keyboard_arrow_uptop

I have a friend who I like to hand things to. Every so often, I’ll hold out a blank sheet of paper, or a random object I picked up off a table, and she will inevitably take it from me, study it, then frown as she tries to figure out why she’s holding it. It works nearly every time. It’s not that she’s a particularly gullible person (although maybe she is, as well); it’s that she works on a contextual level. Patrick is handing me this, her brain decides, and therefore it must be important. She doesn’t actually think this, because then her brain would also remind her that I’m kind of a jerk. By the time that thought kicks in, it’s too late.

You can’t escape context. Here’s a verbal equivalent of being handed that blank sheet of paper: the statement, Giancarlo Stanton is currently hitting .211. There, now you’re stuck with this thought, wondering what to do with it. It’s not a Fun Fact, so that’s out. Is it a statement about how well Giancarlo Stanton is doing this year? Is it a statement about how well batting average encapsulates how well Giancarlo Stanton is doing this year? Does it include a value judgment about how well Giancarlo Stanton is going to do going forward? I’ve told you a simple fact, and yet by bothering to tell you, there must be something else. I have to be saying something.

In general, we tend to think of baseball statistics as falling into a spectrum between predictive and descriptive. FIP (and its advanced variant, cFIP) seeks to strain out the external forces surrounding a pitcher and quantify what he controlled, giving us an eye toward how he’ll do given a more neutral setting in the future. ERA, by comparison, accepts the world as it is, and pens its dutiful and literal history of baseball. But even ERA is tainted (in terms of objectivity) by that connotative E. Every statistic, as it turns out, bears its own motivations.

This is true since the dawn of statistics, and Henry Chadwick, who unapologetically employed his box scores to direct the game in the way he thought best. From the beginning, Chadwick’s intention in his accounting was not to describe the game, which could be done by the newspapers, but to evaluate which players, over the course of a season, were the best at their trade, and he consciously modeled his categories based on what traits he felt should be rewarded. In the gloveless, pebble-strewn early days of the sport, for example, fielding statistics dominated hitting statistics, while pitching skill was hardly tracked. Sacrifices were lavished upon, while stolen bases, which Chadwick felt to be the fault of the defense rather than prowess of the runner, went uncounted.

Chadwick’s biases were outdated before his own time was done; fielding improved rapidly by the turn of the century, and pitching became more vital to the defender’s causes, but box scores were slow to change. And some of his biases carry down to the current day, particularly in one underquantified realm: the error. In an era where batting average titles had lucrative repercussions, the scorekeepers took it upon themselves to mete out justice upon the hitters who earned their times on base.

At the time, this made sense: Batting average was the all-purpose stat, meant to describe and herald at the same time. Being a .300 hitter didn’t just mean that the batter got three hits out of ten; it meant (and still means, to a degree) that the hitter was good, that he was a better hitter than that .280 hitter. It was impossible to separate the effect from its invisible causes.

Fortunately, we don’t live in such dark times. We have a wealth of options to evaluate our players: True Average, for their hitting prowess, cFIP for pitchers, WARP for overall contribution. We no longer need our simpler statistics to instill a sense of justice for us. And so perhaps it’s time to revisit those older statistics and cleanse them of their connotations. Perhaps batting average is too loaded, even in our era, to erase the internalized benchmarks of .200, .300 and .400. And it’s hardly a stat worth fighting over anyway. But on-base percentage: this is our Maginot Line.

This is the point where I explain a new, pure on-base percentage, one that counts reaching on error, catcher’s interference, even stretch for fielder’s interference and the dropped third strike. The rate at which a player reaches base without making an out: nothing more or less. This is where I dig out the astronomical odds of Pat Corrales earning first base eight times on CI over his abbreviated playing career. It’s where I question the gradient of skill from being hit by pitches, which do count, to forcing errors, which don’t. It’s where I throw out everything we know about OBP, everything we feel about it.

But it’s meaningless, because this new OBP is easy to calculate. We could have used it a century ago. We just don’t want it. Even understanding that the law of averages will have its day, that the sun doesn’t care whose doubles it causes, that the talent of pitchers is just as capricious as the throws of rushed third basemen, it feels wrong to reward a batter for something we feel he didn’t deserve. After all, we already have an unjust set of statistics: wins and losses.

But is this justice? Is it fair to appoint official scorers as philosopher-kings in unventilated booths, to decide what should have been a hit and what should have been an out? Is it fair to place such an emphasis on the clumsy mistake of the fleet, rangy shortstop when a sluggish fielder with a smaller range kills with a thousand paper cuts? Social philosophers have argued that this is exactly what justice is, that fairness is defined by the categories we use: ultimately, that justice is a self-fulfilling prophecy. OBP is fair because we’ve all signed off on it, gave our implicit approval when Chadwick signed it into law with his quill.

