CFM has run over a couple of significant bumps in the road on the way to becoming a member of THFJ’s EUROPE50 ranking of the biggest managers of single-manager hedge funds. The bumps in the road will be covered after we have heard about the influential background of the principals of CFM.
Research at CFM
It is a cliché of finance to describe someone who is numerically literate as a “rocket scientist”, but in several areas of finance the phrase has some truth to it. One of the areas which gives veracity to the cliché is in high-end Commodity Trading Advisors (CTAs), or more generally in areas where computer science and statistical methods are used as the core approach to markets – quantitative approaches.
The computer-enabled systematic trading approaches dominate the rankings of CTAs by size (see THFJ website for profiles of Aspect Capital and Winton Capital Management), and CFM is up there through this approach. Marc Potters carries the title of Managing Director, Research and Dr Potters has a PhD in physics from Princeton University. Jean-Philippe Bouchaud is Chairman and Chief Scientist of CFM; he and Dr Potters supervise the research team, while maintaining strong links between CFM’s research and the academic world. Jean-Philippe Bouchaud obtained his PhD in theoretical physics from the École Normale Supérieure in Paris, France, and has worked at Service de Physique de l’État Condensé at the Commissariat à l’Energie Atomique, and at the Cavendish Laboratory in Cambridge in the UK. Dr Potters co-authored “Theory of Financial Risk and Derivative Pricing” (Cambridge University Press) with Dr Bouchaud. This book includes statistical models of returns and correlations, extreme risk control and option pricing beyond Black-Scholes. All these topics have become important to the success of CFM.
The background of the two men who head the Science Centre of CFM has been a template for those chosen to join them. CFM’s researchers are highly trained scientists, with PhDs in physics and other hard sciences from highly regarded academic institutions. The academic flavour does not just reflect from where the staff have come, but academic engagement is considered a key part of the way the research department operates: the key researchers often teach and supervise PhD students and post-docs at academic institutions such as École Normale Supérieure and École Centrale de Paris, as well as various summer and winter programs, and professional training courses. These links to academia allow CFM’s research team to stay on top of new concepts and in touch with scientists who are inclined to consider a career in finance. CFM has approximately 34 researchers, mostly physics PhDs, and 32 IT engineers, out of a total headcount of over 90 people. Over the years, the team has published over 40 papers in international academic journals and documented numerous other internal concepts that serve as an institutional memory at CFM.
In order to pay back to the scientific community from which it draws CFM has set up a foundation for research. The foundation will pay for students doing PhDs in physics or maths for three years at a level to give a decent standard of living.
CFM does not recruit staff in research with a background in finance or markets. Instead they want to bring in fresh minds for problems that the inductees have not seen before. Chief Executive Jean-Pierre Aguilar is quite specific about what he is looking for in potential new recruits. He says: “Quantitative management of capital is a very challenging business. Everybody is looking for exploitable market anomalies. To be successful in this business you have to identify anomalies which have been around for several years so that they can be statistically validated. Everybody is looking for them, so the commercial angle is that in addition you have to identify them before others are able to. And to do that you need smart people, but being smart is not enough,” he emphasises.
“We do not believe in recruiting “genius” staff who cannot communicate their ideas, who cannot motivate the people working around them,” he continues warming to his theme. “We are looking for other qualities – they must be able to create, and to interact with their colleagues well. We are proud of our hiring record – people have been working with me here for 20 years, and our staff turnover is very low. Those things, and the fact that 20 people who knew me from working for me at the software firm (Ubitrade) wanted to come over to work with me at CFM when I sold the firm, that makes me more proud than the amount of money we generate or manage.”
Starting in finance
The CEO relates how he started off in finance. “I’m a computer scientist, and in 1986 when I started at a small brokerage and asset management firm in the bond department I was asked to head up the firm’s effort on MATIF as I was seen as a technocrat. Bond futures were a new market at that time, and so naturally there were terrific opportunities. I was able to put on cash-and-carry trades that returned twice the risk-free rate. It didn’t last, and within 18 months that financing margin had been competed away.” So there was an early lesson that markets change and you have to adapt to carry on making returns.
Other market opportunities were recognised by Jean-Pierre Aguilar: “there was a chance to make good money from baskets of the cheapest-to-deliver bonds for bond futures, though there was some market risk. My boss at the time preferred not to take that kind of market risk; his preference was for very low risk strategies like tax arbitrage. So we didn’t put on these any of the proposed strategies. I was still working on things like linear filters on US treasuries, but they were not put into practice. I was grateful to my boss though, because he warned us about a possible crash in markets (the 1987 crash) and we had positions on that made us money individually.”
