Notice: Trying to get property 'display_name' of non-object in /var/www/html/wp-content/plugins/wordpress-seo/src/generators/schema/article.php on line 52
keyboard_arrow_uptop

“A false conclusion: I hate it as an unfilled can.”
-William Shakespeare’s “Twelfth Night”

One of the objectives of the Basics series is to sort of rehash everything that is very basic: what we know now, and how did we get to the point that we know it? Filling in some of the back-story of what’s up in terms of player analysis serves a few important purposes. First, it helps eradicate some of the potential barriers anyone might have to analysis: take a look, and you that this isn’t all rocket science. If even a non-math person and ex-Teamster like me can get it or get some of it, I’m willing to bet that everybody else can too.

But if you like the flavor and you want more, there’s a really important second goal the Basics series can achieve if you’re new to this. Or, if you’re already familiar with this sort of stuff, the series serves as a general reminder to those of us who think we know it all. That second lesson is: When in doubt, don’t quit early.

Whether you call the line of inquiry about baseball that we’re involved in here “performance analysis” or “sabermetrics” or snarky and insufferable, one of the perils of working within this community is that it’s stocked with bright people devising ever-better mousetraps to define player value statistically, particularly offensive value. As a result, you run the risk of getting lost in the inevitable alphabet soup of different newfangled metrics. And rather than try to sort through them all, it’s perhaps easier to settle for a figure that some people refer to as simple and elegant: OPS, or On-base percentage Plus Slugging percentage. And perhaps worse yet, if you’re an analyst, it’s probably easiest to use OPS, because it’s the easiest to explain. As we mentioned earlier in the series, OPS winds up doing a pretty decent job of mimicking a description of overall offensive value. So it works, right? And if it works, and it’s simple, why not use it as a gateway stat to introduce fans to the broader, more diverse world of statistical analysis?

This is where I get off of the bus, and leave most of the analysis community and even most of my colleagues to their own devices. Why is that, and why should you care?

A lot of my colleagues, fellow statheads or bona fide performance analysts, go to the trouble of re-slaying the dragon of Batting Average every year. As dragons go, it’s pretty toothless, if persistent, and not everyone has seen it done, so why not do it again? In the wake of that sort of easy victory, there’s unfortunately an understandable desire to erect a new stat to substitute for what Batting Average did: to give you and me, fans and commentators and analysts alike, a number that we can all understand quickly while conveying something about relative player value. My concern is that OPS gets sucked into a breach because we just have to have something to use to say everything. Any why not OPS, because it’s supposed to describe offensive value simply and elegantly? Add two numbers we all like–on-base percentage and slugging percentage–and we get a big number that dodges all that heavy math while giving something we can all get and that sort of works.

It’s the “sort of” part that bothers me. Maybe it’s my old experience to medical professionals, and their commitment to “do no harm.” What does OPS tell us that we don’t already know? We don’t need OPS to tell us that Barry Bonds is better than Larry Walker. So what does OPS tell us? Is it actually a worthwhile quick-and-dirty descriptor of value?

For a completely arbitrary and unfair sample, let’s take a look at the following groups of players with similar OPS figures. We’ll see their triple crown rate stats (plain old Average, OBP, and Slugging), OPS, and two of our homegrown favorites from among the statistical alphabet soup, Keith Woolner’s Marginal Value Lineup rate (MLVr) and Clay Davenport’s Equivalent Average (EqA). Strictly speaking, I haven’t made an apples to apples comparison here: Equivalent Average assigns value to stolen bases and the like, which triple crown rate stats, OPS, and MLVr do not.

Let’s start off with a two different trios of star quality outfielders within a point or two of OPS:


Group A          AVG  OBP  SLG  OPS  MLVr  EqA
----------------------------------------------
Sammy Sosa      .279 .358 .553  911  .256 .303
Carlo Beltran   .307 .389 .522  911  .225 .310
Vernon Wells    .317 .359 .550  909  .241 .302


Group B          AVG  OBP  SLG  OPS  MLVr  EqA
----------------------------------------------
Jose Guillen    .311 .359 .569  928  .289 .304
Lance Berkman   .288 .412 .515  927  .222 .312
Magglio Ordonez .317 .380 .546  926  .279 .310

Now, remember, we’re just talking about offensive value here, not potential for growth, or adjusting for age. MLVr and EQA adjust for ballpark, which helps explain a major part of the differences, and Equivalent Average includes the running game. OPS is the only number that says Sosa, Beltran, and Wells were virtually identical in terms of offensive value, despite radically different skills sets (reflected in their basic triple crown rate stats), and in stark contrast to what MLVr and EQA say about their offensive contributions.

Looking at Group B, we’ve got some pretty big differences. Lance Berkman takes a lot more walks, but he plays in an air-conditioned juice box. OPS sees them as equally valuable, but again, their individual basic rate stats or their production in the advanced stats couldn’t be more different.

