Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Wednesday, November 29, 2017

Aaron Judge, José Altuve, and the Next Battle in the War Over WAR - The Ringer

Maybe we should stop calling different systems by the same name. We have different abbreviations for BABIP and AVG because, although they are both measuring a player’s ability to get on base by hitting the ball, they are calculated differently and are different metrics. Fangraph’s WAR and Baseball Reference’s WAR are two different calculations. At least Baseball Prospectus calls their version WARP to differentiate their version.

James’ Win Shares is/was a decent concept but its construction was fatally flawed. As Rany mentioned in his article, we now have the ability to make a more nuanced assessment of individual player’s contributions, related to the timing of their performance.

But I think it’s time to do away with half measures, and decide fully what WAR is supposed to represent. If it’s supposed to represent value, then it needs to evolve to account for the fact that all players, not just relievers, can perform in ways that alter the relationship between runs and wins. WAR should reward a hitter who bats .400 with runners in scoring position and penalize one who hits .136 in high-leverage situations. If the day comes when we can evaluate for how a player performs defensively in high-leverage situations, we can account for that too.

Why not develop new metrics which project performance without needing to be tied to the WAR name? In scouting teams often use terms like Overall Future Potential (OFP) and Future Value (FV). Why not tie our statistical assessment to the common scouting lexicon?

We can, and should, have a “predictive” version of WAR that evaluates a player’s performance based on skills that will carry forward into the future. This would not only strip away “clutch” and situational hitting that doesn’t carry over much from one year to the next and strip away luck on batted balls in play, but as our data set improves would also account for Statcast data like launch angle and exit velocity, so that the player who hit a ton of at-’em balls or the pitcher who gave up a lot of windswept home runs into the first row would have a statistic that says, look, this guy might have sucked last year, but if a butterfly had flapped its wings last April he would have been really good. As Nate Silver suggested, maybe we can call it predictive WAR, or pWAR.

Jim Furtado Posted: November 29, 2017 at 11:20 AM | 16 comment(s) Login to Bookmark
  Tags: sabermetrics, war

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. Ziggy: The Platonic Form of Russell Branyan Posted: November 29, 2017 at 12:02 PM (#5582654)
Maybe we should stop calling different systems by the same name.


How about 'bWAR' and 'fWAR'?
   2. BDC Posted: November 29, 2017 at 12:18 PM (#5582671)
We can, and should, have a “predictive” version of WAR that evaluates a player’s performance based on skills that will carry forward into the future


Isn't that what the various projection systems already do?
   3. PreservedFish Posted: November 29, 2017 at 12:45 PM (#5582693)
BDC, Not exactly. I think he's talking about a WAR that gives no credit for bloop hits, and gives credit for at'ems.

A projection system might already use that information, but there's nowhere where you can look up "oh, for projection purposes, Avisail Garcia actually only had 3.2 WAR this year."
   4. fra paolo Posted: November 29, 2017 at 12:45 PM (#5582694)
Win shares looked like the ultimate all-in-one statistic when he introduced it, but it suffered from one fatal weakness: It didn’t account for replacement level.

That's not exactly true. Win Shares does account for Replacement Level. The actual objection raised at the time of publication was that it zeroed-out performances below its Replacement Level, and that these should be better represented by Loss Shares. Also, I believe people also argued its Replacement Level was too low. Win Shares Above Bench was an attempt to insert a Replacement Level more in line with the generally accepted quantification of such players.
James’s win shares approach is that if a team wins more games than we would expect from its underlying statistics, we should apportion the extra wins to all of its players in proportion to their value. That approach might have been the best method in 2002, when we didn’t have the situational data we have now.

That's another misrepresentation. The Runs Created portion of adding value in fact uses some situational data.

Coincidentally, I was reading just this part of Win Shares last night. James adjusts the initial Runs Created calculation by what the batter did with runners in scoring position (for non-HR hits) and men on base (for HRs).

Now it's possible the article is still right in suggesting that Hosmer is over-valued by Win Shares relative to Merrifield and Perez, but their superior hitting in high-leverage PAs is definitely going to be reflected to some extent in the calculation of their batting Win Shares.
   5. Zach Posted: November 29, 2017 at 04:55 PM (#5582935)
But I think it’s time to do away with half measures, and decide fully what WAR is supposed to represent. If it’s supposed to represent value, then it needs to evolve to account for the fact that all players, not just relievers, can perform in ways that alter the relationship between runs and wins. WAR should reward a hitter who bats .400 with runners in scoring position and penalize one who hits .136 in high-leverage situations. If the day comes when we can evaluate for how a player performs defensively in high-leverage situations, we can account for that too.