Aguilar decided to use his profits from hedging against a crash to start his own business, and so in April 1988 he started a software business for listed derivatives called Ubitrade. It provided front-office tools that give option greeks and values for listed options and futures, and back-office systems to handle them. Through the course of time Ubitrade built to the point of having over 200 banks as customers, and eventually the company was sold to GL Trade in 2004.
However whilst running Ubitrade Aguilar had not given up his idea of using “systematic filters”, as he calls them, to trade markets and so he formed CFM in January of 1991. The concept of this form of systematic trading was well understood by high-end trading managers, and so Aguilar went to New York and obtained backing at launch from what were some of the biggest names in the hedge fund business (firms that are behemoths of trading today). At this point, whilst he saw his firm somewhat isolated as a CTA based in France, there was a perceived and real advantage in trading European futures markets systematically. The returns offered genuine diversification for existing CTAs prepared to farm out some capital, as the CTA sector was nearly exclusively an American phenomenon trading American markets at the time. So for a while investing in this European CTA was a hedge for American capital, before the big traders moved into Japan and Europe themselves.
A few years after the founding of CFM Aguilar thought that he was just not good enough himself as a researcher into markets to sustain the business long term. He decided he needed to set up what he terms a “science centre”, or research department, and he needed a person of real quality to run it. Jean-Philippe Bouchaud was approached, but he was not the complete solution because he wanted to keep active with his scientific interests. So the company Science & Finance (S&F) was set up in 1994 so that Dr Bouchard could carry on with his scientific work on a consultancy basis whilst working exclusively for CFM on financial research. The “person of quality” that did ultimately take the lead on research was Dr Mark Potters who started with CFM in 1995 as a quantitative finance researcher. At this time S&F was conducting research for major US trading groups as well as for CFM. The research for outside clients was very useful to the CFM staff who worked on it. Through these projects the CFM staff learned about how to trade stocks and how to trade options, because prior to this the French money manager only traded futures. In particular they gained experience in how to set up their systems to address these new asset classes. These research clients were at the leading edge of market-making practice for US listed options, and through gaining an understanding of how to price options, wholly different investment strategies became open to CFM (see box describing strategies within the Stratus Fund).
Concentrating on CFMThe end of the last millennium was a significant time for financial markets and no less for Jean-Pierre Aguilar and CFM. The dual role for S&F carried on for a number of years until the CFM CEO wanted to concentrate efforts on producing research just for CFM. So S&F was merged with Capital Fund Management in 2000. Research is singularly important at systematic traders because markets are efficient and will adapt to the strategies and tactics applied by market participants, even if it is smart and/or fast money. So it is important to keep trading strategies fresh, as doing exactly the same thing over and again will produce diminishing returns in time.
Around the year 2000 Aguilar also wanted to realise the value of his software business Ubitrade, and in the fevered atmosphere of the time it was easy to raise cash for Ubitrade, but less easy to use it as a means to take over other software businesses, which was the intention. The problem for Aguilar was the valuations placed on other businesses at the time – all based on clicks and multiples of sales and imputed benefits of an accelerated time-to-market. It was difficult for a rational businessman in the fevered atmosphere of the time. Aguilar passed on day-to-day management responsibility for Ubitrade to someone else so that he could concentrate on CFM, even though the money management firm only ran $150m then. Aguilar kept his majority interest in the software supplier until it was sold to GL Trade in December 2004. So from the early noughties Aguilar was able to concentrate his attention on CFM.
According to the CFM Chief Executive the key to success in the CTA strategy is to manage the research on a dynamic basis. The application of this idea has had an impact on the firm that is perhaps surprising. One of the central tenets held by investors in CTAs in the 1990s was that systematic approaches to markets required the application of discipline; in fact an iron discipline was the preference. “The perception of investors was that sticking rigidly to a static system enabled systematic funds to ride through the fire of stress periods,” says Aguilar. “In fact, changing the system was seen as a red flag to investors in CTAs. They thought that there was a danger in using something that may not be thoroughly tested, that was not proven with real money.” Each time CFM introduced an innovation to the trading systems applied to the funds (as opposed to testing systems on out-of-sample data for quality assurance) investors withdrew capital from the funds when they heard about the changes.
