OPS Should Only Be Used For Convenience
February 1st, 2010 | Published in Daniel Gettinger, Kevin Kouzmanoff, Sabermetrics, San Diego Padres, baseball, links | 8 Comments
by Daniel Gettinger
From an offensive perspective, players who get on base are good. Players who hit for power are too. Players who both reach base frequently, and hit for power are really good.
Because a good measure of getting on base is on-base percentage (OBP), and a decent measure of power is slugging percentage (SLG), it has become somewhat mainstream to evaluate offensive performance by combining the two measures into OPS (OBP+SLG).
Unfortunately, adding SLG to OBP, while easy to calculate, is not an entirely accurate measure of offensive performance. The main reason is OPS treats OBP and SLG as equals that are scaled in the same manner. In fact, OBP is actually more important than SLG. Given two players with the same OPS, the one with the higher OBP is better offensively.
Inside the Book Blog recently published an excerpt from PAAPFLY that illustrates the issue:
Continuing my bashing of Bengie Molina, allow me to show you how his terrible OBP can be quite detrimental. Bengie Molina posted a .727 OPS in 2009, which isn’t very good. Ryan Theriot managed to post an even lower OPS of .712 in 2009. He must be the inferior offensive player. Wrong. Molina’s wOBA is actually .308 to Theriot’s .318. Though Theriot slugged 73 points less than Molina, his OBP was 58 points higher, and, wOBA shows us that his 58 OBP points to Molina’s 73 slugging points were actually worth an additional 10 points in wOBA. This is just a quick example and a good way to illustrate just how much Molina’s extraordinary out making skills truly do hurt his team, offensively of course.
From a Padres’ perspective, I liken the Molina/Theriot example to that of Chase Headley and Kevin Kouzmanoff. This is how Kouzmanoff and Headley stacked up in 2009:
| OBP | SLG | OPS | wOBA | |
| Kouzmanoff | 0.302 | 0.420 | 0.722 | 0.312 |
| Headley | 0.342 | 0.392 | 0.734 | 0.328 |
Although not a perfect example (because Headley’s OPS was slightly higher than Kouzmanoff’s OPS), we can see that Headley’s wOBA is 5% greater than Kouzmanoff’s wOBA, compared to only a 1.6% difference in OPS. The reason is Headley’s superior on base skills.
OPS is an okay estimate of offensive skill. But the only advantage it has over a stat like wOBA, which properly weights OBP and OPS, is ease of calculability. It would be nearly impossible to calculate a player’s wOBA based on the information given by the stadium scoreboard, or the Channel 4 stat-line. However, outside of convenience, there is no good reason to use OPS.
February 2nd, 2010 at 12:15 pm (#)
Is there any such stat as a “positive result average”? This would include hits, walks, sacrifices, advancing runners, HBP, any result other than an out that doesn’t advance a runner.
February 2nd, 2010 at 8:13 pm (#)
Mark-I do not know of such a stat (although I could be forgetting something). Although somewhat interesting, I am not sure what purpose ignoring outs would have.
February 3rd, 2010 at 9:46 am (#)
Good points, Daniel, although if something is going to be used in the mainstream, I’m glad it is OPS and not BA (which it was when I first started watching baseball). Despite its limitations, OPS is a pretty good back-of-the-envelope estimator of offensive utility.
You and your readers may find these articles on the topic interesting:
Run Estimation for the Masses (THT)
The great run estimator shootout (part 1) (THT)
Baseball Prospectus Basics: About EqA
February 3rd, 2010 at 10:35 am (#)
Geoff-Very true, for all of my complaining, OPS is vastly superior to stats like BA and RBI. Also, thanks for the links-all great articles!
February 3rd, 2010 at 10:42 am (#)
The weighting would be tricky. Any stat that gives equal weight to a HR and a slow roller to the right side with a man on 2b wouldn’t be very informative.
Most positive events are already accounted for in other measures. Sacrificing doesn’t count as an AB, HBP counts for your OBP, etc. The only thing missing are the outs in which a runner advances but the hitter didn’t square around, and those may not be common enough to be statistically meaningful.
There’d also be plenty of noise. Man on first, batter takes a full cut and the ball bounces thirty feet in the air. Batter out at first, runner advances to second. Giving the batter credit for a positive result in those cases would be like giving a student a passing grade because he spelled his name wrong at the top of the test paper.
It sounds a little like Win Probability Added, which measures the impact of every play on the team’s chances of winning. WPA isn’t great for predicting future contribution, though.
February 3rd, 2010 at 11:28 am (#)
Name “right” at the top of the paper. Grr.
February 3rd, 2010 at 5:28 pm (#)
Daniel, I wasn’t meaning to ignore outs, just calculating an “average” that counts SOME outs as “hits” such as sacs, and the other things I mentioned. The only things counted as outs would be outs that had no positive effect, didn’t advance a runner, etc. I guess I’d call it the “Good at-bat average”?
Tom, I see your point, too.
February 4th, 2010 at 1:19 am (#)
Friars: Thanks for posting my piece that Tom Tango added to his site. I just started blogging in December and was stoked to see how well received that particular post was.
Mark: I like what Keith Law had to say on this subject. Law: “There are no productive outs. Just discard the term. Unless a run scores directly on the play, an out nearly always puts a major dent in your run expectation for the inning. So GMs who take productive outs into account are … bad GMs.” I also read him in a chat once say something along the lines of… the benefit of a truly productive out is so little it isn’t even worth measuring.
There are maybe very very very special cases at a critical time with the right guy on deck and the right guy on base, etc, when they are worth while. But, over the long haul, outs are highly unproductive regardless of what they accomplish.