Machine learning study points way to smarter beta

‘Boosted trees’ method uses same metrics as conventional factor investing but mixes them in new ways

When it comes to designing smart beta investment strategies, asset managers should consider ceding their role to machines, the results of a recent experiment suggest.

Researchers used a simple machine learning algorithm to pick stocks systematically using many of the same metrics as those used in conventional factor strategies, but with the algorithm learning how best to mix and match them in response to changing market conditions. They found the method resulted in higher average returns than t