This is actually a very telling argument. For the most part, WAR measures value by apportioning a constant number of runs per win. But for relievers, we have to multiply that number by ~1.4, or else we end up with relievers being tremendously undervalued. So which is it -- is leverage important or not?
   6. GuyM Posted: November 29, 2017 at 05:11 PM (#5582955)
This is actually a very telling argument. For the most part, WAR measures value by apportioning a constant number of runs per win. But for relievers, we have to multiply that number by ~1.4, or else we end up with relievers being tremendously undervalued. So which is it -- is leverage important or not?
Umm, that's because top relievers are the only players who generally have an average leverage significantly different from 1.0. And that's because they are deliberately inserted into games at those times. There's really no inconsistency here. But ig you want to multiply every hitter's batting runs by their average leverage, knock yourself out -- it won't change much.
   7. Zach Posted: November 29, 2017 at 06:13 PM (#5582988)
Umm, that's because top relievers are the only players who generally have an average leverage significantly different from 1.0.

That's an ad hoc rationale, not a consistent mapping from runs to wins.
   8. Zach Posted: November 29, 2017 at 06:21 PM (#5582992)
By which I mean that it's a patchwork -- for most players, a run is a run. But for relievers, a run is actually a run times a leverage index. So there are actually two treatments for a run in the definition of WAR, because if we used a consistent treatment we'd obviously undervalue relievers relative to their usage in the game / salary / trade value, etc. What you want is to have the same mapping applied for all players.

But ig you want to multiply every hitter's batting runs by their average leverage, knock yourself out -- it won't change much.

The question of whether we should consistently account for leverage and contribution to wins is exactly the subject under discussion.
   9. This is going to be state of the art wall Posted: November 29, 2017 at 06:21 PM (#5582993)
I mean, aren't the projection systems

We can, and should, have a “predictive” version of WAR that evaluates a player’s performance based on skills that will carry forward into the future.


Doing this. As mentioned upthread, nobody thinks Avi Garcia is going to have a BABIP of nearly .400 next year. Regress him down to a more reasonable (for Avi at least) .340 and yea he's around 2-3 WAR not 4-5.
   10. Ulysses S. Fairsmith Posted: November 29, 2017 at 06:33 PM (#5583000)
Win Shares didn't zero out performances below its replacement level; it zeroed out performances below its zero level. It didn't start from replacement level; it started from zero level. There was some discussion of replacement level in the book, but that's not how the system was built.

   11. Jay Z Posted: November 30, 2017 at 12:30 AM (#5583164)
Win Shares didn't zero out performances below its replacement level; it zeroed out performances below its zero level. It didn't start from replacement level; it started from zero level. There was some discussion of replacement level in the book, but that's not how the system was built.


Yeah, that was my memory. A 50 win team might be close to replacement value overall, but there were still be 150 win shares. Actual replacement value would be more than zero win shares.

It's certainly possible to have a player so bad that he really should be negative win shares. Particularly pitchers and defensive players, I suppose. A team that gives up 10,000 runs a game will cancel out any number of good hitters. So those players should have negative win shares. Players that bad would be very rare.
   12. fra paolo Posted: November 30, 2017 at 10:59 AM (#5583306)
Win Shares didn't zero out performances below its replacement level; it zeroed out performances below its zero level. It didn't start from replacement level; it started from zero level.

Yes, you're right. James denied any connection with the zero level and replacement level in the book. But what's interesting about that denial is how he went about it.

On 23 November 2001, David Smyth, of Retrosheet fame, wrote this on a Fanhome thread discussing the preliminary summary of Win Shares in the New Historical Abstract:
And the derivation of the method has nothing to do with consciously selecting a replacement level of .200 for offense, or whatever.

From internal reference on that thread, which does not survive complete on the Internet Archive, I believe the reference to the 'replacement level of .200' refers back to something written somewhere by Patriot, who subsequently would post a 'win shares walkthrough' at his blog, that is still one of the better analyses of Win Shares one can find.

So on page 107 of the Win Shares book James writes:
Some people have said that the Win Shares system assumes that the replacement level is a .100 player, and other people have said that it assumes a replacement level of .200....But both answers are wrong, because the 'marginal run level' in the Win Shares system is not a replacement level; it's assumed to be a zero-win level.

This would bear out the point, that there is no replacement level in Win Shares. Except, further on, on page 109, James has this to say:
I once theorised that a replacement-level player should be about a .300 to .350 player, but this was assuming that a player who created runs at a 70.7% of the league norm should be considered a .333 player.....But if a team was 71 % of league norm offensively and also 71% of the league norm defensively, their likely winning percentage wouldn't be .333, it would be .200. [my italics]

And just before that he writes:
...we have no firm idea what 'Replacement Level' is, even given an older, established theoretical concept to analyse the game. Exactly what 'replacement level' is is difficult to establish, because the distribution curve of actual baseball players simply falls away into a gas on the left end of the spectrum, the sub-.500 side.

