90 –0.15 1.00 0.76 0.64 DJ UBS Commodity 0.33 0.69 0.60 0.74 0.75 –0.03 0.76 1.00 0.46 DJ US Select REIT 0.09 0.81 0.83 0.76 0.67 0.11 0.64 0.46 1.00 Source: FactSet

Table 4: Risk/Return Statistics, 2009‒March 2013 Annualized Annualized Portfolio Return St. Dev. Sharpe Ratio Max Drawdown 60/40 S&P 500/BarCap Agg 12.7 9.9 1.27 –11.5 60/40 S&P 500/BarCap Agg w/ 20% Managed Futures* 10.1 8.3 1.20 –9.5 *HFRI Macro: Systematic Diversified Source: FactSet

JULY / AUGUST 2014 43

performance of managed futures in such periods as interest rates generally declined Figure 10: Treasury Price Curve, as of August 13, 2013 since CTAs gained prominence 30 years ago. 100 Theoretically, one would be led to believe that CTAs still can be effective in such envi- 99 ronments. Studies such as those by 98 Campbell & Co. (2013), which evaluated the results of a simple trend-following sys- 97

tem through a variety of interest rate cli- Price ($) mates, affirmed as much: “CTA perfor- 96 mance in relation to the direction of rates suggests that the strategy has not histori- 95 cally been rate-regime dependent.” The rationale for this thesis is fairly intuitive: 94 2YR 3YR 5YR 7YR 10YR 30YR

1. Managed futures strategies are designed Source: Bloomberg to capture trends both positive and negative. Should a persistent increase in Impact of Negative Carry and Roll Yield Ultimately, offsetting impacts will be at interest rates occur, trend-following Because yield curves were positively sloped work in a structurally rising fixed-income strategies should be able to capitalize. (i.e., a negatively sloped or backwardated environment. Managed futures funds will 2. Because the majority of actual portfolio futures price curve) during most of the benefit from higher collateral returns and assets are held as collateral, increases in evaluation period, CTAs also benefited trend-following bond spot prices lower (if cash rates should provide a higher return from the embedded positive carry roll there is a sustained move). However, a on that component of the portfolio. effects of futures prices slipping down the decline in diversification characteristics yield curve across the passage of time (see and the negative carry associated with In reality, however, investors have a few figure 10). In a positively sloped yield- being short a backwardated market will be other important considerations: curve environment, however, a CTA effec- a new headwind for the industry. This is tively must “pay the points” embedded in not an immaterial development and should Correlation Impact of Short Bond the futures market rolldown cost to remain be considered when investing in CTAs. On Positioning short. A CTA might for example benefit balance, a new regime in fixed income Because CTAs generally trend-followed from a 10-percent decline in a given bond, should adversely impact managed futures bonds higher over the past three decades, but if it cost 3 percent to carry that short in returns and diversification benefits. CTAs may have developed a portion of embedded carry/futures roll costs, then the their historic alpha-producing negative net return becomes only 7 percent. To the Proposed Path Forward correlation attributes to equity markets due upside of fixed income, this roll cost would Given the possibility for a structural shift in simply to their long fixed-income position- have been additive (10% move higher + 3% CTA efficacy, how does one address these ing. Since 1990, U.S. fixed income (as mea- embedded carry = 13% total return) issues within portfolios? Fortigent has a few sured by the Barclays Aggregate Bond instead of negative. immediate suggestions. Index) exhibited a negative daily correlation to the S&P 500. Long bond positioning In the future, the pace of any bond decline First, do not panic and completely eliminate of managed futures funds entering equity obviously would be important (fast moves CTAs from your portfolio. The above issues downdrafts proved pivotal in mitigating the overpowering negative carry aspects) and may mean that CTAs in general are facing a initial negative impact of long equity expo- the shape of the yield curve obviously more difficult return-generating environ- sure. In the future, a similar scenario where would be important (inverted yield curves ment, but they do not eliminate all of the CTA bond exposure is flat or even short (a would make trend-following far easier to diversification benefits of CTAs. As long as likely phenomenon in a rising-rate envi- the downside of fixed income). But for general market correlations loosen up some- ronment), losses would not only be fully now, with a steep yield curve, CTA trend- what—as they appear to be doing at pres- felt but also could be exacerbated as a flight following to the downside of fixed income ent—CTAs should start to perform a bit bet- to safety trade sends bond prices higher. will not be as easy as it was to the upside ter. One of the key attributes of CTAs is their The historic downside performance and of fixed income. This is a mathematical ability to generate profit in a variety of asset diversification benefits of CTAs, therefore, fact until the time that yield curves classes, not just fixed income; the aforemen- may be structurally eroded in a new interest- invert—an ever so remote possibility at tioned increase in dispersion of individual rate regime. present. commodities, currencies, and equity markets

