Let's say I have a set on n strategies and an account of size x, has anybody come up with an optimal way of allocating capital based on the performance of the individual strategies? Let us assume the account is too small to assign some very small fixed fraction to each strategy. I can think of some arbitrary schemes such as some kind of normalized allocation based on kelly, but I was wondering if anybody else had solved this problem, or if there was some kind of literature on it?
Let us try an analogy. Suppose that you have three willing and ordinarily equally satisfying female sex partners and you can get off twelve times a week (allowing the overworked member a well deserved rest on Sunday). How do you allocate your precious twelve opportunities to get off to the three women? Hint: assume they are over twelve and under sixty and therefore subject to cyclical behaviors.
trade the swing position on extremes and scalp the rotations inter day with 1/4 of it say a 60 lot position and you scalp 5 lots(total of 15) between the extremes
Eckhardt addressed this issue when somebody asked him how to weight their indicators. Paraphrasing his advice: if the indicator is worth consulting, weight it equally with the others, otherwise get rid of it.
Why doesn't modern portfolio theory apply on a strategy level? Markowitz' work seems appropriate. I mean, you want to be optimal in the sense that you want the most return for a given level of risk. So rather than treat the problem as one where it's the assets you're balancing, normalize the strategies to appear as if they are assets, then look at the efficient frontier and make your adjustments accordingly.