I was parsing bookmakers that provide historical odds and match results, based on the results I built a model that determined the total number of goals scored in a match with a fairly high accuracy. The model does not promote any logic, its work is mainly related to statistics, and considering that I collected data from all the top championships for almost five years, I managed to come up with a pretty good result of 68% of winnings. But the pattern is always unstable when the season is just starting and the teams are not yet in optimal shape.
Actually, my advice is that you can do it this way and improve the functionality, for example, predict matches in which both teams will score. Good luck!