Hong Kong native Chris Tam is the cautious, practical type; he gets things done. His partner, New York native John Philippakos, is the creative, passionate type; he comes up with things to do. Since 2003, they have run Future Gate Consultants Limited (FG), a Hong Kong-based CTA averaging 28% per year. The two met while pursuing MBAs in quantitative finance at the University of Southern California (USC) in 1998.
“It was the middle of the tech boom,” Philippakos recalls. “Everybody was day-trading, making a fortune in the markets, and I started to hear about quantitative trading while [I was] designing satellites.”
That was at Hughes Space and Communications, where he’d gone to work after earning a degree in aeronautical engineering from the California Institute of Technology (Caltech). Then he started experimenting with trading strategies, and wound up at USC.
Tam had come to USC after earning a masters in Finance and Economics from the London School of Economics, and the two spent long hours talking markets. Philippakos tried to persuade Tam to go into business with him even then, but Tam took a job as a financial risk and valuation consultant with Arthur Andersen. So, Philippakos showed some of his ideas to a wealthy California investor who bankrolled him.
“That saved me from having to go the normal investment banking route,” he says. “I got to work on my trading strategies, and Chris and I kept in touch. By 2001, my stuff was working and Chris was ready for business, so we began trading managed accounts.”
In 2003, they formed Future Gate, and in early 2008 they implemented the current trading framework, dubbed “Energetic State Analysis” (ESA). The system tries to identify the mode that a market is in, such as trending, counter-trending, or mean-reverting, and then applies both trading and money management techniques based on that assessment, which is made using a trick that Philippakos learned in engineering school.
“Think of it as a bridge between traditional quantitative strategies and pattern recognition,” he says. “Traditional strategies are constantly generating signals, but ESA tells you which signals to listen to.”
It does that by “decomposing” the market into constituent parts and then “recomposing” it using different algorithms designed to identify underlying patterns associated with different market stages. The result upon re-composition is either a complete mess or something resembling a parabola, a sine wave, or some other recognizable pattern.
A mess means the market is not in the stage being tested for, but a pattern means it’s probably in that stage, and the amount of original price data present in the pattern gives a clue to the reliability of the signal.
“If it’s 80% of the original signal, then we have a signal-to-noise ratio of 4:1,” says Philippakos. “That tells us how much money to allocate to which strategies in which markets at that particular time.”
All engineering students learn decomposition, he adds, but it took a lot of trial and error to find the algorithms that work. “By decomposing certain signals related to dynamics or shocks, you can better see what are the frequencies of these shocks or motions.”
ESA tracks more than 100 markets, and all in different time frames. The strategy usually applies a combination of approaches to individual markets, so if two strategies are giving mildly strong signals, the system applies both strategies. If one strategy is long and one short, they are flat, but if both are long, they have a strong long position.
“You basically mix the parts as you like,” says Philippakos. “It’s freaky. Some markets behave really well with weird combinations.” Those combinations determine the size of positions, and the system is constantly adjusting them based on gradually shifting states.
“We don’t use traditional price-based stops,” says Tam. “Instead, we continuously rebalance because every day the state of the market changes a little bit, so positions may drop from 20 contracts to 18 as they evolve over time.”
While Philippakos brings the engineering know-how, Tam conceived the risk management strategy and also manages the process of automating Philippakos’ ideas.
Tam takes the initial idea and fine-tunes it so it can be implemented into a strategy. “We work together to see how we can make these creative ideas practicable,” Tam says. They give the dynamic approach credit for their success in 2009, when many trend followers have been getting clobbered.
“In 2008, we were 70% trend following,” says Philippakos. In 2009 that dropped to 40% with most of their trades coming from the counter-trend and mean reversion strategies.