Michael Rulle is an emerging commodity trading advisor (CTA) with 30 years of experience in the trading business, most recently serving as president of a CTA with as much as $8 billion under management. Today, he is founder and CEO of New-Jersey-based MSR Investments LLC., a quantitative systematic short-term CTA.
From 2002 to 2007, Rulle served as president of Graham Capital Management, where he oversaw its discretionary portfolio managers. He also worked closely with Graham’s quantitative research group. Prior to that he was president and chief information officer for Hamilton Partners, served as CEO of CIBC World Markets and ran the derivatives business for Lehman Brothers.
In 2007 he parted ways with Graham and shortly thereafter decided to research his own quantitative strategies. "I always interacted with traders that had a technical bent. I always would have an [interest in] anyone who did technical trading or if they were technically part of the quantitative research," Rulle says.
So when he set off on his own he took a scientific approach. "I am a quantitative guy by birth," Rulle joked.
Rulle says he always has had an internal argument over building trading systems. "One [side of me] says it is impossible to make models that work consistently. That is the tough guy side and that guy is right 98% of the time. Then there is the optimistic guy who says there are anomalies that can be found if one uses the right disciplined methods to do so."
He began by studying one market, the S&P 500, and concentrating on the distribution of returns. "We study the distribution of returns in markets and we look at how those distributions differ from each other over different time frames," Rulle says. "If those return distributions — which overlap each other because they can be daily, hourly [or] monthly — differ from each other in a repeatable predictable way, you should be able to make money trading the difference between [them]."
Sounds simple, but it is not, as the risk of curve-fitting, somewhat of an obsession for Rulle, always must be considered. "The question is: Are they really different or are they randomly different? We believe that they really are different and our method of analyzing it persuades us that they are."