“I realised that over the course of the 12 years we had been running money that after each phase of drawdown we had changed something and improved our return profile: we always rebounded well in terms of returns,” relates Aguilar. “We needed to react to drawdowns and we were able to react by changing systems appropriately.” This realisation empowered the CEO to suggest a new way of operating. “I said to my partners that we were wrong in not dynamically introducing innovations to the trading systems as they occurred, rather than waiting for a degradation in return to give us a reason for change. We had been stuck in a way of operating for fear of upsetting our investors.”
But investors did not react well to being told that CFM was going to change the trading systems on a more timely basis, that is not wait for a drawdown before introducing an enhancement to the trading systems that were applied to investors’ capital. Given the track record Aguilar thought that they would persuade the vast majority of investors to stay in the funds where they were invested, but that was not the case. Investors can come very skittish when they hear of any change at a manager, for many investors the effect of changes to trading systems need to be observed from the safe perspective of non-holders — they got out. Assets under management (AUM) dropped by more than a third in less than a year. This was the first bump in the road experienced by CFM in the firm’s second decade.
The management decision to change the timing of significant enhancements to trading systems was a point of maturation for CFM as a quantitative money manager. Aguilar’s view is that by running money over the 12 years to that pivotal point in 2003 they had already been operating for four times the life-span of a typical hedge fund. And by doing so successfully they had earned the right to run the money as they saw fit within their chosen strategy and style. This new thinking liberated the firm to introduce new ideas to the production versions of their trading systems as they saw best themselves. This is a key conception that all superior hedge fund managers utilise, some from day one, but for some it dawns over a period of time. They are ultimately running money in a style that satisfies them — that they are happy with themselves.
To the extent that their way of operating is compromised by outside pressures such as that from investors the managers will not be true to themselves. Ultimately investors buy into the managers and their way of doing things. If you like, the style is an independent variable and the dependant variable of AUM is a function of the style of the manager and the fit of the risk profile desired by investors and that delivered by the manager over the course of time. This concept is classically reinforced in the hedge fund business by the managers themselves owning a significant portion of the capital they trade/invest, an idea that goes back to A.W. Jones and was taken to a new level by George Soros, and to its apogee by Jim Simons.
Renaissance Technologies’ Medallion Fund is now exclusively owned by Jim Simons and others on the staff of the firm as external investors have been gradually bought out. In the case of CFM, CFM’s partners and employees represent 8% of the Stratus Fund’s AUM.
It turns out that Jim Simons is seen by Aguilar as an inspiration: “Jim Simons is a very, very bright computer scientist,” enthuses the majority owner of CFM. “He is innovative, he is able to put very good ideas into production in the market, and his track record with Medallion is very impressive. The fact that he was able to return real money in buying out external investors in Medallion showed that it was a real business delivering real returns. Plus he has been achieving what he was supposed to with Renaissance Institutional Equity Fund — beating the S&P500. Jim Simons is above everybody else doing the same job. His achievements are very motivational in showing what is possible in applying a structured quantitative approach to markets. He is a real role model for those doing what we do.”
The second bump in the roadIf the first bump in the road of the story of CFM could be described as partly self-inflected, as they chose to introduce innovations on a dynamic basis and lost investors as a consequence, the second was certainly an external event. All CTAs use futures to take market exposures, against which exposures they post margin with exchanges. The margin posted with exchanges is a small proportion of the value of the contract value of the exposures. So trading with futures allows the trader to leverage exposures (exposures greater than capital), and CTAs except in extreme circumstances have a lot of cash left after posting margin. The cash can be left on deposit, but is often posted with a “cash enhancement programme” that aims to deliver cash-plus returns with very low risk. CFM had cash from the CTA program Discus with Sentinel Management Group, a cash-management firm based in Northbrook, Illinois. Sentinel invested for clients such as managed-futures funds, high-net-worth individuals and hedge funds that want to be able to withdraw their cash quickly. Investments included short-term commercial paper, foreign currency, investment-grade bonds and treasury notes, according to the Sentinel website. Sentinel froze client withdrawals on 14 August 2007, citing credit market volatility.
