Forecasting football matches by predicting match statistics

This paper considers the use of observed and predicted match statistics as inputs to forecasts for outcomes football matches. It is shown that, were it possible know in advance, highly informative outcome could be made. Whilst, practice, are clearly never available prior match, this leads a simple philosophy. If can pre-match, if those predictions accurate enough, follows that Two approaches prediction demonstrated: Generalised Attacking Performance (GAP) ratings set based on Bivariate Poisson model which named (BA) ratings. both provide suitable methodology predicting advance they enough information beyond reflected odds. A long term robust gambling profit demonstrated when combined with two betting strategies.