I think there are two things you can try:

Change your hyperparameters and see how this affects your model performance.

But I think more importantly generate new features. For instance how often did each team win in the previuos ten games, how is it ranked (like what place does it have in its league) and other statistics you can generate from the data you have. Be creative. You could also try to include external data like the weather, or how far away is the visting team from its hometown (less supporters).