44 INVESTMENTS&WEALTH MONITOR

still should offer fertile ground for alpha capture and returns independent of other tradi- Figure 11: Rolling One-Year Return—Macro vs. CTA tional portfolio allocations. 60 CS Managed Futures Index However, in the selection of a CTA, it may 50 CS Global Macro Index be increasingly important to choose man- 40 agers with less embedded fixed-income 30 exposure and more exposure to other assets 20 (soft commodities, metals, energy, etc.). At the very least, investors should seek manag- 10 ers with an appreciation for the above Return (%) 0 fixed-income issues and a proposed model- –10 ing solution to it still either via trade time –20 horizon, sizing, or other methods. –30 Fortunately, many managed futures firms are taking it upon themselves to adapt to Source: FactSet the changes occurring in financial markets. The universe long has been dominated by intermediate-term trend-followers, but models rely on different systems to generate English wheat prices going back to 1209” asset managers are spending considerable buy and sell signals and also can rely on non- (Wilkes and Fletcher 2012). time and effort in developing new alterna- price data. For example, some strategies tives. This may include different time peri- quantify fundamental information to identify Non-futures contracts: One other develop- ods of evaluation (shorter or longer term), potential trading opportunities. Counter- ment within the industry is the use of new the addition of new contracts or nontradi- trend models rely on price data to predict instruments within CTA portfolios. For tional data sets, or the establishment of market trend reversals. Other systems aim to example, prominent managers have begun so-called non-trend strategies. These identify similar historical instances of price trading cash securities such as individual efforts are designed to increase the efficacy movement or fundamental data to select the stocks and bonds, other derivatives such as of systematic strategies and to provide most optimal trade for that environment (i.e., credit default swaps, as well as more eso- diversification to traditional intermediate pattern recognition). The algorithms for such teric instruments such as energy power trend-following. We discuss some of those models and methodologies can become contracts. The risk with such trades is that initiatives below. exponentially complex as researchers refine they are often less liquid, may have higher and add to their techniques. transaction costs, and can require higher Short-term trading: Shorter-term trend-fol- capital commitments than traditional lowing strategies are gaining greater promi- Nontraditional data sets: A so-called arms futures and forwards. nence because of their success in the recent race has emerged among the largest CTAs, market environment (perhaps because of the in which teams of PhDs and highly trained Additionally, Fortigent believes that within shorter half-life of trends due to central bank academics are scouring nonstandardized the CTA space it may be appropriate at this activity). Increased computing power, the data sets to discover new anomalies to time to include more discretionary manag- rise in high-frequency trading capabilities, exploit. Indeed, nontraditional data is a hot ers who potentially are better adapted to and greater accumulation of intra-day data new avenue for managers to differentiate handle the current shifting political winds are contributing to the growth of these strat- themselves. Analyzing data that tradition- and artificial interest-rate environment cre- egies. These systems may follow data as ally have not been used in academia or in ated by central bankers. Macro managers short as minutes or hours out to a few financial studies can produce new opportu- only performed marginally better than months, whereas most intermediate trend- nities for profit. This may include weather CTAs over recent years (see figure 11), but followers trade within a 6–12 month time forecasts, social media trends, or historical they are likely the ones who would first frame. Many prominent managed futures data that have not yet been databased. benefit from any initial cracks developing managers today implement some combina- London-based CTA Winton, for example, in global confidence in central bankers. tion of shorter- and longer-term models to the largest in the world with more than produce more robust signals that succeed in $20 billion in assets, reportedly “sends We argue this because discretionary man- different market cycles. researchers to libraries and archives across agers are more likely to recognize the the world to find numbers held in books unsustainable aspects of certain global Non-trend strategies: The proliferation of and on microfilms. It has found barley and fixed-income markets and may be first in non-trend strategies has been swift. Such sesame prices from ancient Babylon, and line to profit from the initial reversal in