Lest we keep our heads in the clouds of greatness, how about a trio of backup catchers?


Catcher          AVG  OBP  SLG  OPS  MLVr  EqA
----------------------------------------------
Mike DiFelice   .254 .299 .397  696 -.172 .234
Bobby Estalella .200 .294 .400  694 -.197 .233
Miguel Ojeda    .234 .331 .362  693 -.041 .254

OPS sees a bunch of players with similar value, whereas eyeballing the rate stats will tell you none of these guys is an offensive dynamo or much of a slugger. Since Ojeda gets on base and makes fewer outs, he easily ranks as the better player on either an instinctual/old stats level, or using the advanced stats.

Forgive me a bit of tedium on the subject, but here’s a deliberately selected example pair:


First Baseman    AVG  OBP  SLG  OPS  MLVr  EqA
----------------------------------------------
Scott Hatteberg .253 .342 .383  725 -.020 .258
Ken Harvey      .266 .313 .408  721 -.089 .240

The two worst-hitting first basemen in the AL don’t have a lot to say for themselves, but where OPS sees two roughly equal players, again, it manages to undervalue Hatteberg’s ability to get on base while overvaluing Harvey’s marginally better but still noxiously feeble power.

Beyond the old-fashioned straight stuff–the rate stats–or our own homebrewed stats, you can play this game with almost any metric. All of them reflect some element of its designer’s thoughtful research, and they’ll say generally similar things (Barry Bonds is good, Lenny Harris is not), with distinctions between whether the system is scaled to an absolute standard or against that year’s league average.

So why, in the face of that diversity, and with a wealth of information with which to describe players and their ability to contribute, do we have to erect a contraption like OPS? The older numbers–Average, OBP, and Slugging–might individually say less about player value, but in concert, they describe players in terms of their abilities to get the ball in play, get on base, not make outs, if they hit for power or lots of singles…basically, a good snap-shot of the player’s value. The shape of a player’s performance is well-described by triple crown rate stats, which you have to have in order to generate OPS in the first place. What tells you more about what a player brings to the table: .700, or .270/.350/.350 versus one who hit .260/.300/.400? Why build OPS, since it eradicates those distinctions while creating phony similarities? I guess it makes up in laziness what it misses in accuracy. But at that point, why bother? Let’s consider this the Reese’s approach to things that have value by themselves and in concert: if you have your chocolate and your peanut butter, you do not have to put them in a blender to enjoy them together.

On the other side of the coin, like a lot of lazy reductionism–deficit spending is good, all carbs are bad, Scott Hatteberg and Ken Harvey are equally valuable hitters–OPS misses a lot. Ballpark effects, league context, contributions on the basepaths, all of that gets ditched in the names of “simplicity” and “elegance,” while generating wrong answers. And if you want to parse the numbers with some selective special pleading about nagging injuries or that year’s platoon split or whatever, you’re missing the point of OPS, which is to give you a simple answer to a complicated question. Once you start parsing the distinctions between everyone who has a similar OPS, you’re going to find that everyone has a real-world explanation, a possible problem, a circumstance, a contingency.

The reason for that is pretty simple: baseball players are collections of virtues and flaws, and by extension, baseball statistics are tools that generally describe those virtues and flaws. Why use a stat that ignores those things?

You can certainly consider this a matter of taste. I admit, I do not see the value in a system that reduces a player to a single number, and instead enjoy the sticky diversity that we’ll always find in reality. Similarly, I hope to encourage all of you to explore, to evaluate, and to pick and choose. Different offensive stats describe different skills in slightly different ways. Performance analysis isn’t going to give you an ultimate answer, it’s going to give you a lot of answers. The challenge to you is to learn to discriminate. When analysts start using Road OPS or OPS in late-and-close situations, however much you will have learned from them, when they start leaning on a bad tool to make a point, they wind up making their point badly. The middle road doesn’t automatically go somewhere: triple-crown rate stats will tell you something concrete about results, while advanced metrics will tell you something descriptive about the value of those results. What is OPS supposed to have told you in that situation? You can do better, and so can we.

Consider the lot of big league teams; that’s where the basic real-world issues get involved. Specific teams have specific needs. Who goes into an offseason saying “Wow, we need more OPS!” They might need more power, they might need more men on base, they might need a whole new lineup, but what would be the point of questing for OPS? The point at which a team says “Wow, we need a guy with an .800 OPS” is the point at which you need to know they’re swapping out “800 OPS” for “a decent player,” because it sounds downright math-y, and that’s today’s flavor when it comes to jargon. You can do better, and so can we.

Which is why, having gone through the Basics series and what it has to offer, it’s my hope that you’ll feel welcome to think about the game in a new way. I’m going to tell you to stick with the most important lesson any of us who came before learned: keep doubting, keep asking, and refuse to settle for the easy answer.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe