Most trend followers are systematic traders: their trading decisions are driven off a rule-based, quantitative model they have developed to produce an edge by identifying and exploiting momentum in the markets.
Systematic trading can be a very rewarding way to invest in the markets. For example Renaissance Technologies’ Medallion Fund, often described as one of the most successful hedge funds, has produced over 30% annual returns since inception in 1988. It is thought that trend following played a significant part of the early success of the Renaissance trading strategies. However, many less well-known systematic managers – Commodity Trading Advisors (CTAs) – have been successful for decades using the trend following strategy.
Commodity Trading Advisors evolve from a misnomer
The CTA designation is a legacy term dating from the 1970s. For many years the only futures markets were on commodity markets – before the development of financial futures. So it was natural that any investment advisors using futures were registered with the Commodities Futures Trading Commission (CFTC). This regulatory anomaly continued after futures on bonds and equity indices were developed. Most of the diversified CTAs/trend followers now trade global diversified portfolios of futures, representing products in asset classes other than commodities such as interest rates, currencies or equities.
Not all CTAs are trend followers, but the strategy is so prevalent amongst CTAs that the terms are usually used interchangeably. The nomenclature also includes the terms Managed Futures and Commodity Pool Operators (CPO). Here we refer to this group of alternative investment managers as CTAs or trend followers.
Trend following, CTAs’ main strategy, has been used for centuries, with the main motto for “staying with the trend” (“Cut short your losses, let your profits run on”) being attributed to 18th century British trader and economist David Ricardo.
It was not until the 1950s that the strategy was formulated as a more complete investment philosophy by Richard Donchian, often dubbed the “father of trend following”, and considered as the creator of the managed futures industry. Donchian started the first publicly managed futures fund, Futures, Inc., in 1949.
Donchian’s strong influence on the managed futures industry early days triggered a large number of CTAs to follow in his footsteps. One such manager is “Market Wizard” and ex-Commodities Corp trader Ed Seykota. Inspired by one of Richard Donchian’s newsletters describing a moving average trend following system, Seykota pioneered computerised trading to test Donchian trading ideas on early punch card computers, in 1970. He went on to manage client money very successfully (reportedly growing a client model account from $5,000 to $15,000,000 in 12 years).
Another early influence in the world of CTAs is “Prince of the Pits”, Richard Dennis. A successful trader, Dennis, together with Bill Eckhardt, set up an experiment, later known as the Turtle Trading programme. Dennis, arguing that trading could be taught, hired a dozen applicants from diverse backgrounds and taught them a trend following system so that they could trade for him. The experiment was a success and spawned off many CTAs run by ex-Turtle Traders and still in activity today. Some of the most successful ones include Jerry Parker (Chesapeake Capital), Liz Cheval (EMC Capital Management), Paul Rabar (Rabar Market Research) or Tom Shanks (Hawksbill Capital).
Europe, and especially London, was also a strong centre for trend following. In 1983, then commodity broker ED & F Man partnered with Mint Investment Management, set up by Larry Hite, and moved into a successful financial services firm to become the Man Group. After purchasing AHL, another London-based CTA set up by Michael Adam, David Harding and Martin Lueck, Man AHL became one of the world’s largest CTAs. The three AHL founders went on to build other successful CTAs based out of London. Adam and Lueck started Aspect Capital, while Harding founded Winton Capital. Other large European CTAs include BlueCrest’s BlueTrend fund and Transtrend.
Initially using mostly “sophisticatedly simple systems”, a new trend of managers are now investing heavily in research and development, to stay ahead of the game, in what has become a more competitive market. According to BarclayHedge estimates, CTA total assets under management stood at a record $267 billion as of Q4 2010, having grown by more than 400% in the last seven years ($51 billion in 2002). As opposed to the example of long-running Dunn Capital Management reportedly still trading a very similar system to the one he has used since the 1980s, London-based David Harding, of Winton Capital, has an “R&D” team reportedly staffed by over 100 researchers, including PhDs, mathematicians, statisticians and even astro-physicists.
