Many young professionals lost at coin tossing though the odds were stacked in their favour. papers.ssrn.com/sol3/papers.…
2
25
7
55
@gojomo paper is painfully wrong, Kelly is not optimal for this game. You can get the optimal solution by dynamic programming.
1
1
1
V(w,b)=max([0.6*V(min(w+x,250),b-1)+0.4*V(w-x,b-1) for x in 0..w]) for the value of having wealth w and b bets left.
3
1
Replying to @gwern @DegenRolf @gojomo
you can compute U(w,b) in O(b) if you allow bets to be any fraction of w.

Jan 22, 2017 · 6:52 PM UTC

1
Replying to @ArthurB @gwern @DegenRolf @gojomo
it's piecewise linear, the cutoff points and slopes follow a very predictable pattern.
1
more replies