I understand how the Nash tables work with short stack situations. Ie the best thing for SB to do is shove X and the best reactive thing for BB to do is call with Y. However, how are these tables generated? Do programs like holdemresources just play random hand ranges a billion times and then sort out how well hand ranges do? I'm a bit confused about the algorithm they use to generate these. Any explanation would be very welcome.
There are a few algorithms used. The 2 I've heard of are called "counter factual regret" and "fictitious play". They are run iteratively. You start with a table and calculate it's EV. You adjust and calculate again. You keep repeating until the result is "close enough" to optimal. The magic sauce for each algorithm is the means it uses to do the updating between runs.
