Marc Levitt has been interested in trading and building quantitative models since taking graduate level courses in chaos theory at the University of Illinois, but it took him a long time to launch his commodity trading advisor (CTA), Silicon Valley Quantitative Advisors, and an even longer time before focusing on it full-time.
Though he devoted much of his academic effort to applying chaos theory to financial markets, after he earned his doctorate in electrical engineering in 1990, Levitt took a position with Sun Microsystems to learn about business and technology.
"I kept my interest up in financial markets research while working at Sun," Levitt says. He would correspond with hedge funds and CTAs, exchanging system ideas with them while working in Silicon Valley and looking to get more involved in trading.
"Sooner or later I had to decide what to do," Levitt says. "That is when I decided to start High Frequency Finance, my software company, to develop software that could help traders."
Levitt worked with hedge funds and developing overlays concentrating on forex markets, actually developing his own carry trade benchmark. He left Sun in 1997 to concentrate on his software business, which helped several hedge funds improve their risk-adjusted returns. But, the Russian debt and Long-Term Capital Management crises in 1998 caused business to dry up and Levitt decided to develop his own trading program.
He went back to the semiconductor field during the day and moonlighted developing his CTA that would look at a longer time frame as opposed to his high-frequency research. "That was the time frame that fit my profile, working with daily data. It had nothing to do with those [high-frequency] techniques not being successful; it was just not how I decided to go about building a CTA," Levitt says.
His U.S. Quantitative Program (UQP), which trades 36 U.S. futures markets in every sector except stock indexes, got off to a great start, earning 83.08% in 2000, its first full year of trading. UQP basically takes a medium-term systematic trend-following approach, but Levitt built in some countertrend elements that helped it outperform classic trend followers.
"Risk-on/risk-off is very popular these days but there always have been cycles like that. In the old days people would talk about the market trending or not trending," Levitt says. "One of the key things is detecting various market states and then applying the right trading strategy within those market states. That is what we look to do."
He also was able to apply some of his higher frequency modeling to the system. Levitt says his core model is more of a breakout system, but adds, like most trend followers, his core trending signal is not the key differentiator but how he applies risk management and how the model chooses what state the market is in.
"Where you really find differentiation is how you go about all these other parts of the total trading system. When to know trend-following should be working pretty well so go big, or don’t go big, or don’t even follow the trend and when to know that on a probabilistic basis."
Levitt’s different models get enabled at different times but it all is running under the hood of his system. "I don’t go and say ‘it is a risk-on market (to use the current terminology), let’s use this set of trend models; it is risk-off, let’s use this other set of trend models or counter trend models,’ it is the models that determine that, it is all quantitatively driven," he says.
While working to get his CTA off the ground, Levitt took a position overseeing a global team of 250 employees at Cadence Design Systems. When his main project was completed in 2006, Levitt left Cadence to concentrate full-time on his CTA.
The program has been the same since 1999 except for a reduction in leverage following a -33.59% drawdown in 2005. Levitt did not apply the change until after the program quickly worked its way back, returning 55.43% in 2006.
Now he is concentrating full-time on marketing his program and growing strategically.
"I have been an executive at companies like Sun and Cadence. I also have been part of start-ups that literally were growing 100% per quarter and understand that kind of growth rate," Levitt says. He adds that it is as important to grow the business infrastructure as it is to grow assets under management.
He also is concentrating on developing new techniques, models and strategies, including a short-term program where he can utilize a lot of his high-frequency research.
Behind all his work has been a passion for modeling trading strategies. "Starting in grad school and continuing on, it has always been my second full-time job," Levitt says.
His UQP program is up 17.5% through July and has produced a compound annual return of 17.42% since its November 1999 launch. Something he was able to accomplish while also being a high-profile designer and executive in Silicon Valley.