We've tested this also, but it won't improve a CT strategy. Reason is that "more mean" can also mean "more random".