A month ago many computer-driven investors watched astonished as markets were wracked by a financial tempest, with once-hot stocks tumbling and previously shunned sectors enjoying a revival. But for Michael Kharitonov, the chief executive of San Francisco-based Voleon Group, the rapid rotation out of momentum stocks that wrongfooted many investors was "boring". The $6bn-in-assets hedge fund hardly noticed the brief but dramatic reversal. "We saw nothing," he says, with a chuckle. Voleon sees itself as at the vanguard of the next wave of quantitative investing, using machine learning to unearth patterns too faint for others to detect.

Financial firms have generally been slow to accept artificially intelligent stock pickers. They have already invested billions collectively to bring in quantitative analysts, who do major number crunching as well as develop powerful non-AI algorithms. Some argue that there isn't a lot of easy money left on the table for AI to pick up. Even so, hedge funds are now starting to turn to AI to give them an edge. Hedge fund managers, with their high fees (typically 20 percent of profits and 2 percent off the top of whatever an investor puts in), need to have healthy returns to justify the costs.