FanPost

Murray Chass Was Right, or, How VORP Got the 2008 Royals Wrong, with an Open Letter to Rany

Original_image_mediumI'll admit it: the title (the first half, anyway) is meant as an attention grabber. But there is a fundamental truth contained therein: VORP, Baseball Prospectus' most well-known stat for valuing players, contains some serious flaws. I won't go over all of them here. If you want the general theoretical and practical background (including the well-established problems with VORP), check out my latest post at Driveline Mechanics, JustVORP versus Just VORP. There, I review the problems with VORP that are become evident when compared with a version derived from Justin Inaz's Total Value spreadsheet. There, you can find the reasons why something like Justin's spreadsheet (and Fangraphs' WAR follows many of the same principles) is a better gauge of the value of position players than VORP (which also has issues with pitching, but that's another can of worms). Here, I simply want to show how VORP misvalues several of the Royals position players, and not in a an innocuous way.

Yes, it must be admitted that on one (very superficial) level, know-nothings like Murray Chass and Jon Heyman are right: VORP is problematic. But rather than being discarded, it needs to be (and has been) superceded.

You really should read my post at Driveline, where I give all the details. To quickly summarize the relative problems of VORP as opposed to the version Sky Kalkman put together from Justin's stats called "JustVORP":

  1. VORP doesn't use the correct positional adjustments, leading to absurd results such as first baseman and DHs being  overvalued relative to other positions, catchers being undervalued, and more.
  2.  VORP doesn't adjust for league difficulty, so NL hitters are overrated, and AL hitters are underrated.
  3. The best established (and least debatable) problem with VORP is that, irony of ironies, it undervalues walks -- by half. It also tends to undervalue doubles, I believe, but given the stereotype of Baseball Prospectus as being one of the leading proponents of OBP back in the late 90s/early 00s when people had forgotten all about Branch Rickey, well, "ironic" seems to be too mild.

A couple of brief notes before I get to the Royals.

  • Defense doesn't enter this discussion. This isn't becaues I don't think it's important, but simply to focus on the problems of VORP in measuring offense. Moreover, people feel so differently (and strongly) about whether or not to use defensive metrics, and if so, which one to use, that bringing that in would only confuse matters. Both VORP and JustVORP are set up so that you can simply "add in" the defense stat or our choice if you so desire.
  • Most of this work is old hat in the "hardcore" sabermetric community. I'm not a hardcore sabermetrician, although I like what they do. Here, I'm simply tagging along. For the sake of this post, I'm simply assuming that Justin's version of a Baseruns  Run Estimator, which generates custom linear weights for the era, is better than the basis of VORP. See the post...
  • Finally, this isn't meant as a broadside or attack on Baseball Prospectus or Keith Woolner (a brilliant and nice guy who created VORP and now works for the Indians). Yes, I think it's problematic that this is their flagship stat, and I do think that FanGraphs has more than passed them by. However, this is fixable, as people have noted, and I hope they do so.

So.... how does VORP  "screw over" the Royals? And what does it matter? Remember that for the sake of this post, I'm simply assuming that "Just VORP" is the better valuation tool (it is, by the way). Here is a chart of how VORP and JustVORP, respectively, value the 2008 Royals position players (pretty much all of them are on there), and the difference between the two. A "+" indicates that VORP values a player more than JustVORP, and a "-" the opposite.

 

Name VORP JustVORP Difference
Mike Aviles 35 30 +5
David DeJesus 29 30 -1
Alex Gordon 20 28 -8
Mark Grudzielanek 11 12 -1
Miguel Olivo 7 10 -3
Ryan Shealy 7 6 +1
Alberto Callaspo 7 8 -1
Billy Butler 6 -4 +10
Jose Guillen 6 7 -1
Kila Kaaihue 1 1 0
Mitch Maier -2 -1 -1
Mark Teahen -2 7 -9
John Buck -3 7 -10
Jason Smith -3 -2 -1
Esteban German -5 -2 -3
Joey Gathright -6 -1 -5
Ross Gload -8 -8 0
Tony Pena, Jr. -25 -19 -6

I hope you're as outraged as I am. We should not stand for Tony Pena, Jr. being besmirched in this fashion!

But seriously, folks, this isn't about players I like or don't like being over- or underrated. It's about accuracy. If you buy the arguments about the better run estimators, the way to set replacement level, positional adjustments and stuff, then maybe you still think all this is just theoretical. Differences like being off on DDJ by 1 run aren't a big deal, true. But there are potential practical consequences if a team has the wrong value (and if you're a sabermetrically-oriented writer about the team, you might be making a mistake by using VORP, too).

