Hedge Funds Review European Single Manager Awards 2016
Systematic managed future funds continue to attract equal amounts of praise and criticism. Lyxor Asset Management's Epsilon Global Trend strategy has attracted more of the former. The managed futures fund was launched originally in 1997 with a Ucits version added in 2011. Guillaume Jamet, the principal manager of the Epsilon programme, supervises the systematic medium- to long-term trend following the strategy's research programme.
Like many other trend followers, the investment style is focused on pure price momentum, both long and short, trading equities, rates, currencies and commodities globally, although commodities are excluded from the Ucits product, which also has a lower volatility cap of 10% compared with 15% in the strategy.
By trading more than 40 listed futures contracts, it is similar to many other managed futures funds, but there are subtle differences. Epsilon operates as a dedicated investment team within the structure of Lyxor Asset Management. This gives it access to the infrastructure of a global specialised asset manager. Jamet reports directly to Lyxor's chief investment officer, Nicolas Gaussel. Jamet also has a close working relationship with Lyxor's head of absolute return and solutions, Guillaume Lasserre.
The Epsilon team runs the model, trading process and system, managing its own research programme in collaboration with the broader research team of Lyxor. Sales, marketing, legal, compliance, middle and back office and IT support are handled by Lyxor, giving the Epsilon team the space to concentrate on research and making improvements to the programme. It also gives the fund the institutional strength that infrastructure investors look for during due diligence.
The Epsilon programme itself embeds a comprehensive risk management system, but an independent risk management group that does not report to Jamet also supervises the team and the fund.
Jamet oversees the day-to-day operations of the programme and leads research initiatives. He is the main creator of the current model, which was implemented since late September 2012. Like many in this area he has a doctorate in mathematics. His senior team hold a mixture of degrees in maths, accounting, computer science and business.
One of the ways Epsilon is different is in its evolution. Barep Asset Management first conceived the model in 1994. In 2009, the programme transferred to Lyxor, and Jamet joined the team in 2011 to help develop the third generation of the model, which was implemented in 2012.
It is this third generation that Jamet believes makes the programme more robust. The 2012 upgrade and evolution of the model were based on two significant decisions that were derived from research and the team's investment philosophy. The first was to stick to mid- to long-term trends. "This is where many CTA [commodity trading adviser] managers drifted away. Some went short term, started contrarian models, looked at more illiquid or the over-the-counter markets. We decided to stick to the mid- to long-term and only on liquid markets," explains Jamet.
Big models, big theory
Being a "math guy", he was naturally attracted to the big models, big theory. Studies have shown there have been trends in all sorts of markets for centuries. This led him to decide to stick to trend following.
Jamet was also conscious of the impact that changing the model would have on investors. A lot of technology and high-level research is needed for this kind of model. What investors look for are managers that are good at detecting the trends, since predicting markets is "very tricky".
"Diversification is important, but less obvious is taking into account the correlation on portfolio concentration and management. We put equal importance on these," says Jamet.
The programme has an average holding time of around 70 days with an observation window of one year. It trades only in listed futures. The Ucits version concentrates on interest rates, equity indexes and currencies globally, focusing on the most liquid contracts. The trade decisions are price-driven and based only on price data. Correlations are tracked at every stage of the process.
The relative size of the "bet" depends on correlation. Do that on a global portfolio scale and there will be correlations across different markets, which are less important than across the assets traded. Interesting diversification opportunities pop up and are "something worth looking at".
For example, in January and February this year, the programme had no equity exposure in the US and Europe. The model identified the start of a weak downside on equities and it was so strongly correlated to energy and currencies, according to the model, it was not worth the gain balanced by the risk already held. The signal was weak and the risk was similar to that already held in the portfolio, according to Jamet.
"When equity markets went down, we made money on Japanese bonds or the Mexican peso. That can be explained by correlation. This is something that happens frequently," he notes.