We’ve spent 150 years in baseball thinking of justice as an application of Plato’s Form of the Good: that there is an ideal offensive performance, one we should aspire to and seek out in our hitters. It makes on-base percentage a pretty stat, but not an entirely honest one. It’s a shadow world we’ve built, baseball on a cave wall. Life and baseball are full of injustice; to polish them out of our statistics, especially the descriptive ones, seems unjust, somehow.

You need to be logged in to comment. Login or Subscribe
WoodyS
5/24
This isn't a bad idea, but it doesn't solve the biggest problem with OBP, the "situational" intentional walk.
Lassaller
5/24
Since the author's argument is that getting on base is all that matters, then getting on base via any and all intentional walks also counts. Carrying this argument to other stats, defensive indifference should be rescinded. A steal is a steal.
bhalpern
5/24
How about some love for Leo Cardenas? http://bbref.com/pi/shareit/VQneT
SabreTorres21
5/24
I love this idea. I'm all for factoring out anything having to do with the Error in our baseball stats. I'm not sure about dropped third strikes or catcher's interference, but OBP should absolutely include ROE. I'm surprised this hasn't happened a long time ago. bWAR includes ROE. I don't think fWAR does because their version of wOBA doesn't include ROE. It probably won't make a significant difference for any player, but I think it's important in terms of accuracy.
Schere
5/24
Just for kicks - Lou Brock ROE'd 163 times in his career, including 19 times in 1974, the same year he stole 118 bases. So his "new" OBP would be .395 vs .368 that year. Career OBP goes from .343 to .359, with lgAvg going from .330 to ??? over his career.
Schere
5/24
Boog Powell, by comparison, ROE'd 60 times in his career, in about 3/4 the at bats of Brock. So, he ROE's at about half of Brock's rate.
penski
5/25
I like it a lot. Context would still be important in determining value since not all outs (manager-ordered sacrifice bunts, e.g.) and "safes" (intentional walks, e.g.) are the same. For this reason I would also want to have a complementary "bases advanced" stat that simply counts how many bases the batter and any runners were advanced as a result of a batter's plate appearance (without subjective judgment). Examples: A) Grand slam counts for ten bases; B) successful sacrifice bunt with a man on first counts as one; C) runner on first goes to third on a single counts as three; D) runners on first and second both advance on a ground ball out counts as two; E) GIDP with a man on first counts as -1; F) runner on third thrown out at home attempting to score on a fly ball out counts as -3; G) a walk with runners on 2nd and 3rd counts as 1. H) a walk with runners on 1st and 2nd counts as 3. There could then be an advancement rate stat that takes into account opportunities for advancement. So instead of dividing TB by PA, it would be "bases advanced" divided by "advancement opportunities", with the runners-on-base situation determining the number of opportunities in any plate appearance. Bases empty = four opportunities, bases loaded = ten, runner on first only = 7, etc. Rates for the above examples would be A) 1.000, B) .143, C) .429, D) .222, E) -.143, F) -.600, G) .143, H) .333. I'd be very curious to see which players are thought of differently by virtue of their non-subjective "pure on-base pct." and "advancement pct."
morro089
5/25
At first I thought this was silly because then you'd be overvaluing a single to right field (vs left field) when a runner is on first...but would it? A single to right actually is more valuable. Maybe not a whole base more valuable, which leads to partial bases and if you want to go down that route might as well just use tAv. Depends on the stat's audience I guess.
penski
5/25
Yeah, it would just be a collection of data by observation. The runner either is on second or he's on third - one extra base or two - nothing partial. Context would still be important to determine "skill" or "true talent". This wouldn't replace Win Probability Added, for instance. It would be similar to RBI as a counting stat and similar to RBI per opportunity as a rate, just more granular (at the level of each base advanced rather than just crossing home plate). Like you said, a single to right with a runner on first is more valuable than a single to left, but we currently don't have any simple data that accounts for that difference.
BCulhane
5/25
Should we not count a hit when a batter is thrown out trying to stretch a single into a double? After all, if the purpose of OBP is to measure whether a runner (or run) results, a batter thrown out stretching really didn't accomplish his goal.
penski
5/25
I would classify that as being on par with a single followed by a caught stealing (as if on a pickoff without throwing a single pitch to the next batter), so a baserunning out, not a batting out. The batter was safe at first and any runners were able to advance prior to the batter-runner being thrown out at second.