3 ms·
The issue is that even in the case of RPS-game you should use 3 events model: A-win, draw, B-win. This implies optimization over matrix with 2 component vectors
by based_basis 3y ago
The issue is that even in the case of RPS-game you should use 3 events model: A-win, draw, B-win. This implies optimization over matrix with 2 component vectors. You can reduce the problem with some measure over this vector, for e.g. favoring wins if you have a strong team, or draws otherwise. But then it's unclear, what is the best measure for your exact case, turning optimization problem into measure study problem.