There’s no GPS for Outperforming Managers. But This Can Get Investors Closer.

There may be a better answer than past performance to the central riddle of active management — how to identify managers who have the skills to outperform in the future.


Illustration by II

It’s tempting to think that the past returns of managers will tell investors something about how their funds will do in the future. But the ubiquitous disclaimer in fund prospectuses warning investors about the uselessness of picking funds based on past returns is telling.

Investors, however, may be able to turn to another metric that may be helpful in finding promising funds.

New research has found that managers’ demonstrated skill at past investment decisions, which of course contribute to returns, can persist. According to Clare Flynn Levy, CEO and founder of Essentia Analytics, the research shows that the median manager who makes good decisions in the last six months is likely to continue to make good decisions for the next two years. Essentia Analytics uses data from active managers to help them improve their performance by reducing behavioral biases in their decision-making process.


The new findings grew out of a benchmark that Essentia Analytics developed a year ago based on managers’ actual trading decisions — a measure of decision-making skills in stock-picking, sizing, and timing. The methodology behind what the firm calls the behavioral alpha benchmark has been peer-reviewed and appeared in the Journal of Investing. A behavioral alpha score reflects a hit rate — the frequency of decisions they make that create value for the portfolio — and payoff, which measures how much value was added or destroyed by every decision. The benchmark was originally designed to tackle the perennial question of whether the investment returns of managers are the result of skill, which can be repeated, or luck.

Flynn Levy said the decisions are an additional guide for investors, in part, because there are just far more of them to judge than monthly return data.

“If you have a high behavioral alpha score now, your elevated score is likely to persist for a meaningful period of time — nearly two years for the median portfolio manager in our study,” said Flynn Levy. “After that period, the relationship falls below statistical significance: your skills don’t necessarily diminish ... but the statistical relationship we observed is no longer strong enough for us to say your high score is likely to continue.”

Measures of true skill are rare, even in an industry whose central objective is to generate investment returns above what an inexpensive index fund could provide.

“There’s a common-sense angle to this: if you have developed and demonstrated skill in anything (not even necessarily investment-related ... could be writing, singing, practicing law, whatever — anything that is achieved through skill, not luck), it’s reasonable to assume you’re not going to suddenly lose it in a very short time frame,” said Flynn Levy.

“Again, we’re not saying that the skill itself necessarily fades after two years -- we’re just saying that’s the point at which the relationship fades below statistical significance.”

According to Isaac Kelleher-Unger, a data scientist at Essentia Analytics, the research analyzed daily trade data from 123 portfolios, including stock picking, the timing of entry, scaling in, sizing of positions, adjusting those sizes, scaling out, and timing of exits. The researchers calculated the results of the decisions on the portfolio and compared that to the fund’s benchmark. After splitting the results into discrete time periods, the research found a statistically significant relationship between the quality of decisions made by a manager from one period to the next.

Essentia Analytics’ founder stressed the distinction between measuring investment skill, the basis of the firm’s new study, and investment returns, which are a combination of skill and luck — everything that can affect investment returns that the portfolio manager cannot control. “Largely because of the impact of this luck (good or bad), past returns are no indication of returns in the future,” Flynn Levy said.

But managers do have control over their skills.