Spring Training Stats Tell Us Almost Nothing

Last night I was poking around at baseball-reference.com, looking at Brandon Belt, for whom I hold a lot of hope.  I guess they’ve been there for a while, but it was on his page that I first noticed 2013 spring-training stats as part of the mix of data. I can’t say I cared all that much at first; spring-training stats are almost useless and, in many cases, hinder our understanding of players and teams.

At the end of the stat line was a number I’d never seen before: “OppQual”. Rolling my mouse over the abbreviation, I learned that Sean Forman had yet again found a way to make my life better. “OppQual” is a 1-10 rating of the quality of opposition a player has faced in the spring. Every player gets a rating based on where they played in 2012, weighted by playing time at each level. “OppQual” averages those ratings to give you an idea of what type of opponent a player has seen in the Cactus and Grapefruit Leagues. Variance of opposition is one of the biggest problems with spring-training stats, because there are so many sub-MLB performers getting playing time. Forman’s method, while not perfect — I’d want to know who the player was having his success against — gets you close enough to be useful.

Here’s the scale:

10: MLB

8: AAA

7: AA

5: High A (California, Carolina)

4: Other full-season A

1.5 to 3: Rookie, short-season

1: Opposing hitter is a pitcher

Using Belt as an example, his 63 plate appearances have seen him face an OppQual of 9.2 — indicating he’s faced a mix of major-league and Triple-A pitchers.

That’s what the number serves as, a go-to check when you hear about some player, like Belt, having a big spring. Alex Gordon is going nuts in Surprise? 9.4, one of the highest marks for hitters with significant playing time. Jackie Bradley Jr. should make the Red Sox? 8.2 — we already have a pretty good idea he could hit Triple-A pitching. How are the Nationals going to make room for Anthony Rendon? At a leisurely pace; his 8.4 OppQual score means his 1.287 OPS has come largely against minor-league hurlers.

The same idea works for pitchers. Allen Webster has looked really good for Boston, with a 1.64 ERA and a 14/1 K/BB…but against an 8.2 OppQual, again, Triple-A. There are people trying to elevate Michael Wacha off of 11 2/3 shutout innings…against an 8.5, closer to Triple-A than anything else. (I actually like Wacha long term as a mid-rotation starter, but he’s a year, at least, from that.) The phenomenon of overrating players off of spring training stats is worse with young pitchers because they’ll often getting their work in the late innings of early spring-training games.

The mistake would be to take this new metric too far. Just because we’ve roughly quantified opposition quality doesn’t mean you can normalize spring-training stats by that number and decide they’re legitimate. They’re still not. They’re all, every one of them, too small a sample on which to reach conclusions. Even the OppQual number doesn’t tell you if the player is hitting good pitchers. Even if the player is doing so, those good pitchers didn’t go to the mound that day trying to win; they went to the mound trying to prepare.

Performance analysis works because the statistics are created in games where both teams are trying to win. Take that away, and the statistics become meaningless very quickly. The OppQual number is useful not because it’s a key towards divining meaning from spring-training stats; it’s useful because it’s a tool to show why you cannot.

This entry was posted in Uncategorized and tagged , , . Bookmark the permalink.