The prices of the instruments listed on the website would not suffer disproportionately from the “effects of credit market volatility.” Something else was going on. It might be recognised that this was the time when distress in mortgage-backed securities started to be widely evident. The cash manager, like many fixed in money management were “reaching for yield” by holding higher yielding but lower credit quality bonds. But what was going on there was more than just operating beyond the investment mandate. Late on Friday, 17 August 2007 the firm filed for Chapter 11 bankruptcy protection. The US Securities and Exchange Commission sued Sentinel, claiming the firm, which managed $1.6 billion according to pre-bankruptcy financial filings, used client assets as collateral to obtain a $312 million credit line from the Bank of New York. In addition the SEC filed a complaint in court that Sentinel had improperly commingled $460 million in securities from clients’ investment accounts into Sentinel’s proprietary “house account”.
CFM was exposed via the CTA program Discus, and then via the Stratus Fund that had an allocation to Discus. Jean-Pierre Aguilar experienced the nightmare of having to write letters to his 600 investors informing them that his $4 billion company could lose as much as $407 million, of its Discus Master Fund. Then came the series of phone calls to hold the hands of all those investors.
CFM was not a passive bystander in the process of recovering assets for creditors of Sentinel. CFM was instrumental in whistle-blowing to the National Futures Association as Sentinel imploded, and then a senior executive from CFM chaired the investor committee after bankruptcy of Sentinel. Still, there was damage to CFM’s business even though they themselves had been victims of Sentinel providing falsified account statements.
Aguilar decided to be very clear with investors at the time. Rather than say nothing and risk lawsuits against his firm Aguilar decided to be transparent with his investors and tell them about the problem. He let them know that 10% of the Fund may have been jeopardised, but that CFM would fight to recover it. With 80-90% liquidity in the remainder of the Fund, redemptions in the Fund were suspended for only a month. No investors were trapped in the CFM funds. Yes, assets under management at CFM fell by more than 20% straight away, but within six or seven months they were back to the same level. To date 26% of the cash lost has been recovered, and further lawsuits are outstanding against various parties.
Whilst CFM has a track record of 18 years for the managed futures program Discus, of more importance is the track record of the more diversified product Stratus Fund. The returns shown in Tables 2 and 3 have some very attractive attributes. There is no correlation with the S&P 500, and in the analysis of returns it is feasible to see a smoother return series than is typical for a fund that utilises just directional strategies like a typical CTA.
This is very clear in the proportion of months with positive returns. The Stratus Fund has achieved 70% positive months, as against 58.2% for Winton Futures Fund and 56.8% for Aspect Diversified Fund. The latter two pure CTAs have produced higher absolute compound returns – 18.37% and 11.56% respectively versus 9.15% for the Stratus Fund (USD) according to HedgeFund.net. However the standard deviation of 12-month rolling returns for Stratus comes in at 7.16%, as against 15.25% for Winton and 14.29% for Aspect. So putting the absolute return against the volatility of that return gives Sharpe ratios of Stratus 0.80 (Sentinel bump included), Winton 0.97 and Aspect 0.62 according to HedgeFund.net.
As ever one has to be careful with statistics, and for this case it is important to bear in mind that the Winton and Aspect track records covered periods 1998-2002 with some great years for CTA returns. For further comparison the Credit Suisse Tremont Managed Futures Index has produced an annual average return of 6.62% since January 1994, with a standard deviation of 11.85% to give a Sharpe ratio of just 0.24. Even so the inference is that for allocations to CTAs/systematic strategies investors might obtain a more consistent return profile by investing with CFM’s Stratus.
Last year and this are quite indicative in illustrating the different returns coming from Stratus as a Fund diversified beyond directional strategies. The Credit Suisse/Tremont Managed Futures Index was up 18.33% in 2008, and is down over 6% this year. Stratus was up only 8.36% last year, but has had four up months in a row this year to give a YTD return of 3.49%. As ever in fund choice there is a give to get.
You have to do it over every day
Jean-Pierre Aguilar describes his role at CFM. “My job here has several elements. I am here to hire the right people and organise the company to operate efficiently. The second part is to manage the investment risk we take. This is very important because there is no one system that is going to tell you that we are in a very uncommon market environment. It doesn’t happen very often, but every now and then you have to put your hands on the leverage and turn it down. It’s always in the same direction: you must take your risk down. A third element is communication with investors, both new investors and existing ones. The last part is one that, hopefully, we don’t see very often which is crisis management. When you have a very, very big problem you are very alone as an executive. This is the same for any business with a major problem, not just a hedge fund business – as Chief Executive you have the whole team looking to you to find a solution. It can feel like you are in a tsunami, when 600 investors all want to speak to you at the same time! You ask yourself over and over: “have we done the right thing?”
