A related, but different project is my own nflgame [1]. Related in that it provides play-by-play data, but different in that it can be used to get it in real time. Also different in that the data only goes back to 2009, so it can't really be used to do the cool kinds of statistical analysis being done in the OP.<p>The source is JSON data from NFL's GameCenter pages. It contains its fair share of errors, but it is at least structured.<p>[1] - <a href="https://github.com/BurntSushi/nflgame" rel="nofollow">https://github.com/BurntSushi/nflgame</a>