That's the hypothesis this experiment is trying to validate but so far I have no reasons to believe they will behave much worse than human portfolio managers.