July 12, 2005
Swing and a Myth: The Home Run Derby Effect
When the phrase "conventional wisdom" gets used, it tends to conjure up associations of grizzled old managers passing on the secrets of the game from the grizzled old managers they played for back in the Good Old Days, when home runs happened once a month and runner-advancing ground balls to the right side of the infield drew standing ovations from nattily-attired crowds.
Conventional wisdom isn't necessarily a turn-of-the-century relic, however, just something that gets repeated because, hey, that's what everybody else is saying. A modern example is the concept of a "Home Run Derby effect," in which sluggers who participate in the All-Star Game sideshow somehow get their timing messed up or their swing mechanics altered from all those mighty hacks, to the point that they experience a slump coming out of the break.
Is there anything to it? It would certainly be useful information in fantasy terms, so let's take a look.
The Home Run contest went to its current three-round individual format in 1996, and with 80 participants since then we've got a decent sample size even if we toss out two injury-shortened seasons, Ivan Rodriguez's 2000 and Greg Vaughn's 1996. I've calculated each player's HR/100 PA rate both pre- and post-All Star break, and broken the data into three groups: all 78 player seasons, the 36 who made it to the second round, and the 18 who made it through all three rounds.
Group Average 2 SDs Below 1 SD Below 1 SD Above 2 SDs Above All Players -0.817 0% 19.23% 10.26% 3.85% Two-Round Players -0.705 0% 16.67% 5.56% 2.78% Three-Round Players -0.705 0% 11.11% 11.11% 0%On the surface it looks like conventional wisdom might be on to something. The average Home Run Derby participant loses 0.817 HR per 100 plate appearances after the All-Star break. Before anyone gets too excited, though, take a look at that effect in real terms: The average batter in the sample saw 297 (or so) PAs after the break, which means they lost all of 2.4 home runs on average--hardly a massive downturn.
Keep in mind too that a small drop-off is to be expected, since this is in a sense the ultimate self-selecting group. Players generally get asked to take part in the Home Run Derby specifically because they hit a lot of home runs in the first half. The players who had big second halves, who would have helped balance out the sample, aren't even part of the sample. (Think of it like flipping a coin 100 by 100 times to find out whether it was loaded or not, but disregarding any set of 100 that didn't land on heads at least 30 times in the first 50 flips.)
Even more damning for convention's case, though, is the fact that none of the hitters experienced downturns two or more standard deviations below the mean. While in raw totals a number of hitters saw double-digit drop-offs (none worse than Jeromy Burnitz's 19 home-run loss in 1999) those deficits tend to be as much, if not more, a product of fewer plate appearances than an actual power drought. With every hitter keeping reasonably close to the average, it's hard to make a case that any of them had what you could call a huge slump. Injuries and a late scheduling date for the All-Star Game did far more damage to those home run totals than the Home Run Derby ever did.
We'll give conventional wisdom one more chance. Maybe the swing-altering effects are most strongly felt among the 39 first-time participants; hitters who have been through the ringer once might know how to prepare for the Home Run Derby, and avoid becoming a mechanical mess at the plate in the second half.
Group Average 2 SDs Below 1 SD Below 1 SD Above 2 SDs Above First-Timers -1.325 0% 17.95% 10.26% 2.56%While the drop-off is certainly more noticeable (working out to 3.9 fewer home runs post-break), it's hardly earth-shattering. And again, the lack of any players more than two SDs below the mean doesn't do the theory any favors. But we can't rule out the possibility that maybe a first-time Derbyer might get himself a bit out of sync by swinging for the fences for half an hour. That doesn't mean you should immediately start shopping Bobby Abreu around, of course. It just means that in this particular case, there's just a little too much noise to determine conclusively that there's no signal at all.