Classification for different thresholds

Betting markets offer betting lines for football matches, where you can bet over or under x offside for a team. For example, for one match they can offer U4.5 offside with odds 2.0/2.0 (lets assume there's no rake). Other matches, where for certain reasons there will be lower likelihood of offside during the match, they could offer U2.5 offside for same odds.

Hence, I want to create a model (I have the data) which can give probabilities which can be compared to the odds. If the line was the same for every match, eg U3.5, I would simply use any classifier algorithm with binary classification (0 or under 3.5 offside, else 1). However, each match can have different threshold.

The most obvious solution to me is to create a classifier as mentioned above for every possible threshold. My question is if there is any better approach to solve this?