Friday, April 16, 2010

Is OPS a Useful Statistic?

What makes a statistic useful? I'd say two factors go into it.
  1. How well does the statistic reflect the characteristics it's trying to portray? On-base percentage is a pretty good example; it does a great job of representing how frequently a player reaches base.
  2. How easy is it to derive that information? Again, let's look at on-base percentage. If a player has a career OBP of exactly .400 (like Brian Giles), then you can say that, ignoring anomalies, he was able to reach first base and avoid generating an out in 40% of his plate appearances. Simple, straight to the point.
The statistic we're talking about, OPS, is the result of adding two other statistics together: on-base percentage and slugging percentage. The idea is that, by combining those two numbers, you can determine a player's value as far as avoiding outs (OBP) and hitting for power (SLG). In theory, this kind of action makes some sense. If you're able to find a player who doesn't cause outs and gets extra-base hits with regularity, he's clearly a hitter who's going to benefit your lineup.

Which brings us to our question. Is OPS a useful statistic? The answer, for the most part, is no.

I have no qualms with the theoretical side of OPS, and I have no problem with an attempt to generate a new statistic that helps people (both insiders and outside observers) know more about players. My problem is that adding those two statistics together is the most simple and illogical thing you can do.

I offer this illustration of my point. I have two quarters and four pennies in my pocket. How useful is it for me to say I have six "moneys" in my pocket? Or should I maybe try to think of some other way to convey what I've got in my pocket (like 54 cents)? OPS just seems like a lazy stat.

Furthermore, we all learned in elementary school that you can't add fractions that have different denominators. In order to add 1/3 to 1/4, you have to convert them. The maximum possible OBP is 1.000, if you reach base in every plate appearance. The maximum possible SLG is 4.000, if you were to hit a home run in every plate appearance. OPS basically says that .4/1 + .6/4 = 1. The math itself tells you that the statistic is flawed.

Let's go back to the two criteria I cited.

First, does OPS reflect the characteristics it's trying to portray? Well, sort of. The guys with the highest OPSs are generally the best hitters, but I have a hard time saying that the order of players' ranks in OPS is indicative of how effective they were at the plate. I can't really get behind the idea that Joey Votto had a better 2009 than Mark Teixeira. And it seems impossible that J.D. Drew was a better hitter than Evan Longoria, Ryan Zimmerman, and Victor Martinez last year.

But I think it's the second question that puts OPS completely out to pasture. What exactly does OPS mean? Ryan Howard had .931...whatevers? Units of OPS? OPS percentage? I'm willing to yield a little bit on slugging percentage (which as I mentioned is out of 4.000, rather than 1.000), because it's been around forever, and because it's got a specific formula whose result makes sense (total bases/plate appearances). It should probably be called "slugging ratio" or "slugging rate," but the name won't be changed. Regardless, the actual rates are useful; they tell us who makes the most out of their PAs. Prince Fielder slugged .602 last year, which means that over ten plate appearances, he would generate six total bases (e.g. a home run and two singles, or three doubles, or a double, two walks and two singles).

I think there is a stat out there waiting to be created that can combine this information into a useful, effective number. Heck, it probably already exists as VORP or runs created, two stats I don't really appreciate because they're not on Yahoo's player pages. I just know one thing: the stat we're looking for is not OPS.

1 comment:

penwrecker said...

So i understand the concept of OPS but i think its a terrible stat as well. Reason being that a player could have an OPS of say .800 by hitting 2 triples in 10 at bats. He could have have zero RBIs/runs, sure the stat looks great but thats not contributing to the team at all....or my fantasy stats.

Top 500 Songs - Dave Matthews Band

This was always going to be the hardest of my band lists, because I like so many of DMB's songs, and have liked them so differently over...