The Home Run Derby and Second Half Slumping

The Home Run Derby took place very recently. Pete Alonso won. But this post will not really be about this year’s Home Run Derby. Rather, it will be about the wake of the Home Run Derby for those who participated in recent years.

Due to back issues, Christian Yelich had to pull out of this year’s competition before it started. Before those issues though, he shrugged off the popular suggestion that the Home Run Derby can lead to prolonged swing tweaks, a reversal of fortune, and second half slumping. Specifically, he pointed to Bryce Harper’s campaign from a year ago, in which Harper won the whole competition, and subsequently played considerably better in the second half.

In response to the seemingly annual discussion that takes place around this issue, this post aims to address the question of whether or not players do in fact experience second half slumping following HRD participation. The short answer: yes. The less short answer: it’s complicated.

First, here is compiled data for all players who participated in the Home Run Derby over the last five complete seasons (2014-2018). Their first and second half wRC+ are provided in separate columns. The annual winners are highlighted. Note too that some player data is repeating given multiple years of participation.

This chart is pretty interesting. As we might expect, by wRC+, participants are enjoying strong seasons. The “worst” offensive player-season listed is that of Todd Frazier from 2016, when he participated after just a 109 wRC+ first half. That season is somewhat of an outlier though, given that the median and average wRC+ of participants are both in the upper 130s.  

Lo and behold though, these players are, on average, worse offensively in the second half. One notable exception (as Christian Yelich effectively cherry picked) is Bryce Harper’s 2018 season, but it stands at the far end of the spectrum. This drop-off fits the narrative around HRD participation, but should the slip be attributed primarily to participation? I don’t think so.

Bias has been covered here before. Bias is one part of what feeds our perception of the Home Run Derby. In this case, weirdly few people make selection bias part of the discussion. Very simply, those Home Run Derby participants are part of a biased selection. They represent a selection of players performing very well, with prodigious power as a proportionately large part of their game. So the population of participants is, by definition, a subset of players having atypical seasons in relation to the wider league.

The other half of this is regression to the mean. When examining a small group of players off to (by default given their HRD selection) very strong starts, it only makes sense that they should generally come back down to earth in the second half.  This regression is likely to take place for any given player to some extent regardless of their Derby participation. I have not compared Home Run Derby participants to other All Stars, but it would probably be a interesting and worthwhile exercise.

This next chart presents the same set of players, but featuring their career first and second half splits. 

This chart clears things up further. Dating back to the last 5 years’ worth of contestants, HRD players, on average, hit worse in the second half throughout their careers as a whole, not simply just the year of their participation. So in addition to performance regression, at least some of the second-half drop off can be (more recently at least) attributed to the players themselves as opposed to wholly the Home Run Derby.

This third chart aligns the seasonal and career splits against one another, and also includes the difference between those two stats.

The fourth column takes the difference between the second and third, or the difference in first & second half wRC+ differences. As a result it produces the discrepancy between each player’s HRD splits and their career splits. We can see that the “difference in those differences”, which very generally controls for each player’s career-wide trends, cuts down on the perceived impact of in-season HRD splits.

The figure below plots the seasonal and career half-splits, or the second and third columns of the chart above. The upper-right and lower-left quadrants are most populated; players in those regions had career first/second half splits that align (both positive or both negative) with their Home Run Derby season first/second half splits. As an example, a point that represents a player in the upper-right hand quadrant indicates that that player has played better in the second half throughout his career as well as in the second half of the season in which he was an ASG participant.

While it’s premature to say that the Home Run Derby has no negative impact on a participant’s second half performance, several meaningful confounding factors have been covered that the Home Run Derby often takes the blame for. The first is very simply performance regression for players off to hot starts over a relatively brief sample. The second is an existing trend for recent participants to be more first half oriented players regardless of the Derby.

For rookies like Pete Alonso and Vladimir Guerrero Jr., there is no MLB second half data. But if either get off to slower starts in the second half, in a similar vein of Aaron Judge’s 2017 second half start, it might be worth hesitating before pointing at the Home Run Derby as being wholly or even primarily responsible. Because baseball is hard, and coming back down to earth is an inevitability.