March 18, 2008
It Doesn't Count
Here's a look at a couple of leaderboards:
Home Runs Prince Fielder 8 Ivan Rodriguez 6 Victor Martinez 8 Andre Ethier 5 Gary Sheffield 7 Carlos Guillen 5 Matt Stairs 6 Grady Sizemore 5 Alex Rodriguez 6 Victor Diaz 6
One of those is from spring training, the other from a random stretch of days in the 2007 season. Can you tell the two apart? Yeah, you probably can, since Ivan Rodriguez hit 11 homers all of last year. Let's try another one:
Batting Average Dmitri Young .500 Mike Morse .526 Ruben Gotay .450 Elliot Johnson .517 Dustin Pedroia .436 Alex Rodriguez .500 David Eckstein .432 Craig Counsell .486 Brian Roberts .429 Torii Hunter .483 Casey Kotchman .426 Jesus Merchan .483 Orlando Cabrera .414 Aaron Hill .480 Albert Pujols .403 Yunel Escobar .472 Yunel Escobar .400 Ronnie Belliard .471 Corey Hart .395 Placido Polanco .465
Both of these seem to come from a league in which infielders rule the world. Can you tell which is which? I'm sure the presence of guys no one has ever heard of on the second list helps. One more, just for fun:
ERA Randy Messenger 0.00 Burke Badenhop 0.00 Chad Cordero 0.00 Denny Bautista 0.00 Heath Bell 0.00 Philip Humber 0.00 Brian Tallet 0.61 Hideki Okajima 0.00 Chris Young 0.65 Chan Ho Park 0.00 Sergio Mitre 0.67 Chris Narveson 0.57 Matt Guerrier 0.77 Carlos Zambrano 0.60 C.J. Wilson 0.82 Manny Parra 0.64 Kevin Gregg 0.84 Jorge Sosa 0.75 Billy Traber 0.87 Darren O'Day 0.79
You've probably figured out where I'm going with this. I've been making the point about the meaninglessness of spring training stats all month, and I'm making it again, because I continue to see these stats quoted in stories about baseball as if they have meaning. They don't. Spring training stats, save for that weird "slugging 200 points above your career mark" thing, are meaningless.
The lists above serve to illustrated just one of the problems, the sample size issue. If Victor Diaz can pop six homers in two weeks, or Ruben Gotay can hit .436, or Sergio Mitre can post a sub-1.00 ERA, then how seriously can you take 50 at-bats or 15 innings? That's what the exhibition season is: small samples blown up to have meaning because they're ripped completely out of context. A player's performance in the tiny sample sizes of March is no more significant than it would be in a tiny sample size of June (or, as above, May 17 through June 5 of 2007; thanks to William Burke for the help), but because the industry has it in their heads that March is meaningful, these numbers, these performances, are treated as important.
The sample size is just the most obvious problem. Almost as big a problem is that teams aren't actually trying to win in the Cactus or Grapefruit Leagues, no matter what they may tell you. The framework of performance analysis works because the statistics we use are generated in an effort to win baseball games. Because of this, there are relationships among events that are reliable. In the spring standings in Arizona or Florida, however, winning is at best a secondary goal. The primary goal is preparing players for the season, and that effort is largely an individual one. What I mean by that is that one guy might need to play a lot to get his swing in order, while another may need to play very little while nursing an injury that wouldn't keep him out of the lineup in August. One pitcher might be working on his mechanics, and another his new changeup, and if in that effort they both give up 17 runs in nine innings, well, so what? The games don't count.
It's because of this that you simply cannot use statistics to measure performance, because once you divorce the end goal of winning from the statistical lines, they completely lose meaning.
Another problem, tied to both of the above, is variability in competition. Up until the last week of spring training, there are many, many minor leaguers getting playing time in the games. Most teams travel to road games with just four starters, and maybe that many major league pitchers. No one is seeing a normal distribution of opposition talent, and just on the luck of the draw, some guys might face more Double-A pitchers than MLB ones. As far as I can tell, this is a bigger factor with pitchers, who might run out nine shutout innings while facing a series of minor leaguers wearing MLB colors, and then find themselves in the Opening Day rotation.
All of these issues also affect scouting, observational evidence, and skills analysis. As much as we think that the teams are looking past the stats to make their evaluations, in reality, they get fooled by this stuff. Guys don't make teams or win jobs by hitting .200 or posting an 9.00 ERA. The surprises that pop up in the spring track the stat lines, and the guys who hit .460 are the ones we hear about. You know, the ones who worked in the cage all winter, revamped their swing, had laser eye surgery, rededicated themselves to the game… these storylines are older than BP, and they're nearly as old as baseball itself. They're all post facto rationalizations, though, and they're based on small-sample stats compiled in games where the outcome doesn't matter against a mix of major leaguers and others.
Spring training stats are meaningless. It's the single most important thing to keep in mind every March.