Last year, on my personal blog, I posted about a model I created for evaluating baseball teams and predicting games. I’ve re-run this model for the 2017 season, so let’s take a look at the results! Be sure to check back throughout the playoffs, as I’ll use this model to predict games and series throughout the playoffs. I’ve already posted wild card PML predictions.
Let’s start with a very brief overview of the method. The model, which I originally called the Poisson Maximum Likelihood (PML) Model, assigns scores to each team’s offense and each starting pitcher. The idea, roughly, is that multiplying an offense’s score by a starting pitcher’s score gives the average number of runs we’d expect that offense to score in a game started by that pitcher. As you can imagine, lower scores are better for pitchers (they mean more runs!) while the opposite is true for offenses. These scores can be used to evaluate the performances of offenses and starters throughout the season. For the purposes of this article, I’ll leave the explanation at that, but I’ve posted the original methodology article on Model 284 here.
Overview of 2017 Scores
Let’s take a look at some scores! The table below gives each team’s offense score for the 2017 season. It should come as no surprise that the Astros are head and shoulders above all other teams. They led the MLB in average, OBP, and slugging percentage, posting a wRC+ of 121. The next closest team, the Yankees, ended with a wRC+ of 108. The other high-scoring teams in PML—New York, Arizona, and the Coors-assisted Rockies—come as no surprise either.
Next, let’s take a look at pitchers. The following table gives the ten best starting pitcher scores. Note that this doesn’t necessarily mean these were the best pitchers; rather, if a pitcher is on this list, it means that the combination of him, his defense, and his bullpen limited opponent runs exceptionally well in 2017. See the full methodology for details.
Starting Pitcher Scores
For more PML, keep an eye out for playoff posts and take a look at the methodology article!