Consistently among that illustrious group is the Home Run Derby, a creatine-inject remnant of the steroid era that provides an excuse for Chris Berman to leave the golf course once between the Super Bowl and start of preseason football.
While this year’s edition did have some drama and entertainment (I’m shocked too), it also continues the trend of having little to do with the actual game of baseball. It’s no surprise that the American League had both finalists since the contest is in step with the designated hitter, Arena baseball that is so revered in the junior circuit.
Once again, the major league leader in home runs entering the contest, Jose Bautista, left without the trophy as the league-leader has in each of the previous 24 contests. Failure in the derby of those who succeed during the season is well documented. What’s less clear is the success after the Derby of those that participate and whether or not the night of swinging for the fences has any effect.
Over the last 10 years (80 Derby participants from 2001-2010), 51 (63 percent) posted lower home run totals per game played in the second half of the season. On average, those 51 players decreased their home-run outputs by 0.0758 dingers per game. For 2011 champion Robinson Cano, whose Yankees have 88 games remaining, that would translate to 5.6 fewer home runs after the break. For the 29 players who increased their homer frequency after the all-star break, the rise was slightly less at 0.0467 HR/GP.
It’s important to note that both values are not statistically significant according to a paired T-test (this was done in Excel and I’m pretty rusty from freshman stats class, so clarification is welcomed).
My analysis is rather simplistic. But the one often cited in the attempt to debunk this phenomenon comes from Derek Carty. By comparing the Derby participants from the last nine years to a control group comprised of, in Carty’s words, “similar players who could have participated in the Derby but, for whatever reason, did not,” he yields similar differences in pre- and post-all-star-game home run totals.
So then what else could cause this difference? Several factors are likely to contribute.
The first is the idea of anecdotal evidence, the posterchild of which is Bobby Abreu, who in 2005 won the Derby by popping a then-single-round record 24 and a still-record 41 total bombs and then forgot how to hit home runs altogether, hitting just six the rest of the season. He’s certainly not the only one to lose his power touch so drastically – he’s not even the most drastic drop-off of that year, as Hee-Seop Choi went from 13 homers in 78 games (0.167 HR/GP) to two in 55 games (0.0363). Abreu’s case is in no way representative of the many sluggers whose numbers have improved post-Derby, though they are the minority. But tales like Abreu’s have a prejudicial effect that, as Carty points out, can snowball.
There are also external factors at play, which reduce the sample size below what we can deemed reliable. Prior to 2006, how many results can we use? And how much of the effect realized, one way or the other, is attributable to performance enhancement? For instance, Barry Bonds participated in the Derby three times during this period; his numbers increased once afterward and dipped twice, though both decreases were less than 0.015 HR/GP, a difference of about one homer over 70 games. Alex Rodriguez, a notoriously slow-starter, improved his homer total after both Derbies in which he took part. Of the nine Derbies contested by confirmed dopers Rafael Palmeiro, Sammy Sosa, Gary Sheffield, Jason Giambi and Miguel Tejada, six led to improved second-half numbers. The sample size is too small to provide statistical validity, but the prevalence of drugs in the sport brings more than reasonable doubt to everything.
The other important factor is selection bias: The players participating in the Home Run Derby may have been chosen precisely because their first half numbers don’t resemble anything close to their career numbers. A second-half drop off could be merely normalization of an abnormally prolific first half with a commensurately stingy second upon which a place in the Derby, a three-night bender in Cabo or a quiet weekend at home with the family would have no effect. To fully tease out these effects, the second-half numbers would have to be compared against an index of the player’s career numbers, adjusting for the trajectories of similar players based on age, experience and position that I simply lack the knowledge to do. What I looked at relies on the assumption that the players before the Derby are an accurate control group for the after to be compared to, which we can’t say with certainty.
There is however something that gives me pause. Extending the stats out another half season, 28 of the 51 players who declined from the first half to the second half continued that fall the next season. There are myriad reasons why this is the case. For instance, the selection bias could may not be for players whose numbers are in an upward or at least level trajectory. In fact, they could be big names picked for their value on the marquee regardless of their performance during the season. It could be that these players were on a downward tick when they were selected for the Derby and could have continued that regardless. Or the Derby could have signaled some type of sea change in their approach at the plate or altered their confidence in a way they never recovered from.
There’s also the possibility of injury incurred as a direct or indirect result of the Derby. You’ll notice Jim Edmonds name on the list only once; that’s because an injury incurred in the 2003 Derby severely limited him most of the regular season and made sure he’d never participate again.
Obviously, the anecdotal nature of such data and the uniqueness of each player make the effect difficult to ascertain. At least it deserves a bit more of a look than it has been given.