The Home Run Derby, played the day before the All-Star game, has become a favorite event for many baseball fans, especially younger ones. Even I enjoyed it the first year or two. In recent years, however, there has been significant speculation that the players who participate in the event face the possibility of negative impact on their home run production in the second half. Allegedly, the swing typically used in the Derby, an all-out uppercut, is so different from the swing typically used in real games that some players see a second-half power dip due to their swing getting off-kilter.
Let's take a look into the stats and see if this is true...
First I'm going to list some simple data for a subset of players who have participated in the Home Run Derby dating back to its inception in 2000. I chose to look at the two finalists for each Derby, as these players each competed in 3 rounds, taking the most swings.
I have listed the winner of each derby in the table below, followed by the runner-up of each derby in the second table. I listed them separately because it was easier to do in Excel this way, but otherwise I'm treating all of these players the same for the purposes of this study.
For each player, I list their home run and plate appearance totals before the All-Star break in the given year, as well as the same two totals after the break. I then calculate the change in their HR/PA production from before the break to after. (So a positive number means their home run frequency went up, and a negative number means it went down. Keep in mind the changes are based on home run rates, not just raw number of homers.)
Home Run Derby winners:
before after change 2000 Sosa 23 394 27 311 49% 2001 LGonzlz 35 388 22 340 -28% 2002 Giambi 22 381 19 308 7% 2003 GAndrsn 22 370 7 268 -56% 2004 Tejada 15 353 19 372 20% 2005 Abreu 18 397 6 322 -59% 2006 Howard 28 352 30 352 7% 2007 VGuer 14 368 13 292 17% 2008 Morneau 14 412 9 300 -12% 2009 Fielder 22 387 24 332 27%
Home Run Derby runners-up:
before after change 2000 Griffey 28 381 12 250 -35% 2001 Sosa 29 375 35 336 35% 2002 Sosa 28 370 21 296 -6% 2003 Pujols 27 396 16 289 -19% 2004 Berkman 16 365 14 322 -1% 2005 IRodrgz 6 310 8 215 92% 2006 Wright 20 386 6 275 -58% 2007 Rios 17 389 7 322 -50% 2008 Hamiltn 21 425 11 279 -20% 2009 Cruz 22 265 11 250 -47%
A few observations:
- It appears that none of these players had an injury that caused them to miss significant playing time either before or after the break,which is great. Each guy had a normal season in terms of games played.
- The big increases in terms of HR output after the Derby were for Sosa in 2000 and 2001, Tejada in 2004, Rodriguez in 2005, Guerrero in 2007, and Fielder in 2009.
- The big decreases in terms of HR output after the Derby were for Griffey in 2000, Gonzalez in 2001, Anderson and Pujols in 2003, Abreu in 2005, Wright in 2006, Rios in 2007, Berkman and Hamilton in 2008, and Cruz in 2009.
- Overall, players had an 8.1% decrease in HR output (per plate appearance) after the All-Star break.
A bit more data relative to this last point. Here are the overall MLB numbers for homers before and after the All-Star break:
before after change 2000 3312 101879 2381 88363 -17% 2001 2983 100819 2475 86142 -3% 2002 2649 99812 2410 86794 5% 2003 3045 108472 2162 78965 -2% 2004 2826 101440 2625 87079 8% 2005 2685 100470 2332 85804 2% 2006 2942 102315 2444 85737 -1% 2007 2589 101588 2368 87010 7% 2008 2803 110115 2075 77499 5% 2009 2707 101597 2335 85463 3%
Again, that's HR and PA before and after the break, and change in HR rate. For the 10-year period, the total difference is actually zero. If you throw out 2000, where there was a large discrepancy in first-half and second-half homers, there is actually an average increase in HR frequency after the break of about 3%. All in all, this suggests that the 8% drop-off seen among the Derby participants is meaningful, as this is not what happens with the general population of MLB players. Of course, it's also true that the Derby participants are mostly outliers. These are usually guys who hit a well-above-average number of homers before the All-Star break, so the second-half performance may simply be a regression to the mean.
However, looking at the list of names more carefully reveals another pattern. Guys like Sammy Sosa and Ryan Howard have always hit a lot of homers both before and after the All-Star break. (Howard has had a few huge Septembers so far in his career.) It would appear that participation in the Derby didn't affect their swings, but that doesn't mean others were not affected by their own participation. Many of the players who saw a big drop-off in the second half were not typically among the best power hitters in the game. Abreu has topped 25 HR in a season only twice. Anderson has hit under 20 homers in 10 different full seasons. Wright averages 27 HR every 162 games. There is a suggestion, then, that the big-time power hitters are not affected by playing in the Derby but the lesser power guys are.
But is this true? Let's think again about the selection bias for the Home Run Derby. The participants are usually among the HR leaders going into the All-Star break. (The only obvious exception to this among the group of players in this study is Rodriguez in 2005.) If a player happens to have had a really hot first half and then a typical-for-him second half, it may look like he had a big drop-off in his power numbers when in reality what he had was an unusual rise in the first half of one season.
Alex Rios is a good example of this. Let's ignore 2007, the year he played in the Derby. In the rest of his career from 2004-2009, he had 34 homers in 1563 plate appearance in the first halves of each season. In the second halves 2004 to 2009 minus 2007, the numbers were 26 homers and 1233 plate appearances. That is, on average, a 3% drop from first half to second half. (If he'd had one more first-half homer or one fewer second-half homer, his HR frequency would actually have gone up overall from first half to second half, so essentially his production is level between the two halves.) His overall HR rate over these years is one HR every 46.6 plate appearances. In 2007, in the second half, he hit 7 HR in 322 plate appearances. That's one every 46.0 plate appearances, which is identical to his performance every other year of his career. In the first half on 2007, he hit 17 HR in 389 plate appearances, or one HR every 22.9 plate appearances. That performance, his first-half performance in 2007, is the aberration. That's why he was selected to the Home Run Derby in the first place, and that's why his numbers seem to have suffered after the break--but in reality all they did was go back to normal.
Obviously, the effect of participating in the Derby varies from player to player and year to year. Many other factors, including nagging minor injuries and strength of pitching faced, can affect home run numbers before and after the All-Star break. Overall, though, it seems that since the players with the biggest second-half power outages seem mainly to be the players with overall lower HR rates in their careers, I believe the effect is primarily a selection bias among the players chosen to participate in the event.
This entry was posted on Thursday, July 8th, 2010 at 7:53 am and is filed under Splits. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.