He goes on to describe some other uncertainties about the zero-level player: 'in reality his [win-contribution] is merely near-zero.'

I submit that it is a reasonable interpretation that James didn't want to call what he did in Win Shares 'Replacement Level' because he didn't believe he could yet quantify the contribution of a replacement level player. He already had people trying to undermine his work by asserting his replacement level was too low, and James was wasting the time of serious saberists. (A charge that would be made subsequently, IIRC.)

But the zero-level in fact represents James' attempt at conceptualising how a Replacement Level Team would perform in a major-league environment. Twenty-five 'near-zero' wins is equal to 5-157? 18-136? Or the 1899 Cleveland Spiders 20-134? What constitutes near-zero?
   13. Jesse Barfield's Right Arm Posted: November 30, 2017 at 11:10 AM (#5583309)
I mean, it makes no sense that batting average doesn’t count sacrifice flies while OBP does, but if I proposed a version of batting average called sfAVG, I’d be justifiably defenestrated.


This is impossible - you can't throw someone out of a window from your mother's basement.
   14. GuyM Posted: November 30, 2017 at 12:45 PM (#5583413)
By which I mean that it's a patchwork -- for most players, a run is a run. But for relievers, a run is actually a run times a leverage index.
Because for non-relievers, multiplying by leverage basically means multiplying by 1. What's the point? The results will be the same.

Now, if you think each player's production should be adjusted for their *individual* leverage, that's fine. But that's another discussion. WAR's construction is not contradictory, it's just taking account of a fact unique to relievers, in the same way we account for PB and CS when evaluating catchers but not when evaluating CFs.
   15. Rally Posted: November 30, 2017 at 02:49 PM (#5583584)
This is impossible - you can't throw someone out of a window from your mother's basement.


Some basements have windows. You could throw someone out it if you have a real strong throwing arm. Given your screen name, you should be up to the task.
   16. dlf Posted: November 30, 2017 at 02:55 PM (#5583589)
On 23 November 2001, David Smyth, of Retrosheet fame wrote this on a Fanhome thread ...


Nitpick: David Smyth at Fanhome is the guy who came up with BaseRuns. David Smith is the guy at Retrosheet.

You must be Registered and Logged In to post comments.

 

 

<< Back to main

News

All News | Prime News

Old-School Newsstand


BBTF Partner

Support BBTF

donate

Thanks to
^^^
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogOTP 15 January 2018:Mississippi's anti-gay marriage law is hurting two college baseball teams
(1971 - 7:40am, Jan 21)
Last: Greg K

NewsblogNBC Sports Philadelphia: Morphine found in Roy Halladay's system before fatal plane crash
(29 - 6:47am, Jan 21)
Last: . . . . . . . . . .

NewsblogMLB's plan on how they can make baseball games at least 10 minutes shorter
(26 - 5:53am, Jan 21)
Last: Leroy Kincaid

NewsblogRyan Thibs has his HOF Ballot Tracker Up and Running!
(1568 - 1:38am, Jan 21)
Last: QLE

NewsblogOT - NBA 2017-2018 Tip-off Thread
(2750 - 1:21am, Jan 21)
Last: maccoach57

NewsblogComparing a Player Outside His Era | Articles | Bill James Online
(26 - 11:18pm, Jan 20)
Last: David Nieporent (now, with children)

NewsblogWhy is J.D. Martinez's big bat still available
(39 - 11:07pm, Jan 20)
Last: cardsfanboy

Sox TherapyWhere Is The Offense...It's Right Over Here
(11 - 9:41pm, Jan 20)
Last: villageidiom

NewsblogInside Baseball | Scott Boras Weighs In On Slow Market
(22 - 8:59pm, Jan 20)
Last: cercopithecus aethiops

NewsblogOT - 2017 NFL thread
(1324 - 8:52pm, Jan 20)
Last: PASTE, Now with Extra Pitch and Extra Stamina

NewsblogWhat Is Baseball’s Equivalent of the Vikings’ Miraculous Victory?
(90 - 7:17pm, Jan 20)
Last: Greg K

NewsblogBenFred: Everybody can find a story about Musial's magic | Ben Frederickson | stltoday.com
(3 - 7:15pm, Jan 20)
Last: RMc's Unenviable Situation

NewsblogThe Current MLB Free Agent Market - Perception vs Reality? | Jays From The Couch
(12 - 4:55pm, Jan 20)
Last: Bote Man

NewsblogAndruw of Center Field
(213 - 4:47pm, Jan 20)
Last: Rally

Hall of MeritMost Meritorious Player: 2011 Discussion
(83 - 3:15pm, Jan 20)
Last: caiman

Page rendered in 0.2011 seconds
47 querie(s) executed