Checkdowns: Brian Burke’s “Best Games of the Decade”
Posted by Neil Paine on August 19, 2009
This post is a little behind the curve (the feature originally came out in late June), but Brian Burke of Advanced NFL Stats has a cool application using his Win Probability model that measures the best comebacks and the "most exciting" games since the 2000 season. He talks about the methodology here, and the gist is that the greatest comebacks are ones where the winning team had the lowest WP at some point in the game. The most exciting games, on the other hand, play on the idea that back-and-forth games are very exiciting --therefore the "excitement index" is a measure of how much the WP graph has moved over the course of a game. I love the concept of win probability in all sports, and Brian's work on the topic in football has been very interesting, so I think this feature is one of the coolest ones on his site.
August 19th, 2009 at 9:25 am
Am I reading this thing right? The second-most exciting game of 2008 was the Cincy/Philly 13-13 tie?
August 19th, 2009 at 10:16 am
I wonder if that's because there's a huge movement at the right side of the graph with 0:00 on the clock -- almost like it incorrectly treats it like a Cincinnati win on a walk-off score. Maybe there's a glitch with ties?
August 19th, 2009 at 10:28 am
I was researching some of his graphs a few weeks ago. He has some kind of a bug in games involving teams with zero points, and 21-point deficits.
For instance:
12/26/2004 Houston 21, Jacksonville 0
Houston gets a comeback factor of 33. With 1:08 left in the game, the graph shows that Houston's WP suddenly drops to .03. This has to be some kind of error. I am guessing that maybe the HST and JAX pbp logs must have been inverted or something.
August 19th, 2009 at 6:50 pm
Hi guys. Yes, there is still a bug with 21-point leads. There are also occasions when the NFL gamebook has a mistake and the graph goes to zero or 100% momentarily. Those would appear as big comebacks. I'm in the middle of a major upgrade to the WP model, so I'm going to wait until that's done to address some of those discrepancies.
The 13-13 tie was truly the most dynamic game last year (as defined by the biggest movement in WP). The reason was due to the closeness of the game and the high number of opportunities for each team to win it. It was also a full 15-min sudden death OT. That's a full quarter of "either team could win this at any second." If you're a fan of either team you were glued to your seat. It wasn't excellent football, but it was the very definition of a tight game.