A quant’s guide to detecting a future "Madoff"

Worried about getting taken in by an investment management Ponzi scheme?

With the SEC ratcheting up its fraud detection efforts, it’s less likely that you’ll get scammed, based on what I heard at the CFA Institute’s GIPS conference last week. But the conference also introduced me to a quantitative method for detecting fraud in “The Importance of Risk and Attribution in the Post-Madoff Era” by Dan diBartolomeo, president and founder of Northfield Information Services.

The solution boils down to identifying investment returns that aren’t economically feasible. The effective information coefficient is an important tool for that, said diBartolomeo. 

A personal commitment to preventing future Madoff-style fraud 
DiBartolomeo wants to make people more aware of–and attentive to–risk.  He’s so committed that every year he hires a pickpocket to attend his annual client conference and warns his clients that “Keith the thief” will be targeting their wallets, watches, and other possessions. Despite the warning, each year, diBartolomeo has to return a pile of stolen goods. Keith succeeds because he’s good at distracting people–and Bernie Madoff was good at this, too, said diBartolomeo.

Speaking of Madoff, diBartolomeo’s firm was involved in the efforts of Harry Markopolos to uncover the secret to Madoff’s steady investment returns. At the time, diBartolomeo only knew that he was analyzing the returns of Manager B. But within a few hours, analysis revealed that Manager B’s returns “were either fictitious or had arisen from a strategy other than what was being represented to investors, wherein returns were probably being enhanced by illegal means.” You can read more of the details of this analysis in a March 2009 FactSet podcast with diBartolomeo. 

How to uncover a fraud 
“Do these returns make sense?” That’s an essential question for those who perform due diligence on potential investments, according to diBartolomeo. Returns-based methods aren’t adequate for analyzing this question, he said. Instead, one needs “a risk-based measure of investment performance that can detect manager skill(or lack thereof) quickly.”

The information ratio is one place to start, but it has flaws. The information ratio has nothing to do with making money for investors,” said diBartolomeo. For example, the information ratio would look great for a manager with alpha of 1 basis point and a tracking error of zero, but the manager’s clients wouldn’t benefit much. He also pointed out that “the statistical significance of a ratio is hard to calculate.”

The effective information coefficient (EIC) could be the answer to this problem. For more details on the EIC, read “Measuring Investment Skill Using the Effective Information Coefficient,” which appeared in The Journal of Performance Measurement (Fall 2008). 

I wonder what Madoff’s EIC was. I don’t know if diBartolomeo got an opportunity to calculate it.

Oct. 31 update: diBartolomeo’s talk is now available as a podcast from the CFA Institute.