In the first 111 games of the Yankees’ season, Aaron Judge has played in 107 games, walked into the batter’s box 475 times, and taken an MLB-leading 44 nice, slow jogs around the basepaths.
That astounding production has put Judge into rarefied air, as a strong contender to finish the season with 60 or more home runs, and a chance at beating Roger Maris’ long-held AL record. If Judge can hold pace through August, the final month of the season may turn into appointment viewing as he closes in on his 61st and 62nd homers, tying and surpassing Maris, respectively.
Simulating Judge’s Home Run Production
To estimate the likelihood that Judge beats Maris and hits 62 home runs, we’ll simulate a million potential finishes to the season, and then look at the resulting distribution of performances.
The main factors in our simulation are pretty clear:
- Can he keep up this home run rate?
- Will he stay healthy?
Since both of those factors (but especially the health component) are incredibly difficult to forecast, it was clear to me that we’d want to introduce a range of potential outcomes into our simulation logic. As such, each simulated season will randomly sample the data inputs below, so that we have hundreds of thousands of projections of Judge’s performance using different potential baselines for Home Runs per Plate Appearance and Remaining Plate Appearances.
- this season’s observed data: Judge’s current season has been remarkably healthy and productive - there’s definitely a chance that he’s able to stay locked into groove, and I believe there is also a degree of a survivorship bias in play as well. Having stayed healthy so far this season implies he might be in decent shape to stay healthy for 51 more games
- career average data: it wouldn’t be crazy to see Judge’s home run rate regress towards his career average, and his career injury history (not the shortest) gives pause to the idea that he’ll play all of the remaining games -> I think it’s good to have this case as a potential “regression-to-the-mean” outcome
- projections from a variety of forecasting models: let’s be honest - there are others who have put way more work into building, testing, training, and validating models to project player performance. We’re lucky to be able to access these - so let’s just pull from public projections to see what the market thinks is most likely for Judge’s remaining games
Visualizing Judge’s Simulated Production
After running a million simulations of the data (no quick task for the little computer I run my models on!), we end up with a beautiful set of projected season finishes, estimating that Judge looks to have a 27.4% chance of surpassing Maris somewhere in the course of the 51 remaining games - an outcome that will require Judge to beat a homer-every-third-game pace. We’re rooting for you, Aaron!
