Jump to content


Yo! You're not logged in. Why am I seeing this ad?

Photo

Please Read This Before You Use WAR In a Post


  • This topic is locked This topic is locked
10 replies to this topic

#1 absintheofmalaise


  • too many flowers


  • 10883 posts

Posted 18 October 2010 - 01:57 PM

The use of WAR as a stat to talk about player value is on the rise here. Because of this, we ask that posters who do use WAR to also let people know which version of WAR they are using. Baseball Reference and Fangraphs calculate WAR differently. Since WAR is a combination stat, it would also help tremendously if you'd break down both the offensive component and the defensive component whenever you post using WAR.

Fangraphs uses Park Adjusted Runs Above Average based on wOBA to calculate the offensive component and UZR to calculate the defensive component. You can learn more about it here and here.

Baseball Reference uses the Sean Smith's WAR data. You can learn more about how it is calculated here and more about Sean Smith here.

Thanks

#2 Worst Trade Evah


  • SoSH Member


  • 10834 posts

Posted 18 October 2010 - 03:40 PM

The use of WAR as a stat to talk about player value is on the rise here. Because of this, we ask that posters who do use WAR to also let people know which version of WAR they are using. Baseball Reference and Fangraphs calculate WAR differently. Since WAR is a combination stat, it would also help tremendously if you’d break down both the offensive component and the defensive component whenever you post using WAR.

Fangraphs uses Park Adjusted Runs Above Average based on wOBA to calculate the offensive component and UZR to calculate the defensive component. You can learn more about it here and here.

Baseball Reference uses the Sean Smith’s WAR data. You can learn more about how it is calculated here and more about Sean Smith here.

Thanks

Tango on his blog distinguishes between fWAR (Fangraphs WAR) and rWAR (B-Ref WAR), which is a quick and easy way to identify the method chosen.

#3 mabrowndog


  • Ask me about total zone...or paint


  • 37768 posts

Posted 18 October 2010 - 05:25 PM

A couple of weeks ago Dave Cameron expanded on the differences between the two sites' WAR calculations, explaining why Fangraphs uses FIP in the pitching component.

Details here and here.




#4 jose melendez


  • Earl of Acie


  • 12720 posts

Posted 09 December 2010 - 11:46 AM

Is there such a thins a park adjusted WAR? The reason I ask is because it seems important in the context of Crawford. For a player who derives so much of his WAR defensively, the fact that he is moving to a small left field seems like it would dramatically reduce his WAR.

#5 JimBoSox9


  • will you be my friend?


  • 12423 posts

Posted 09 December 2010 - 10:50 PM

Is there such a thins a park adjusted WAR? The reason I ask is because it seems important in the context of Crawford. For a player who derives so much of his WAR defensively, the fact that he is moving to a small left field seems like it would dramatically reduce his WAR.


Fangraphs WAR uses UZR for defense and B-Ref WAR uses Total Zone. Both are park-adjusted.

#6 SumnerH


  • Malt Liquor Picker


  • 14556 posts

Posted 29 December 2010 - 04:34 PM

Fangraphs WAR uses UZR for defense and B-Ref WAR uses Total Zone. Both are park-adjusted.


Further clarifications from the ongoing "Comparing Players" thread:
Fangraphs uses Total Zone for pre-2002 players, and dRS (defensive runs saved) for catchers.

#7 absintheofmalaise


  • too many flowers


  • 10883 posts

Posted 17 September 2011 - 08:27 AM

One other thing about using WAR, since it's a combo stat like OPS, could you also break it down into its individual components like you would with BA/OBP/Slg/OPS. Using fWAR as an example. They use Batting/Base Running/Fielding/Replacement/Positional = RAR = WAR .
Pedroia's line today looks like this: 28.6/-0.8/17.4/22.5/2.2 = 69.9 = 7.4

#8 Toe Nash

  • 3010 posts

Posted 04 May 2012 - 10:24 AM

Baseball Ref has updated their calculations of WAR:

http://www.sports-re...-reference-com/

All in all, these are major improvements to the system and while there are dozens of details, here are the main points of difference between our old framework and the new one.

  • Switch from BaseRuns for batting to an advanced wRAA metric.
  • Folding ROE, infield singles, SO vs. Non-SO into wRAA.
  • Excluding pitchers' hitting and averaging by league rather than year from the league averages for wOBA and wRAA.
  • Estimation of CS numbers for leagues they are missing.
  • Use of Baseball Info Solutions Defensive Runs Saved from 2003-present (in our view the most advanced defensive metric).
  • Use of a player-influenced runs to win conversion for both batters and pitchers based on PythagenPat.
  • Use of a player-specific park factor for pitchers weighted by actual appearances in each park.
  • After a preliminary WAR calculation, we fine-tune the replacement level on a playing time basis, so the total WAR in each league is very consistent year-to-year.
  • I've changed the oWAR and dWAR formulations. oWAR is now called ndWAR for (no-defense WAR), but is the same otherwise. dWAR now contains the position component of WAR, so the Career Leaderboard is now dominated by SS, C and other great defensive players.

This page has a nice table at the end comparing fWAR and bWAR:

http://www.baseball-...explained.shtml

#9 absintheofmalaise


  • too many flowers


  • 10883 posts

Posted 16 November 2012 - 06:18 PM

This was posted over in the MVP thread in the MLB forum by a devotee of Markov, and in my opinion, it is must reading for anyone who uses WAR.

