Ok to try to win the max number of match, but it’s diffilcult to cont by hand without the ranking.
To fine tune, the ranking is needed.
The different seed can lead to different end for the match. Also there is the possibility that the other player change the code from the time the ranking was done. But in some cases, when the match was just simulated for few seconds, and you win largely, and you try with multiple seed, then you can be pretty close to be sure that there is something strange.
The situation is far better now, but there is still some issues. I’ll try to remove the for in loops, as I use them a lot.
Thanks for your effort to solve this
We discovered another bug with the random seeds that I introduced with my last fix, which would have made the seed change on each simulation. I’ve uploaded a fix for that now to make the seeds consistent for each opponent pairing, so simulations in the next couple days should grow more accurate as more clients get the new simulator fixes and start to behave deterministically again.
