Same Quantitative DNA in Distinct Strategies
Particle physicist Francisco J. Vaca runs two very different investment programs – high-frequency stock trading vs. multi-strategy futures trading – that nevertheless draw on the same skill set. He moved from physics to a commodity trading advisor 15 years ago and founded his own firm in 2000.
He has racked up quite a track record. For the 36 months to April 2010, Vaca Global Diversified has total return of over 77% and a Sortino ratio of 2.62, which makes the program competitive with the highest ranked large CTAs—see Top Ten in this issue for comparable Sortino ratios.
Here Dr. Vaca talks about how he approaches markets and risk as well as prospects for managed futures.
"The basis for back-testing and volatility forecasting is the same whether the instrument traded is a futures contract or a stock."
Opalesque Future Intelligence: What made you get a PhD in physics?
Francisco Vaca: My dream was to help advance particle physics. When I went to graduate school and worked in Fermilab, experimental particle physics was an up-and-coming, vibrant field. In the early 1990s a very large particle accelerator was being constructed in Texas. This Superconducting Super Collider, as it was called, was to be the largest every built. But toward the end of my studies, the Superconducting Super Collider project was cancelled. To do the kind of work I was in, you’d have to wait for 15 years for the Large Hadron Collider in Europe to be built. That was not encouraging to a young new graduate! So I pursued a career in trading.
OFI: How did you move from physics to trading?
FV: Technology and trading, in addition to physics, fascinated me from an early age. As a physicist, you’re always tackling problems. I saw trading as one of those problems and traded stocks for myself. After the Super Collider was cancelled, I was still working at Fermilab and a friend there told me there was a job for a programmer at a commodity trading firm He had thrown away the job ad, so I picked it out from the garbage! It turned out to be at C&D Commodities, which was co-founded by Richard Dennis of “turtle” fame. My professional investment career started there in early 1996.
OFI: What was the next big step?
FV: In 2000 I left C&D Commodities and started Vaca Capital Management. We have two strategies. One is a high-frequency trading program that trades only cash equities. The other is our futures program, which started trading in 2004 based on research I did from 2001 to 2003.
OFI: Is the futures program also high frequency?
FV: No. It includes short term models, meaning five- to 10-day holding periods, and medium-to-long term models, meaning 30 to 50 day holding periods. In developing this program, I thought hard about how to diversify internally. We know that nothing works all the time. I decided to trade in as many styles, time frames and markets as possible. To get diversification, the program uses 16 independent models. These cover four different styles—pattern recognition, counter-trend, trend and momentum. Each of those styles is further diversified by holding period. We trade 65 global futures, forwards and cash markets across all major market sectors.
OFI: How did the program do in 2009, when trend followers as a group lost money?
FV: Last year was not a good time for our trend following component, but having multiple styles is a great advantage; when one doesn’t perform, another tends to pick up the slack. In 2009 our counter-trend models bought some of the markets at May lows and made good profits, so we had positive overall performance. No trend model will give you signals like that. Many CTAs have embraced multi-time frames but not multi-styles—that’s probably the direction for future growth.
OFI: Do you switch models in response to market shifts?
FV: No, we allocate equally to the models—25% to each style. It is all systematic. As scientists, we know that with any model there will be unexpected situations that can circumvent your logic and damage performance. We always have a stop on every position we put on, but stops are not sufficient to prevent big losses, so we use another layer of protection—a set of limitations on risk per trade, per market, per sector and for the entire portfolio. In addition we have a probability overlay that tells us the odds of the market moving in a certain direction within a given period. We adjust the size of a trade according to the odds. This means that we pull out of super-long trends when the probability of continuing movement in the same direction declines, so we avoid the inevitable reversal.
OFI: Has the program changed since inception?
FV: We did make a significant revision in 2006. Originally the probability thresholds were too tight, causing us to filter out many good trades. From our experience by then, we realized that it would be better to loosen some of the filters, so we could take more trades. Our results improved, and not just in higher profits. The greater number of trades brought more diversification, so the risk/return tradeoff improved. That said, I believe that a sound system will work for a long time, on the whole. But you have to recognize that all systems go through cycles and have periods when they lose money— tweaking a system in response to a short-term situation does not typically improve it. Traders are too easily tempted to modify their models in a never-ending, futile chase for perfection.
OFI: How did the high-frequency trading strategy come about?
FV: When I started the company in 2000, my research indicated that I could create a strategy that would make money provided the technology worked. Later, this kind of strategy came to be known as high frequency. Back then it was expensive to develop the technology. If you wanted to do automated trading, you could not buy products off the shelf the way you can now. Trade servers, trade engines, the things we now take for granted were not available 10 years ago. So I developed an engine with a whole suite of software for electronic trading and started trading in 2001.
OFI: Were there other barriers beside technology 10 years ago?
FV: Another challenge was trading cost. The first year we traded that strategy (2001), we made gross 30% but had to pay 15% in brokerage commissions, ending with net 15%. We were paying as much as a penny or more per share. You can’t afford those commissions when you trade high volume. We would change brokers and go to smaller brokers who agreed to give us lower fees because we traded a lot and gave them high daily volume. After the collapse of the technology stock bubble, brokerage commissions came down. About five years ago (in 2005) our commissions stabilized. We don’t have that issue now.
OFI: Why did you start the futures program, given that high-frequency stock trading worked well?
FV: There are pros and cons for each. High-frequency trading has low volatility, but its capacity is limited. The strategy can accommodate around $100 million but beyond that diminishing returns may set in. By contrast, the capacity of our futures program is in the billions of dollars.
OFI: Do these strategies draw on the same methods?
FV: The two strategies are different in many ways such as asset class and method for making money, but under the hood they’re similar. Both derive from the same approach, share a lot of DNA. The basis for back-testing and volatility forecasting is the same whether the instrument traded is a futures contract or a stock. The risk calculations apply in the same way. I say this with some pride because it shows that with our knowledge and techniques, we can find different strategies that work.
OFI: Have investors changed in the past 10 years?
FV: Investors now want more. It used to be that what mattered was whether you could make money; volatility paid little role. Nowadays, the Sharpe ratios and Sortino ratios are the sticking point with investors. They want a good risk/reward tradeoff, as established by those measures. Another recognizable change is investors’ demand for longer track records. When I first started, a two-year track record was sufficient. Then it went up to three years, now it is more like five years.
OFI: How does the future of managed futures look?
FV: Over the years, very smart people came into this space, bringing sophisticated tools. We are using these tools to harvest the risk premium embedded in futures markets. More managers now recognize the potential and are coming into these markets. Anomalies in markets can be arbitraged away if a lot of people do the same thing—that’s a concern you hear about. But we expect that there will be some risk premium to be extracted for a long time and skilled managers will get better bang for the buck.
OFI: Do you ever miss particle physics?
FV: I keep in touch with people at the university and remain interested in their research. If I ever retire from trading, I could see going back to physics.
Opalesque Futures Intelligence
Thursday, June 24, 2010
Founders Q&A: Physicist Francisco Vaca discusses two very different quantitative strategies, one of them high-frequency, rooted in the same skill set.