This approach that I tested was based on identifying significant odds changes just before the start of football match - and then follow the change. For example, if odds drifted, I laid a team. If the odds dropped, I backed the team etc. This was tested on 234 selections.