JULY / AUGUST 2014 45

stretched markets. Trend-following CTAs, As always, ongoing manager due diligence is References by comparison, will be slower to react to paramount to navigate this new and more Campbell & Co. 2013. Prospects for CTAs in a Rising Interest Rate Environment. Campbell White Paper any such entropic shift. The potential for complex world. The universe of discretionary Series (January). http://www.campbell.com/_files/ discretionary managers to capitalize on and non-trend oriented managers is much Prospects%20for%20CTAs%20in%20a%20 Rising%20Interest%20Rate%20Environment.pdf. major market inflection points—either in more heterogeneous than that of intermedi- Chen, Peng, Christopher O’Neill, and Kevin Zhu. 2005. fixed income, equities, or otherwise— ate trend-followers, presenting more poten- Managed Futures and Asset Allocation. Ibbotson Associates. https://corporate.morningstar.com/us/ offers a complementary exposure to tradi- tial pitfalls for investors. Careful evaluation documents/MethodologyDocuments/IBBAssociates/ tional trend-followers. Investors should of individual strategies, and understanding ManagedFutures.pdf. Edwards, Franklin R., and James M. Park. 1996. Do consider layering in more discretionary- how they potentially complement an existing Managed Futures Make Good Investments? Journal of oriented solutions to add robustness to portfolio, remains critical. Futures Markets 16, no. 5 (August): 475–517. Hurst, Brian, Yao Hua Ooi, and Lasse H. Pedersen. 2012. CTA allocations. A Century of Evidence on Trend-Following Investing. Ryan Davis, CAIA, is a senior analyst in the Working paper, AQR Capital Management. http://www. aqr.com/DesktopModules/AQR.FileView/FileDownload. Conclusion alternative investments group at Fortigent, ashx?fileID=266. In our opinion, managed futures strategies LLC. He earned a BS in economics from Keynes, John M. 1930. Treatise on Money. London: Macmillan. continue to play an important role in inves- George Washington University. Contact him Lintner, John. 1983. The Potential Role of Managed tor portfolios. Although alpha generation at [email protected]. Commodity-Financial Futures Accounts (and/or Funds) in Portfolios of Stocks and Bonds. Working paper, has waned in recent years, we believe this Harvard University. http://books.google.com/books/ is primarily due to cyclical factors in the Barclay Leib is vice president of alternative about/The_potential_role_of_managed_commodity. html?id=kqB6GwAACAAJ. marketplace. Preliminary evidence investments at Fortigent, LLC. He earned McCarthy, David, Thomas Schneeweis, and Richard indicates that this adverse environment an AB from Princeton University’s Woodrow B Spurgin. 1996. Investment through CTAs: An Alternative Managed Futures Investment. Journal of may be dissipating. Wilson School of International Relations and Derivatives 3, no. 4 (summer): 36–47. Public Affairs with a concentration in econom- Schneeweis, T., R. Spurgin, and M. Potter. 1996. Managed Futures and Hedge Fund Investment for More importantly, managed futures ics. Contact him [email protected]. Downside Equity Risk Management. Derivatives strategies continue to provide diversifica- Quarterly 3, no. 1 (fall): 62–72. Wilkes, Tommy, and Laurence Fletcher. 2012. The tion benefits for diversified portfolios. Endnotes Algorithmic Arms Race. Reuters (May 21). http://www. Correlation levels for CTAs are signifi- 1. This information obtained in response to an e-mailed reuters.com/article/2012/05/21/us-trading-blackbox- query to centralbanknews.info, June 2013. idUSBRE84K07320120521. cantly below those seen in other traditional 2. Trailing one-year daily correlations, March 2013 vs. hedge fund categories and provide a posi- December 2011 for S&P 500, MSCI EAFE, MSCI EM, Russell 2000, Dow Jones–UBS Commodity, DJ US tively skewed performance profile not evi- Select REIT, HFRX Absolute Return, Barcap Agg, and dent in other strategies. These diversifica- Alerian MLP. tion benefits made up for performance deterioration of the asset class since 2009, resulting in no material decline in risk- adjusted performance for diversified portfolios. This suggests the opportunity cost of holding managed futures allocations is far from unbearable.

With that said, the impending shift in interest-rate regime does present a potential structural issue for CTA performance. One may want to consider managers with less fixed income-centric programs because the Are You LinkedIn? correlation benefits of trend-following that particular asset class could become more NETWORK WITH YOUR PEERS THROUGH IMCA’S GROUP. problematic. Favoring programs with diversifying non-trend exposures and different portfolio time horizons should help Join online at www.linkedin.com/e/gis/1620107 mitigate this issue. For the immediate • Business and Social Networking • Discussions • Jobs future, we also would favor manager pools • News and Article Updates • LinkedIn member directory with more discretionary influences because they may be better equipped to capitalize ® on the dislocations caused by overly IMCA accommodative central banks.

46 INVESTMENTS&WEALTH MONITOR