I think some of the stat types do themselves (and their cause) a disservice by quoting metrics that don't make sense, even to most stat supporters, while making their cases for Trout. Look at the 2012 WAR numbers for Trout and Cabrera and the first thing a non-numbers guy (or a numbers guy) should notice is that the Baseball Reference WAR and the Fangraphs WAR are not the same. If they bothered to look closely, and most won't, they would see there are significant differences in the defensive numbers, discrepancies in base running and what is measured in running, and even differences in the metrics that make up the total WAR numbers.

Focusing on defense, Trout is 21 runs above an average player at his position per BR and + 11.4 runs per FG; Cabrera is -4 runs per BR and -10 per FG. How does a non-stat guy reconcile that? Is the difference 31 runs or 15? Or something else entirely?

Base running is measured differently on the 2 sites - BR looks at SB, CS, PB, WP, and Defensive Indifference, while FG considers SB and CS, then separates out other base running events like going from 1st to 3rd on a single into a new stat called Ultimate Base Running (UBR). BR has Trout 10 runs above average with their running metric, while FG says he is +7 runs on SB and CS and +5 runs in UBR, for a total of +12. Cabrera is even (0) runs in BR, -0.4 in SB and CS on FG, and -2.4 using UBR on FG.

Also, BR uses a batted into double play stat in their WAR calculation, that is not included in the FG number (Cabrera is -5 and Trout is +1).

Got it ;)?

Effectively, WAR, in both forms, is a needlessly confusing and convoluted stat that will never, and should never, catch on with the average fan. The metric adds an offensive/batting component that is pretty accurate (maybe 85 - 90%), with a defensive number that is at best 50% accurate (looking at one year of defensive stats you should regress the number by 50%, so if you see a player who is +10 runs in one full season, regress him to +5), a base running number (UBR) that is also in need of serious regression, a positional adjustment (that could use an update, esp for DH), and (imo) a completely unnecessary and bewildering (to a common fan) adjustment for a mythical replacement player.

Do people really expect Joe Burger King to accept this stat as Gospel when a vast majority of the people using it have no idea what it includes, how it is calculated, and what its strengths and weaknesses are? If you were performing a science experiment, would you add a stat you had 90% confidence in, to one you had 40% confidence in, include another you had 20% confidence in, then call it accurate? I know of one person on the board who might, but hopefully he's the only one.

Trout had a better season than Cabrera in 2012, of that there is no question. Cabrera may have been a slightly better hitter, but Trout played a more difficult defensive position, played it better, and had a huge advantage in stolen bases (49 steals, caught only 5 times). He also played in a home park that was tougher to score runs in (and ergo hit in) than Cabrera. I know this is what WAR is attempting to measure, but it does not work and quoting it as if it were an accurate stat erodes the discussion and turns off the non-stat types who are (rightfully) confused, the baseball traditionalists that see the metric's flaws, and the math heads who despise fallacious formulae.

Is it a huge deal that Cabrera won? Not to me; the Triple Crown is pretty cool, imo, and I don't have a rooting interest for Trout (and URI would tell you I always root for the Latino guys). I do believe Trout deserved the award, however, and tried to explain why in simple terms above.



#10 absintheofmalaise


  • too many flowers


  • 10883 posts

Posted 14 May 2013 - 10:59 AM

Colin Wyers posted an excellent article today on BPro about the use of WAR. Interestingly enough, a Tweet from Heyman inspired it.

 

 

 

As that old pop song goes, “oops, he did it again.” Sports Illustrated’s Jon Heyman is asking questions about WAR:

@JonHeymanCBS

i like war. but heres my war mystery of the week. mark reynolds .988 ops, 0.8 war; elliot johnson .641 ops, 0.9 war.

If this sounds familiar, Heyman asked a similar question about Starling Marte and Bryce Harper roughly two weeks ago. There was a predictable dust-up on Twitter, then and now, and a few articles were written. And while much noise has been made, very little has been said, and that’s sad for a number of reasons:

1. Heyman is actually right on the merits here,

2. These aren’t edge cases or weird outliers but examples of a fundamental problem with the construction of certain WAR metrics and with several of the underlying components,

3. These are problems that have been identified for some time now, and solutions actually exist for them already,

4. And finally, that the response to Heyman is indicative of a larger problem with the sabermetrics movement.

 

Wyers covers all of these and more in a very easy to understand way. 



#11 MyDaughterLovesTomGordon

  • 3948 posts

Posted 14 May 2013 - 03:25 PM

Keith Law's Behind the Dish May 7 podcast is with B-Ref rounder Sean Forman and goes pretty deep into WAR, what "replacement level" means, how much they trust the defensive piece of WAR (a lot, in Forman's case), and basically how they use WAR personally. 

 

Forman says he considers two players within a half-run of each other basically the same, for example, which gave me a better idea of what the margin of error is in their calculations. Also, there's some interesting stuff on how B-Ref and Fangraphs came together to at least establish a baseline for what the replacement-level number actually is, so that they'd be more in line with one another, even if their calculations were different. 

 

Now, he says, both systems should add up to a total of about 1,000 WAR for the year.

 

I found it to be a really great listen and highly informative, even after doing some reading about WAR previous to listening.






1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users