Trend following as a strategy
Trend following is a systematic global macro strategy designed to benefit from volatility and large “fat-tail” moves that exist in market returns distribution. Because of various behavioural biases impacting financial markets (herding and feedback trading, overreaction, confirmation bias), trends tend to persist longer and more often than expected by the Random Walk Hypothesis. CTAs aim to generate profits by capturing part of these trends.
The strategy focuses mostly on price action and is reactive, as opposed to more traditional, discretionary strategies, for which traders or managers attempt to predict the market by making a forecast. Trend followers initiate a long position when an instrument starts trending up, i.e., “Buy high” (with the aim of selling back higher), or a short position when the instrument starts trending down, i.e., “Sell low” (with the aim of buying back lower).
John W. Henry, owner of the Boston Red Sox and recent acquirer of Liverpool Football Club made his fortune as a CTA. He highlights the “Buy High, Sell Low” principle:
“For 19 years we have consistently bought high and sold low. If trends were not the underlying nature of markets, our type of trading would have very quickly put us out of business. It wouldn’t take 19 years or even 19 months of buying high and selling low ALL of the time to bankrupt you. But trends are an integral, underlying reality in life.”
Basic trend following systems use simple technical trading strategies such as moving average cross-overs or price breakout indicators in order to identify trend formations as can be seen in Fig.1.
A moving average cross-over trading strategy naturally follows the trend. It can capture part of large trends both on the upside and downside, as can be seen in this Crude Oil chart. Trendless periods such as 2010 generate small losses because of whipsawing price movements.
Another factor in trend following success lies in the risk management and asset allocation parts of the strategy. Most managers trade a large set of instruments (CTAs can trade upwards of 100 different instruments) in order to benefit from the diversification and reduced volatility this provides. This volatility reduction is useful to keep the overall volatility of the system under control, as the strategy itself can generate volatile results, due to its position management principle: “Cut short your losses, let your profits run on”.
“Cut short your losses” means that positions are exited swiftly if prices start moving unfavourably. This avoids large losses from any one trade but also translates in the majority of trades being losers. “Let your profits run on” gives room to stay with existing positions by only exiting them once the trend is clearly over. This results in winning trades far outsizing losing trades, making the strategy profitable on balance, but being negatively impacted during trend reversals, and subject to volatility during trend oscillations.
These two factors make the equity curve more volatile, explaining why very few long-running CTAs seem to be able to exceed Sharpe ratio values of 1.
Correlation and diversification
As an alternative investment, CTAs offer a diversified source of returns, usually uncorrelated to traditional investments or even hedge funds. The year of the last equity market crash, 2008, being the clearest recent example of this relationship. Fig.2 illustrates how CTAs and hedge funds fared against the S&P 500 index during the harsh market conditions of 2008. Whereas hedge funds display a strong relationship with the equity market – the correlation between the S&P 500 and Barclay Hedge Fund monthly returns being equal to 0.84 in 2008 – the CTAs’ performance is much more decoupled from that of the S&P 500 index with a correlation of -0.17 (and -0.39 against the hedge fund index).
From a portfolio management point of view, under Modern Portfolio Theory, adding uncorrelated investments is an efficient way to reduce volatility (often equated to risk) and/or to increase returns. CTAs can play a significant role in this area.
This is illustrated in Fig.3, which displays a chart of both S&P 500 Total Return Equity Index and Au.Tra.Sy Trend Follower Index (see box) since 1990, as well as a third equity curve representing a hypothetical portfolio including both components split on a 50/50 monthly rebalancing basis.
As can be seen in Table 1, the trend follower index is more volatile than the S&P 500 but also generates higher returns. An equal-allocation portfolio markedly improves the performance of the equity index both increasing return and decreasing volatility.
Correlation within CTAs and differentiation strategies
CTAs, as a group who follow similar strategies, often have correlated results to each other. Considering all individual components of the Au.Tra.Sy Trend Follower Index, the average pairwise correlation between each CTA and the index is 0.79 with values ranging from 0.67 to 0.89 as illustrated by Fig.4. Note that the correlation is calculated on monthly returns from 1990 to 2010.