Another example is metals and equities; these hold the same risk. Diversification on currency, equities and commodities can "sometimes be completely misleading, riding the same macro underlying factors", he says. "That's the importance of correlation. It is deeply rooted. We look for this in each market feedback."
Epsilon does not target volatility but rather looks at global risk exposure – another way this model differs from others. "We don't do volatility targeting. We're not always 8% or 15%. When the model sees fewer trends, we navigate at a lower volatility. For instance, tracking since the beginning of the year, there has been less volatility than in some similar periods." Present markets, Jamet admits, are difficult ones to for a CTA to decide when to buy.
Research is vital
Research is a key component of any systematic strategy and, for Epsilon, it is another differentiator. While there are small incremental improvements to the model on a regular basis, ongoing research is vital. Any statistical metric on where a market is might help to improve the model. But mainly the research focuses on long-term topics. "Here again we are different. The investment philosophy is actually that we don't optimise a large mixture of signals. We want to have a statistical model of markets and then indicators are a natural consequence of that statistical model," explains Jamet.
This is more like a classical science approach, he says. "We find a statistical model is better to describe the reality of financial markets. It is something we can observe on the market. If the statistics are robust enough, we add them to the strategy."
Parameters in the model are updated on a daily basis. The same software is used for historical and live simulation as well as for real trading. Simulations must account for a realistic trading environment, including the actual trading house, execution lag, costs and other factors. The models are tested out-of-sample and live through paper trading before going into a validation phase.
Models are back-tested over multiple time windows and investment universes. Epsilon favours models with a small number of parameters but all of them must have a sound statistics, economical and behavioural justification before being added to the programme.
The last major intervention was the major upgrade of the model in 2012. "That's when we made the important investment decisions. We started putting parts of the risk budget in the new model. We were confident of what we were doing and, by 2013, we were using 100% of the new model. That was a big change in the life of the programme," he says.
With a complex systematic strategy, investors do want research but not major model changes. So this was a brave move by the team led by Jamet. "We knew we had a track record with the model and investors who bought into the fund expected the portfolio manager to be able to make model changes. That's what we did. The investment decisions of a systematic manager is about model changes," Jamet states.
Another area of interest for Epsilon is adding new contracts. Being systematic means the team wants a lot of market data before moving into new markets or contracts. It takes two to three years minimum to consider if a new contract should be added.
"We need historical data, liquidity. These are two very important things. We also don't want one market to be more important than another. Positions are always just a few percentages of the risk of the portfolio. We will look at whether access is easy, if it costs less. We'll also discuss exposure on a market. Unless we know the reasons why we need exposure to a new factor, we won't add it."
The portfolio has more than 40 eligible contracts targeted with around 32 open positions on average.
Jamet recommends that investors into the programme stay in for a minimum of three to five years, even though there is daily liquidity. This timeframe is likely to give investors the best returns over market cycles.
Looking to the future, Jamet expects the programme to attract more assets over the coming months. Ucits vehicles are particularly popular in continental Europe and many investors now recognise the merits of making allocations to systematic managed futures funds. Jamet believes the programme could easily work with $3 billion of assets under management (AUM), a considerable move from the current AUM of less than $500 million.
Like many other CTA/managed futures fund managers, Jamet has also looked across the Atlantic to the ‘40 Act funds, the US equivalent to Ucits. While superficially attractive, there are huge upfront costs and Jamet believes opening a ‘40 Act version of Epsilon is not cost-effective until the strategy has more AUM overall. "I think an alpha absolute return strategy is good for investors. We're thinking about [that], but for now, we don't have any plans in this area," says Jamet.
What Jamet wants to concentrate on is improving Epsilon for its current investors.
Lyxor Asset Management's Epsilon Global Trend also won best macro alternative Ucits fund and alternative Ucits fund of the year at the Hedge Funds Review European Single Manager Awards 2016.
The week on Risk.net, October 6-12, 2017Receive this by email