There is a closed form solution (it's just maximising some logs, so differentiable) but I just did it in excel which is much quicker. GAT
Only the one for the first two moments was wrong ( the full distributional one was correct). And I didn't even use excel for that, just the nearest envelope. GAT
In the first puzzle I just read the peak off a graph. In the second puzzle I used solver to get a more precise answer. While this is intellectually interesting, you and I both know that in the real world you'd be lucky to know your correct Kelly to within an order of magnitude. GAT
The new Kelly formula does better than an order of magnitude in the real world. And it's perfectly safe at half-Kelly.