2019 World Series Pitcher Matchups
Last year I posted some world series projections.
The underlying model was fairly simple: learn a parameter for all 30 teams representing the team’s ability to win from the outcomes of each series during the regular season.
This year I thought it would be interesting to do something similar, but instead build the model at the player level. Conceptually this year’s model is very similar to last’s: learn a parameter for all pitchers representing the player’s ability to get an out given the batter’s on base percentage.
I plan on writing a post at the end of the season using this model to rank trades or player’s consistency but for now I have matchups for each of the first six scheduled starting pitchers against the opposing team’s position players.
The plotted position of each batter represents the probability the pitcher they are matched up against on the left will get an out from them. If you hover over the batter’s face you can see upper and lower bounds on that probability estimated from the model.
You can see that according to the model the National’s are outmatched in the first starts. However, their rotation so far has been dominant this post season so it should be a good series.
<!DOCTYPE html>
You can find the code that built the model and generated this plot on my github linked below.
** Note that Andrew Stevenson leading the Nationals offense is a bit of an anomaly due to his 30 at-bats this season.