
This is something I put together a couple years ago, and I've dusted it off for the tournament. The model's approach is very similar to my football model. Basketball is a much simpler sport, though. There's no field position, down, or distance in basketball. Aside from possession, score and time remaining are really the only significant statistical factors.
It seems to fluctuate way more than football, I am assuming its built the same way as football, actual historical percentages?
ReplyDeleteYes. I have a post somewhere about how they were built, but can't seem to find it. The only real thin spot in the model is in the endgame--the last 60 seconds or so. The play-by-play doesn't have data on who has possession or whether teams are in the bonus, etc.
ReplyDeleteThe last two minutes on the graphs could be someones heartbeat at these games, got to love March Madness!!
ReplyDeleteThinking out loud here - I'm wondering if there's possibly for a momentum effect in college BB?
ReplyDeleteIn addition to the to the possession, score and time remaining, could the score from 1 minute prior also be statistically significant?
For another look at college basketball win probability see
ReplyDeletehttp://kenpom.com/blog/index.php/weblog/in-game_win_probabilities/
Les