So... my method of risk analysis is flawed in some way. I created a program to test 1000 bets 1000 times on each point of the decision matrix to create a wins to losses ratio. These ratios defined a matrix which doesn't really seem to reflect reality. I think my ratios are biased by the probability that you will win given any starting combination.
I'm going to check my code again and possibly rethink my strategy for testing the theoretical case.
In the meantime, I've been using my initial code for simulations to modify my original strategy. I've now tested a new strategy with 1200 iterations of 1000 bets and got a return of 1.325 (stdev 0.033). Which is statistically significant to say that it's a better strategy than the 1.3 that I had before.
I'll be permuting the optimum strategy that I have right now to try to manually find the optimum. The two processes will run in parallel and we'll see which one will win.
Synopsis: Theory Trials didn't work Found a better Strategy through permutation of previous strategy Still working.
ensignhotpants · Tue Dec 18, 2007 @ 01:20am · 0 Comments |