MVN - a statistical and sabermetric baseball blog
Statistically Speaking
The mystery of Troy Percival
While doing a wee bit of research for another piece that I was writing, I came across a small mystery in Troy Percival’s stat lines. Open that up for a moment and take a look at his BABIP column over all the years that he’s been pitching. (Prefer a graph?) The numbers struck me as a little strange. They’re almost all below .300. (The one and only exception is 1997, when he had a BABIP of .313.)
BABIP is generally considered to be a matter of random fluctuation around a league mean, and the league mean is usually .300. There’s a small mountain of evidence that says that there is very little repeatable skill in a pitcher’s performance on this particular stat from year to year, no matter how you slice it. It looks like there’s a tiny bit, but randomness is far and away the bigger piece of what’s going on.
Then there’s Troy Percival. From 1995 to 2004, he generally threw a little more than 50 innings a year and faced between 200-250 batters. Again, only once was his BABIP above .300. In 2005, he started giving up homeruns like crazy for the Tigers and soon was out of baseball… for a year. His BABIP in 2005, however, was .192. Then, last year, he caught on with the Cardinals and posted a .215 BABIP in 40 IP. This year, he signed a contract with the Rays and has put up a mere .152 Assuming that BABIP fluctuates completely randomly and that I’m as likely to be above the mean as I am below the mean (a 50/50 shot), the chances of Percival being below the mean in 12 out of 13 seasons is about 0.17%, round it to 1 in 600. Not astronomical odds, but certainly not odds I’d want to stake my life to.
Now, given that the BABIP reliability numbers that we have aren’t zero (just very very low), perhaps there can be a case made that Percival has a particular skill that we haven’t yet appreciated and that he’s not that lucky. Still, there’s something a little mysterious about his statlines.






4 Responses to “The mystery of Troy Percival”
May 31st, 2008 at 12:12 pm
He has played on some pretty good defensive teams. Weighted for IP each year, the average defensive efficiency for the teams he’s pitched for is .705.
Obviously doesn’t get you all the way to .245, but it’s something.
May 31st, 2008 at 12:46 pm
Elite relievers actually do have an impact on BABIP. See this article (go to the end), for one piece of evidence:
http://www.hardballtimes.com/main/article/pictures-of-batted-balls/
June 1st, 2008 at 12:42 pm
just like a batter with great speed or great bat speed can have higher than expected BABIPs, a pitcher with late movement on a fastball or a deceptive cutter will have lower than expected BABIPs. That is because batters can make contact but they often miss the sweet spot.
Mo Rivera is a great example. Nobody breaks more bats than Rivera. Is he lucky in that regard? No. He has a cutter with very late break. Batters make contact often, but rarely well.
June 2nd, 2008 at 7:01 pm
[…] are hitting just .135 against him with a .191 BABIP. Pizza Cutter discussed Troy Percival’s knack for posting low opponent BABIPs and Dave Studeman commented that it isn’t that unlikely for elite relievers to follow suit. […]
Leave a comment