>> We have lost a number of high-profile games, where the predictive models suggested a 99% probability of winning near the end. If it's any consolation, I don

>> We have lost a number of high-profile games, where the predictive models suggested a 99% probability of winning near the end. If it's any consolation, I don't believe the Falcons were ever truly at 99%+ win probability, in the Super Bowl or in those blown leads this regular season. Most WP models, in my opinion, highly over/understate the true WP as they approach the extremes. Obviously ATL had a very high P(winning), but perhaps it was 96-97% rather than 99+. Seems like a minor difference, but from the NE side, if the Patriots truly had a 3.5% chance instead of a reported 0.5%, that's a 7x improvement. I've done some analysis on these models and written about it here: https://cfnine.com/blog/nfl-win-probability-part-1 https://cfnine.com/blog/nfl-win-probability-part-1