At the Dutch football site 11tegen11 (11 on 11) there is an extended discussion on possession analysis in football and its evolution over the years.  The motivating principle is a simple one — if you don't have the ball you can't score, to paraphrase Johan Cruyff.  But assessing the likelihood of scoring goals and winning matches from possession is not an easy problem to solve.  Initial analysis looked at possession time, but it's become clear from looking at the data that there is little correlation between time on the ball and match outcome.  The alternative process that the author defines is a good one and a step in the right direction, but in the end I believe a more finely-grained data analysis is required. 

The discussion in the comments is also worth reading; there are some perceptive comments and papers referenced that I've read before (and reviewed on this site).