Analyzing In-Game Win Probability – Numbers Game 88
Episode 88 (January 25th, 2017): The eighty-eighth episode of The Numbers Game (@NumbersGamePod), a sports analytics RotoViz Radio podcast.
If you’re a listener of the show who hasn’t subscribed yet to RotoViz, you can get a special 30% discount through the podcast.
Host: Nick Giffen (@RotoDoc) Editor at RotoViz, Top-Five NASCAR DFS player, and Math Ph.D
Host: Josh Hermsmeyer (@friscojosh) Contributor at RotoViz
HOST DISCUSSIONS: Analyzing In-Game Win Probability, Sticky Team Stats
Nick and Josh talk about in-game win probability and whether they think it’s not flexible enough, or if it’s too flexible and should be less variable early in the game. They liken the probability to election forecasts such as at fivethirtyeight.com, and why an approach by Nassim Nicholas Taleb that updates more slowly makes more sense for both election forecasts and NFL in-game forecasts. Next, they talk about Josh’s work describing which NFL team stats are sticky, and which regress to the mean, and what the fantasy implications are for forecasting, especially at the RB position.
Our theme music is by Black Ant.