On Kendry Morales and Regression


Regression, or regression to the mean, is not something that is typically talked about in non-saber baseball forums and blogs, in my experience. The basic concept is that players will have times when they will play above their skill level, and because of things like a small sample size or a player have a stretch of good luck, their numbers will be better than their actual skill level. As that sample size increases, however, the luck will run out or the stretch of playing over their heads will end, and their numbers will start to trend back towards their actual skill level. The “mean” is their skill level, and the “regression” is their stats falling back in line with that skill level.

We can find plenty of examples of this with even a short glace around baseball. In 2007, Joe Saunders put up a 3.41 ERA, despite never before or since posting anything better than a 4.44 ERA. Looking even a little deeper into his numbers, we discover he had an obscenely low .267 BABIP (Batting Average on Balls In Play, which measures how many times a ball put in play will go for a hit), which is a full 25 points lower than his career BABIP, and certainly helped with his lower numbers. Simply put, pitches of his that were put into play weren’t finding holes then like they are now. Darin Erstad is another fine example. In 2000, he hit .355/.409/.541, which is a really good season for a lot of people, but was clearly an out-of-his-mind season for a career .282/.336/.407 hitter. The very next season, his OPS dropped an absurd 260 points, and in the nine seasons of baseball he’s played since 2000, he’s never come even remotely close to repeating those numbers.

So, we turn our attention to Kendry Morales, a player that showed us the text book definition of bursting onto the scene last year. Before last year, Kendry had given the Angels 407 Plate Appearances and a slash line of .249/.302/.408, which wasn’t awful, but for a major league 1Bman, it wasn’t exactly acceptable either. After trading away Casey Kotchman for Mark Teixeira, and then failing to resign Teixeira during the off-season, the starting gig at 1B was Kendry’s to lose. Blowing away anything he’d done previously, he put up a .306/.355/.569 slash line in 622 PA, and was a big reason why the Angels’ offense last season wasn’t completely terrible (a career year from Torii Hunter, another candidate for regression, certainly didn’t hurt either).

He is clearly, and without a doubt the starting 1B for the Angels going into 2010, and rightfully so. Not only is there anyone lurking around likely to replace his production, but there’s no reason to remove a guy that had a season like he did in 2009. The question, then, becomes what we can expect for him going forward. Personally, and this is an opinion that seems to be shared by almost all of the projection systems, I think regression is quite likely. Not many players put up numbers so significantly better than anything they’ve previously done at any level, and never look back. For its part, CHONE is expecting fairly significant regression from Kendry, giving him a .286/.331/.495 slash line with a drop down to 22 HRs. I’m slightly more optimistic, though I wouldn’t be shocked if he didn’t break 30 HR or much over a .330 OBP.

We, as fans, need to expect regression to happen, and not be too quick to call for player’s to be removed or assign blame for it to whatever random thing the Baseball Tonight people are telling us a player is doing (“It seems like he’s just pressing too much, maybe he needs to look at more tape, work on his swing with the hitting coach, because this kid is a great player and blah blah blah.”). Regression is a fact of life, and one we as Angel fans are going to need to prepare ourselves for moving into 2010.

(Nate Proctor is the lead writer for Halo Hangout.  You can stay up to date on all of Nate’s work by following him on TwitterFacebook, or by way of the Halo Hangout RSS feed.)

Tags: Baseball Tonight Darin Erstad Joe Saunders Kendry Morales Regression Regression To The Mean