The most glaring case in my eyes is that of Alex Gordon. If we take the generic 10:1 runs to wins conversion (and assume average defense for the sake of brevity), we'd look at Alex Gordon and say that this is a guy who is barely average, is going to be 25... maybe not much more than a stopgap. But JustVorp has him at +28, or 2.8 wins. That's a big difference! Now we're talking about a player who is clearly above average. That makes (or at least should make) a big difference when a team is thinking about their needs in the near future and/or whether to extend a player.

John Buck is a less crucial case for the Royals, and +7 is hardly a great year. But they are off by a whole win here -- the difference between a guy having a bad year (or just a solid bench player) and a worthless player easily replaceable for the minimum. Yes, 1 win -- teams pay (even this offseason) between $4M and closer to $5M for each marginal win. Mistakes like this can cost you. Teahen is undervalued just as much. Is he a guy a team should simply non-tender (VORP), or is he at least valuable as a utility  guy worth paying more for (JustVORP).

These are just some isolated cases. If you read the Driveline post, you can see other similar situations. But the best players in the league, the guys who are 7-8 WAR players, missing or gaining one win, you know, is still a big deal, but when you're talking about a guy who is around +20, a 10 run difference is the difference between an average guy or an above average guy, a solid starter or a bench  player, a useful bench player and a near-worthless scrub. These are not trivial issues. Especially for a team that needs to pinch every penny... Because yes, these sorts of valuations have financial repercussions that play into a team's ability to compete.

I would be remiss if I didn't (snarkily) note a bit of irony. As RR noted in his post ushering in the Dayton Moore Era, Bill Shanks, the Bravest Sychophant in the Game Today, noted right after the hiring that Moore possessed a Scouts Honor, that he didn't know what VORP was, RR noted in particular. The funny thing is, given Moore's record of position player acquisitions, as well as what we know about Baseball Prospectus' lack of defensive stats, VORP's undervaluation of walks, overvaluation of first basemen and DHs... well, it seems that Dayton Moore may very well know what VORP is, and uses it extensively. Perhaps it's just in his blood.

To conclude -- sabermetrics-friendly bloggers should be aware of the limitations of VORP. The walks issue, in particular, is well-established and no longer controversial. Why it hasn't been fixed, who knows. I hope they do. In the meantime, FanGraphs' WAR, for example, although not identical to Justin's stats, is very close and the way to go. It uses wOBA, which certainly fixes the walks and doubles issues. Even if you don't like the defensive stats (I do, of course), you can simply substract those yourself to get the Runs Above Replacement.

And finally, in conclusion, an open letter:

I'm sure Rany Jazayerli won't read this and has no idea who I am. And why should he? But I know he cares about the Royals, baseball, and, of course, is one of the founders of Baseball Prospectus.

Dear Rany,

If this does get to you -- please advocate VORP being fixed/put out to pasture in its present form. The tools with which to do so are there, as as been pointed out. Even if Dan Fox can't give you guys Simple Fielding Runs anymore (which would fix the embarrassing issues regarding defense and BP), you could at least use RARP (based on EqA/R) in your team audits and as your flagship stat, since it is a stand-in for the correct linear weights (albeit, in the case of EqA, an overly complicated one).

Here's the deal, Rany (which I'm sure won't matter, because you won't read this and really, who cares what I have to say? I know I don't.): If you are the force behind this change at BP, I promise not only to hold off comparing Christina Kahrl's transaction analyses to pretentious versions of the same from Rosenthal or Olney, but I will also stop mentioning that your friend Joe Sheehan, the guy who (rightly) mocked Jose Guillen's Final Vote eligibility in 2008, is the same guy who wrote in the previous offseason a column praising the Royals' signing of Jose Guillen, writing that it 'isn't likely to be a disaster' (ahem). I'll even throw in not linking to MGL's masterful demolition of Sheehan's analysis (written not long after Sheehan's)  that showed just how, um, "questionable" some of Sheehan's assumptions were.

Yes, Rany, if you personally get VORP replaced with something that at least fixes the walks problem, I will stop mentioning those two or three things that expose Prospectus as not quite living up to its analytical reputation. I will cease.

For three whole months.

Yours and BP's (sincere) admirer,

devil_fingers

This FanPost was written by a member of the Royals Review community. It does not necessarily reflect the views of the editors and writers of this site.