With all the discussion of Adrian Gonzalez’ power outage – one homer in 102 ABs since the All Star Break – it’s no surprise that we’re seeing discussion of the Home Run Derby curse. What I haven’t seen thus far, however, is a look at whether there is statistical evidence to support the assertion that the Home Run Derby has a provably negative impact on participants’ home run rates following the contest. So I decided to check.
To save time in data gathering, I picked a single season, 2005. I picked 2005 only because it is the most frequently cited as evidence for the Home Run Derby curse; the winner, Bobby Abreu, had 16 homers at the break but hit only 6 after. The single season means, obviously, that I have a smaller sample size to work from, so the usual caveats apply. I also have made no effort to control for other variables such as games played, so bear that in mind as well.
What I’ve done here is look up the participants from MLB, then compare the players’ career pre/post All Star break splits with their numbers from the 2005 season (all splits taken from Baseball Reference). Here are those numbers:
As you can see, the differences in first and second half home run rates of 2005 compared to their career numbers is slight. For their career, participants have hit 56% of their home runs before the derby; in 2005 that number was 57%. Rather than take for granted that the one percent delta isn’t statistically significant, I ran a simple two-sample proportion test in R. In simple terms, this compares two proportions and determines whether a given proportion is equal for two different groups. The test, the results of which are included below, tells us that there is no reason to suspect that there’s a larger Home Run Derby curse at work; the difference in the observed percentages for the group is not statistically significant.
It’s possible that it affected Abreu – the result if you run the test on his numbers is just this side of significant (P-value of 0.05181), and we can’t prove that it’s not affecting A-Gon. But we don’t have any evidence to say that, in general, there is a curse.
Due to the aforementioned sample size limitations, this study shouldn’t be considered representative. But if someone tells you that Bobby Abreu is proof that there’s a curse on derby participants, you might want to point out that the effects of the “curse,” that year, were around 1% fewer home runs.
Appendix A: Test Results
2-sample test for equality of proportions with continuity correction
X-squared = 0.223, df = 1, p-value = 0.6367
alternative hypothesis: two.sided
95 percent confidence interval:
prop 1 prop 2
Appendix B: R Code for Two Sample Proportion Test
> home.run.derby rownames(home.run.derby) colnames(home.run.derby) home.run.derby
Career 1318 1052
2005 148 110