Welcome to the end of the 2015 season! Baseball is weird, so the Twins and the Mets are making compelling playoff cases. Baseball is extremely predictable, so the Cardinals and Dodgers are playing very well. Kris Bryant is awesome, but also Alex Rodriguez is awesome. That's about everything of note from the year in baseball thus far.
We're also probably seven-to-fifteen years into what we can consider the "sabermetric" era in baseball. And yet, even still, we can't get rid of some of the old statistics that have been burned into our frontal lobes over the past hundred-plus years. While some statistics carry almost no value (wins, saves), some are perfectly okay, and just waiting to be replaced with metrics that are more accurate and more descriptive. Those stats are things like OPS and ERA.
Today, I'd like to issue a brief manifesto on why using OPS is not such a great idea, and yet, why it's still used anyways. For many of you, this may be old hat. For that I'm sorry. But sometimes a little refresher is, well, refreshing.
Reasons Not To Use OPS: Math
It's been said before, but OPS is a statistic that just isn't mathematically sound. OPS is calculated by adding together a ballplayer's on-base percentage and his slugging percentage. While these two statistics are incredibly valuable in telling us just how good a player is offensively, they aren't meant to be mashed together like that. Here's why:
The best possible score a player can have for an on-base percentage is 1.000 -- this indicates a player gets on base 100% of the time he comes to the plate for a plate appearance (PA). No one does this -- a good OBP is somewhere over .350 or so in this era.
The best possible score a player can have for a slugging percentage is 4.000 -- this indicates a player hits a home run every time he comes to the plate ... well, every time he comes to the plate and gets credit for an at-bat (AB). No one does this -- a good slugging percentage is somewhere around .430 or so.
There's one big thing wrong here first: OBP uses plate appearances as a denominator in fraction form, yet SLG uses AB. What we have here are two mixed fractions, and if we learned anything in middle-school math class, it's that you can't really add two fractions with different denominators. If slugging percentage used PA instead of AB, this problem would be solved. But it doesn't.
If you did want to mash these two stats together in some format, you'd first want to put them on an even mathematical playing field. That's what more effective offensive statistics do.
Reasons Not To Use OPS: Context (or, OBP is more important!)
Perhaps the more important reason not to use OPS is that it treats OBP and SLG the same. Even if you disregard OPS's bad math underpinnings, it's easy to recognize that in 95% of cases, a player's OBP is going to be smaller than their slugging percentage. When combining the two numbers, the lion's share of the value almost always comes from slugging percentage, and guys who hit for a lot of power will see a rise in their OPS as well.
This would be fine -- I suppose -- if slugging percentage really is more important than on-base percentage. But, unfortunately, it's not. As Tom Tango has pointed out, on-base percentage is about 1.8 times more important than slugging percentage when it comes to run-scoring. That's not to say slugging percentage -- and power -- aren't important, because they definitely are! But when it comes to the statistics, slugging percentage and on-base are calculated in a way that doesn't exactly jive with linear weights ... which we can use to find the expected run outcome of any batting event.
So all in all, using OPS is bad math, and it pulls focus away from OBP and gives "extra credit" to players who hit for power, even though their offensive contributions may not be as important as a player who has a high OBP but a low(er) SLG. That's not precisely to say that on-base is more important than power ... we're only talking about these particular statistics here. Let's find an example from this season!
Nolan Arenado of the Rockies has a fantastic .889 OPS this year. That breaks down to a .317 OBP and a .572 slugging percentage. Nice work! Meanwhile, Buster Posey of the Giants has an also-fantastic .879 OPS this year. Posey's stats break down as a .392 OBP and a .487 SLG. Also excellent!
If one were to use OPS to rate these two players, you'd say that Arenado is a superior offensive player. But he's not. But we (read: Tom Tango) has established that OBP is about 1.8 times more of a driver of offensive success than SLG, making Posey a far superior offensive force than Nolan this season. (Of course this is even before we take into account each player's home park, etc. -- Posey is far and away the better hitter this season.)
The best part is that we're not forced to use OPS anymore. There are plenty of other offensive statistics that use linear weights, or take into account the differences between slugging and on-base. So OPS is bad and other stats such as wOBA (found at FanGraphs) or True Average (found at Baseball Prospectus) are better, right?
