[grayc]

That looks nice. You could improve things a lot by using even a simple model to account for opponent quality, home field advantage, etc.: a simple fixed-effects regression model with team-level effects would be one approach and would take only a few lines of R code.

For an example, look at: http://www.pseudotrue.com/details which is my blog, so this is a pretty shameless plug but I don't really know of a different reference.

The point of using model isn't because you believe the model is true or expect it to necessarily be accurate, but because it can give you a better measure than just looking at (say) yards per game.

But your site is cool as is, so good job.

edit: tweaked URL to point to the right page.