CTAs mainly target High Net Worth Individuals (HNWI) and institutional investors. Given the relatively high (6 to 8-figure) minimum account size to invest with a CTA, it might not be possible for a HNWI to diversify across a large range of different CTAs. However, the fact that their returns are closely correlated would indicate that picking just one or a few CTAs should ensure a performance close to that of the index.
On the other hand, institutional investors or fund allocators most certainly aim at diversifying within the CTA space in order to benefit from increased risk-adjusted returns. But the positive effects of diversification substantially drop off as the inter-correlation between the individual components of a portfolio increase towards 1. And as we have seen earlier, correlation runs fairly high amongst CTAs.
However, correlations are not stationary and can evolve dramatically through time, as was experienced during the credit crisis of 2008, with large correlation spikes in various financial markets. Instead of looking at correlations over the whole historical performance, studying correlation over shorter time spans should give better insights into the dynamics of how CTA correlation evolves.
The next chart (see Fig.5) is based on rolling 12-month correlations between each CTA and the index: every month the correlation is calculated using the previous 12 monthly returns. The monthly average across all CTAs is plotted on Fig.5 and gives an overview of the CTA correlation changes through time – or rather non-change as the value appears fairly stable.
The distribution of all rolling 12-month correlation values is plotted on Fig.6. Over half of these correlations are very strong (median value is 0.83), with less than 10% registering a value below 0.6.
The main reason why the return series are so similar is due to the typical asymmetrical trade returns distribution: many small losses and few large winners create a positive skew in the returns distribution. As trend followers do not usually attempt to predict the magnitude of trends, they often stress that they cannot afford to skip a position, as any trade has the potential to turn into a large winner and provide a large share of their returns for a given period.
This last point highlights the fact that CTA results can be strongly influenced by a few trades only, with the bulk of their performance being driven by these large outlier trends.
With the critical importance of diversification to the strategy, there is a large overlap of instruments traded by most CTAs. As a result, it seems that CTAs mostly end up making money in the same trends, which they have to follow in order to ensure their profitability, thus creating these strongly coupled return streams.
These tight correlations have implications for building portfolio of funds (diversified funds of funds or those dedicated to CTAs). From a portfolio management construction point of view, one of the most desirable attributes of a new addition is a component with negative or low correlation to the current portfolio, and positive returns. Adding another strongly correlated CTA to a portfolio already containing CTA components will not provide much improvement.
Simple simulations can help quantify that point. Consider a fund allocator already holding a few CTAs, producing a composite performance similar to that of the broader index (average monthly return of 1.53%, with a standard deviation of 5.79%). Adding to the portfolio a fifth or sixth CTA, with identical return statistics and a strong correlation of 0.8, would leave the average return of the portfolio unchanged and only slightly reduce the volatility (monthly standard deviation of returns decreasing to 5.49%). Alternatively, a manager with a correlation of 0.6 would decrease volatility by twice as much (monthly standard deviation of returns decreasing to 5.19%), with near-identical average return.
In order to become more attractive to fund allocators and investors, CTAs attempt to differentiate themselves from the pack, and generate less correlated results with the use of different timeframes, sector allocations, leverage, or combination of different strategies. One example is London-based CTA Beach Horizon, who attempt to offer diversification within the CTA space with an original portfolio construction:
“One of our great differentiators is that we have got quite a large weight in commodities, typically averaging 60%, which is a lot higher than some of our CTA peers. That involves trading things like palladium, platinum, lumber, oats, pork bellies, milk, all sorts of weird little futures markets. We see ourselves as a provider of diversification.”
Venturing into more exotic markets is a way to seek extra diversification, but liquidity can often be an issue. This leaves smaller CTAs at an advantage on this front. One strategy employed by larger funds to expand their universe of trading instruments is to follow spreads (inter-market or calendar spreads), potentially providing more robust diversification – spreads being more likely to move independently of the market during correlation spikes. Typically more subject to arbitrage and mean-reversion strategies, spreads can also diverge for longer than expected on a historical basis, as experienced during the demises of Amaranth or LTCM, for example. An episode, one can speculate, which was surely beneficial to trend following CTAs.
