After a multi-decade long slope upward in performance, AQR took a hit in 2018, and Founder and Managing Director Cliff Asness, who opened Thursday’s Morningstar Conference with a keynote session “Understanding Intuition in Quant Investing,” began with a mea culpa. From there, the speech, interspersed with wry humor, broke down the workings of factor and multi-factor investing, and why multi factor is better.
“When quant strategies lose, it’s often hard to understand why … Feels like black box when losing,” he told the audience.
In explaining multi-factor (smart beta, a term he hates) investing, he used only two factors: value and momentum, and how they work together. He noted they use other factors, such as carry, defensive, trend and volatility, and apply them to markets beyond stocks: fixed income, futures, currencies and commodities.
He noted that “when the sum of a few of these is positive vs. negative is going to weaker than any one, is your intuition for when value does well or poorly strong? Yes it is.” He said they would rather do more factors over more markets. “It’s not a non-intuitive investment process, but specifically when it wins and loses is less intuitive.”
So based on a simple hypothetical stock portfolio, using value and momentum factors, each measured one way. He illustrated the performance of various factors for each factor separately, for example, the value portfolio had a 4.5% return with volatility of 7.2%, while the momentum, which has been weaker, had a return of 2.9% with a volatility of 9.9%. He noted it was “gigantically negatively correlated with value.”
However, combined, the factors, taking top third of these, tripled Sharpe ratios and doubled values. He further showed using standard deviation if the value portfolio had an SD of 2.5, and momentum had an SD of -1.5, that would be a good year for the combined portfolio. However, if value had a SD of 1.5, and momentum’s SD was -2.5, that would be a bad year for the combined portfolio.
He wrapped up saying that advisors can improve building a process that’s intuitive by:
- Building a process that’s intuitive and robust to data mining
- Sizing the bet reasonably
- Confront intuition head on
- If they have followed these, and tested to make sure there is no smoking gun, “stick like grim death to your beliefs.”
— Related on ThinkAdvisor: