Quants clash: machine learning or linear models?

Some studies say the algorithms beat the common models; other studies say the opposite

There is little harmony among quants on what delivers the sharpest forecasts of stock prices: machine learning models or conventional linear models.

And the two camps came to Risk.net’s Quant Summit on March 5, armed with studies buttressing their positions.

Giuliano De Rossi, former head of European quantitative strategy at Macquarie who recently joined Goldman Sachs, told delegates there was “convincing evidence that machine learning can beat linear models”, even accounting for the