This content is from: Innovation

A Strategy for All Seasons (Including Inflation)

The 30 year evolution of Discus from trend follower to a sophisticated, low correlated strategy that’s not easily reproduced.

Investors around the world are focused on the likelihood that inflation will stick around a bit longer than whatever length of time “transitory” might have indicated. As investors mull what might be best for their portfolios in an inflationary regime, a venerable systematic strategy merits revisiting. Known as Discus and created by Capital Fund Management (CFM), it has evolved into a remarkably sophisticated and difficult to replicate alpha-generating strategy since its inception 30 years ago. Inflation is among the countless scenarios it assesses with its models that include nearly 200 predictors across all asset classes.   

Inflation is one of many things we try to anticipate,” says Yves Lemperiere, Head of Research, Alpha Strategies, at CFM. “We have lots of different ways to forecast it – backward-looking, forward-looking, scraping the web, using market products. The sheer breadth of data we look at within Discus makes it much more of an all-weather product than a standard trend follower.”

Along with resilience in various market regimes, Discus is far nimbler than a typical CTA. When it was first hatched, Discus was essentially a trend following strategy. Had it stayed that way, it would likely function as most CTAs do, taking a long or short position across an entire sector. But Discus morphed. Despite its roots, it has greatly reduced its exposure to trend models. Rather than being overly directional, Discus incorporates relative value in a significant way.

“Because our predictors and our portfolio construction are quite specific, we are able to take a relative value position. So, for example, inside each sector we mitigate the net long on a short bias by taking relative value with two indices,” says Lemperiere.

What that might look like in action, says Lemperiere, comes into view in market neutral equity strategies that match indices against each other. “The CAC 40 [French] and DAX 40 [German] may be correlated, but clearly CAC minus DAX is something pretty new. This is a model that leverages the difference between two countries, focusing on idiosyncrasies of each and hedging with other similar stocks. It could be Europe versus U.S, it could be anything. Because we have such a broad range of predictors, and some predictors focus on relative value components, it turns out that about 30% of our risk is on relative value modes. And to play relative value requires much bigger positions to deliver a certain amount of risk than if you’re fully directional.”

Yves Lemperiere

Decorrelation at heart of Discus

In challenging environments such as those driven by inflation, investors become more keenly aware of over-correlation within strategies and across their portfolios. When Discus came into being 30 years ago, stock indices globally weren’t excessively correlated. However, the coupling between markets in different countries increased thought the 1990s peaking in 2008 with the global financial crisis until, broadly speaking, most economies and interest rates were moving together. 

Combine the vast number of models and predictors in Discus, and the massive quantity of data that feeds what Lemperiere lovingly refers to as “the beast,” and you are working with a recipe for low correlation.

“When we say Discus offers strong decorrelation, it’s not just talk,” says Lemperiere. “It not only has low correlation from traditional risk premiums and the main asset classes – so bonds, indices, etc. – but also has reasonably low correlation to trend following strategies. And we are also decorrelated in the tail. Many investors are long equities and really want to know what’s our performance when the market tanks. So, we’ve conducted several case studies, and it turns out Discus has very low correlation. It shows slightly positive performance during extreme negative events in the market. Overall, you are only mildly correlated to the trend.”

There is a lot under the hood of Discus. A constant focus on the strategy being more dynamically adaptive than any other creates a mixture of technical and fundamental aspects that may be a little daunting to some investors – but for the same reasons it’s a differentiator for a portfolio says Lemperiere.

Yves Lemperiere

Vigilance around cost efficiency

Discus – and all strategies at CFM – are built upon three pillars: research, technology implementation, and trade execution. Minimizing transaction costs is increasingly important not only to investors, but to the success of investment strategies, too.

“We put a lot of effort into research because you can find wonderful alpha, but you can blow it up in execution,” says Lemperiere. “Transaction costs can noticeably alter the performance of a program, especially for something like Discus which is a faster paced, medium-frequency program with a holding period of around 20 to 25 days. We have a team of researchers dedicated to putting algorithms in place to execute orders, and then analyzing the results to help us model the transaction costs.”

The process Lemperiere describes has resulted in a greater understanding of how the trades themselves affect transaction costs. “For sizeable trades, not only do you pay your fraction of the spread, but you move the price up when you buy. The more you buy, the more the price goes up and the higher your average price. We call that impact, and we’ve spent a lot of time modeling it.”

Similarly, the knock-on effects of the impact Lemperiere describes have permanent and transient qualities, the latter of which will revert once buying ceases and the price is no longer being pushed. “Modeling execution cost is a major part of portfolio construction at CFM,” he says. “We can do it because we understand costs and have invested a lot of time studying it. We model everything we can. For example, you trade one asset it affects another asset. We call that cross- impact. There’s a whole range of complexity in modeling transaction cost – in short, just saying that it’s a quarter of the spread or half the spread is naïve. If you want to run a relatively high- or medium-frequency trading strategy, you need a more sophisticated model.”

Research driven

Much of the success of Discus – which recently received the nod for The Hedge Fund Journal CTA performance award for best risk-adjusted returns over five years in the systematic short-term trader category – can be traced to the research team and process at CFM. Autonomy is prevalent. If a researcher comes upon a topic they find interesting, or stumbles across a time series or data product that could prove useful, time is carved out for them to pursue their instincts.

In the complex and highly nuanced strategies of Discus, it’s no mean feat to do the supporting research and make sure the data is there each day at the right time. The research team at CFM consists of three groups equal in importance: a predictors team, portfolio construction team, and execution team.  

“It’s important that we keep a relaxed and collaborative culture,” says Lemperiere. “We want the team to interact and challenge each other, point out possible mistakes, and find solutions together.”

Lemperiere says the next big game-changer for Discus is alternative data – not that it’s entirely new to the strategy.

“What has changed is the coverage we get from alternative data providers,” he says. “Five or six years ago we noticed the data quality on futures wasn’t as high was we wanted. It took a few years to find the right data providers – this is not low-hanging fruit. Today, the quality has greatly improved. I’m sure brokers realize it, too, and they’re revamping themselves as data providers. They have data catalogs with insights into flows, analyst reports, and so on.

“To extract value out of that takes a bit of skill,” he continues. “Discus is well positioned for that to be very additive.”


Any description or information involving investment process or allocations is provided for illustration purposes only. There can be no assurance that these statements are or will prove to be accurate or complete in any way. All figures are unaudited. This article does not constitute an offer or solicitation to subscribe for any security or interest.

Related Content