I am experienced counter, I have read in many places that spreading your action out over multiple boxes maintains ev while reducing variance, but when I simulate with CVdata I do not get that kind of result, indeed the ev goes down, thoughts as to where I might have gone wrong?
