In the end, it doesn't actually matter if corsi or fenwick are an accurate measure of possession, because the idea of measuring possession is based on a sensible but unproven assumption that possession results in wins.
What is important for predicting future events is whether or not corsi or fenwick correlate with wins. The answer to that question right now is that yes, there is a correlation between high fenwick and corsi scores and winning percentage. It's not perfect, and there are obviously outliers, but there is a correlation, therefore there is value in the statistics.
|