Reasons To Use OPS: Simplicity
While we've discussed why OPS is calculated in a "bad" way, we have to acknowledge that it has one big advantage due to that calculation: it's pretty easy to figure out on your own. Nearly every statistical outlet you turn to provides an on-base percentage and a slugging percentage, from gory math stats sites to your local TV broadcast. These numbers are everywhere. In addition, those two stats themselves are relatively easy to calculate from raw data. You can almost always find OBP and SLG splits for any particular circumstance, whether it's day-night, left-right, monthly, or whatever. And literally, all you have to do to calculate OPS is to add those two values together.
So OPS is simple to calculate, but it's not the best, most descriptive way to measure a player's ability. But how much do you "pay" for the simplistic answer -- are you still able to use OPS to effectively describe a hitter's offensive prowess?
I guess so. While OPS doesn't give you the right picture of how good a player's offense is, it's not like it takes you extraordinarily far afield. If a player has a high OPS, they're a good hitter. If a player has a low OPS, they're a bad hitter. This is still accurate. Where OPS suffers is in the middle, not in the margins. While it's tough to use OPS to parse the offensive difference between Player X and Player Y with similar scores, you can certainly use it to say "Joey Votto has a 1.012 OPS, so he must be a damn fine hitter." And that's worth something.
Reasons To Use OPS: Availability
OPS has one major contextual advantage over other statistics that can be used to determine a player's overall offensive value: the might of Baseball-Reference. FanGraphs and Baseball Prospectus are phenomenal statistical resources, but they probably still live in the shadow of the statistical monolith that is Sean Forman's wall of data.
At the risk of sounding negative towards B-R, which is the best, I am of the opinion that it is not the best place to go to if you want the clearest, most detailed data regarding contemporary players and statistical analysis. The metrics at B-R tend to fall on the side of simpler, older, and less-detailed compared to some of the things you can find at FanGraphs or Baseball Prospectus.
And yet, I'm not sure there's a more useful baseball research tool than B-R's ubiquitous Play Index ... which is quite literally the best small investment any baseball researcher could possibly make. And if you want to use the Play Index to sort / measure a player's overall offensive ability OPS (or OPS+) is one of the only good ways you have to use the tool. The options that I consider "better" ways to evaluate a player's overall offensive contributions -- including the aforementioned True Average or wOBA -- can't be Play Indexed, because they're unique to BP and FG.
That's not really a complaint! It's just that many writers and analysts are often going to default to a simpler, easier choice when researching. Because OPS is used on Baseball-Reference, it is probably more likely to be used.
In closing, I don't like using OPS in my writing because I try to be analytical, and am usually shooting for the most accurate statistics I can summon ... even when I'm writing for non-analytical outlets. I prefer accuracy over simplicity, especially to the extent that I can use other, readily-available statistics to tell the right story.
This is the important part here: the other options that exist to use rather than OPS are really, really good these days. Especially if you use statistics like weighted runs created plus (wRC+), you can easily just break it down to a single number like 146 -- that's Buster Posey's wRC+ -- that tells you how much better a player is, offensively, than league average. In addition, it takes into account pesky things like ballpark and league-average.
If you like OPS, that's fine too. It's not one of those statistics that is downright misleading (I'm looking at you, wins) ... its kind of like using ERA as a judge of a pitcher's talent level or effectiveness rather than RA9. You'll lose something in the calculation -- something important, perhaps -- but you'll probably get it 75% right, at least.
I'll say this, however, the more that we use inferior stats like OPS, the less likely it is that the "better" statistics will become a regular part of the lexicon. Or at least it will take longer for them to hit the mainstream. That's not that important -- things like hunger and refugee crises and pollution are important -- but it does matter, a small bit, to some people at least. Using something like OPS is a very, very tiny win for science and math. But one can see a reason (or two) why folks might want to default to OPS.
(Just, whatever you do, don't use wins.)
* * *
Bryan Grosnick is the Lead Writer for Beyond the Box Score and a columnist for Baseball Prospectus - Boston. Seriously, don't use wins.