One of the first things that a talent evaluator will notice about potential quarterback prospect is their size. This not only includes height, but also hand size and arm length. This got me to thinking about how much each measurement actually correlates to future NFL success. I decided to get some data and see what I could find. The metric I chose to correlate these numbers to is win probability added per game.
Height comes in at an R^2 value of 0.0052. This is an extremely weak correlation and can realistically be assumed to be at about 0. This effectively means that 0.5% of the variation in WPA/G can be explained by the variation in height.
Hand size has another low R^2 value, 0.0372. Better than height, but still very low.
Last but also least, arm length has an R^2 value of 0.0030.
When looking at these graphs, it makes you wonder why NFL teams even bother to look at these numbers. Just to show the contrast between these numbers and useful ones, this is the WPA/G graph for my adjusted college completion percentage metric:
Adjusted completion percent has a decent R^2 value of 0.2352, meaning that 23.52% of the variation in WPA/G can be explained by the variation in adjusted completion percentage. For only using one aspect of a quarterback, adjusted completion percent shows a very solid correlation between NFL and college success, especially compared to all of these physical metrics.