What next for CTAs?
Despite some skeptics, CTAs have been around since the 1970s, generating positive returns with low correlation to other asset classes, including hedge funds, thereby providing a great source of diversification in portfolio construction for investors and fundsof funds alike.
The industry has been evolving, with a breed of CTAs spending a lot of resources on “R&D”, seeking to exploit new techniques and academic advances into mathematics or statistics and the markets. This has however not materialised into great differentiations amongst CTAs’ performance results,
This has led some critics to postulate that CTAs do not really generate alpha, but rather trend following-style beta, and as such do not justify their typical 2/20 fees. Studies have been published, showing that simple public domain trend following strategies are able to achieve a high degree of correlation to CTA/trend following indexes, with a large share of managers not generating significant alpha to these system-based benchmarks. In August 2010, Newedge published a study showing that the relationship between weekly returns of their CTA Index and their CTA Trend Follower Index is tight, with an overall correlation of 0.97 over the previous 10 years.
These studies go in the same direction as the “hedge fund replication” development (replication of specific strategy returns with factor replication), but using rule-based models instead. A possible evolution of the industry might be in the creation of more benchmark-like trend following funds/ETFs using these simple trend following systems, with a lower fee schedule, more similar to passive investment funds, and creating competition to CTAs in their current form.
Conquest Capital Group is a manager marketing such a fund. “Conquest MFS is designed to become the “index fund” of trend following; it is over 75% correlated to most CTA indices while outperforming on an absolute and risk adjusted basis. The goal is to provide pure trend following “beta”, which is an important allocation to any hedge fund portfolio”.
This approach is gathering pace.
Also trying to tap into this concept, and providing access to retail investors, the Cambria Global Tactical ETF (GTAA) was launched in Q4 2010. While having a different structure to CTAs (it is an actively managed “ETF of ETFs”), its core strategy uses long-term trend following principles to allocate funds between 50 to 100 underlying ETFs representing different global asset classes. Both the Conquest Capital and Cambria funds have fees/expense ratios of around 1%.
As the CTA sector is attracting more inflows in the aftermath of 2008, investors look forward to seeing how the industry landscape shapes itself in the future – and which category comes out ahead in the battle between “old school” trend followers, research-heavy CTAs and simpler replicator funds.
The Au.Tra.Sy Trend Follower Index is a list of well-established successful CTAs (“Trend Following Wizards”), many of which can be traced back to the period when Trend Following started becoming more popular in the 1970s/1980s.
The list of individual components is as follows:
Abraham Trading Diversified Program
Altis Partners Global Futures Portfolio
BlueTrend Fund Limited
Campbell & Company Trend Following Portfolio
Chesapeake Capital Diversified Program
Clarke Capital Millenium Program
Drury Capital Diversified Trend-Following Program
Dunn Capital World Monetary and Agriculture
Eckhardt Trading Standard Program
EMC Capital Classic Program
Hawksbill Capital Global Diversified Program
Hyman Beck & Co. Global Portfolio
JWH & Co. Financial & Metals Portfolio
Man AHL Diversified Futures Ltd
Millburn Ridgefield Diversified Program
Rabar Market Research Diversified Program
Saxon Investment Diversified Program
Superfund Q-AG
Tactical Investment Management Institutional Commodity Program
Transtrend DTP – Enhanced Risk (USD)
Winton Capital Diversified Program
Legal disclaimer: The performance figures are collected from CTA databases and not provided by the CTAs. No reliance should be taken as to their accuracy, and as a consequence the figures may not be in accordance with any CFTC / NFA performance reporting requirements
Jez Liberty is a London-based independent trader with a strong interest in managed futures and trend following. A keen blogger, he can be found on www.automated-trading-system.com where he publishes systematic trading research and CTA performance tracking.