“Over all, this is a very interesting job, but you can never sleep when you have it. You can never say you have achieved a position, rather it is a process, and you have to do it over every day. I can never be out of touch. Even when flying over the Kalahari desert I have to be in contact, and I check the Fund performance and my e-mails every day. Yes there are problems to deal with, but the satisfying thing is to find solutions to them. The pleasure of dealing with pressing problems, and finding things in research that other people have not been able to find is very gratifying.”
Today CFM is a $2.5 billion business, and the management of the firm has been stress-tested several times over the span of its existence. This tempering in the fire will stand CFM in good stead to handle whatever is ahead for it.
Whilst recognising that there is a trade-off between size and what can be delivered, CEO Jean-Pierre Aguilar sees scope to grow further. “We want to stay in the high-frequency trading space, and continue to be successful,” he states. “For us that means we could be running $6 billion in two years time without impairing returns on current capital. We may have to add more scientists to get there, and market liquidity needs to hold up; but it is possible, yes.”
CFM by numbers
• CFM has six data centres to reduce latency issues on dealing.
• Over 70% of the capital in the Stratus Fund comes from Funds of Funds (40%) and Pension Funds (30%).
• Stratus AUM dropped by $700m from its 2008 peak through redemptions in the second half of 2008.
• Stratus minimum investment is $1m
• Straus Fund has monthly dealing (subscriptions and redemptions) with two months' notice.
• The trading universe of Stratus includes 3,000 equities, 20 FX pairs, 100 different futures, and options on 1,000 equities.
Stratus strategies
CFM has a track record for the Discus CTA program that goes back to 1991. Discus is a component of the Stratus Fund, now the only open /investable Fund CFM offer to investors. Stratus invests in the three core programs, all of which are purely systematic:
• A high frequency global macro statistical strategy that trades the 100+ most liquid futures, FX crosses and on-the-run US treasury bonds around the globe
• An equity statistical arbitrage program across a 3,000 stock universe (US, UK, Europe, Japan, Australia) constrained by market neutrality
• A US equity option arbitrage fund across 1,500 optionable stocks constrained by delta, theta and vega neutrality
The three strategies, each with the same risk weightings and uncorrelated returns, are combined within Stratus to deliver a smoother return series through diversification. CFM have observed that this EquiRisk portfolio is actually very hard to beat out-of-sample. There is a 99% daily VaR budget of 1.25% and Stratus has a return objective of >15% p.a. The probable worst intra-month month loss is estimated to be 4.5%.
Discus: A high frequency strategy trading futures, FX and government securities that takes short-term positions (about three days average holding period) generating a turnover of approximately 8,000 round trips per year per $1 million; at present it trades approximately 100,000 futures contracts per day. Discus is purely systematic and takes short-term directional positions in the major international futures markets; spot and forward FX electronic trading platforms; as well as on electronic platforms for trading US treasuries. Discus uses two types of models, one being purely price based and applied universally across all markets, while the other is more fundamentally based, applied to a subset of markets where certain prices are conditioned with fundamental data. Resulting signals can be mean reverting or trend following with patterns ranging from a few minutes to several days.
Ventus: A global equity statistical arbitrage strategy which is medium-term, positions being held from one day to one month (average holding period seven days). Each day, on average, 75% of its assets under management are traded in the market. The portfolio is constructed to be dollar neutral and sector neutral, with the particularity that industry sectors are dynamically calculated from observed large correlation matrices. Ventus uses two types of models, the first one being purely price based, with statistical filters picking up market mean reversion on different time scales, as well as lead-lag effects. The other is fundamentally inspired, using quantitative models that use fundamental data to condition prices and generate trading signals.
Nimbus: An equity volatility arbitrage strategy whose objective is to extract value from trading options (delta neutralized) using CFM’s proprietary model for predicting future realized volatility. The program trades approximately 120 x AUM per year with stock hedges and exchange-traded funds (ETFs), and approximately 4 x AUM per year in option premium. The US single stock options’ universe is scanned for opportunities in the volatility space; positions are taken balancing longs and shorts, and traded options are held until expiry (making the average holding period two months). Profit opportunities are detected by identifying spreads between observed option prices and computed prices using CFM’s model, and idiosyncratic name risk is managed by limiting the amount of theta risk per single name. CFM aims to further control risk through diversification, continuous stress testing and correlation limits.
