You should think about whether the number of matches one player wins is modeled by a Binomial random variable.
If each match is independent and each player has the same probability of winning, then this is a Binomial distribution with n matches and probability p for a fixed player.
If the probabilities change, but are independent, you need to use a probability generating function.
If the probabilities depend on what happened in the outcome of the last match, you need to use a Markov Chain model.
If it is more complex than this, you will probably need to use simulation models on a computer.