Fangraphs WAR vs. Baseball-reference.com WAR

[adambr2]

Can someone clarify what the difference is and give an opinion on which is more reliable?

 

Sometimes they are frustratingly different. If you go by Fangraphs, Carlos Gomez took a big leap last year. If you go by base-reference.com, he was roughly just as valuable last year as in 2011. If you go by Fangraphs, Rickie Weeks had a down year last year, but not awful. If you go by baseball-reference.com, he was easily worse than an average AAA replacement.

 

I realize WAR is just a measurement, but I'm surprised there would be so much disparity in some cases between the two.

[sveumrules]

The main difference on the offensive side is the defensive input. On the pitching side fangraphs WAR is FIP based so it strips fielding out of the results while bref uses runs allowed so pitchers get credited or penalized based on the results of their balls in play. At least that's how I understand it.

[TooLiveBrew]

FanGraphs' WAR usually seems to pass the smell or eye test better for me. But that's not exactly an opinion I can back with tons of raw data. I just feel like I scratch my head more often with BBRef's version.

[logan82]

Try this. My biggest problem with rWAR was it used TZR for the defensive side. I don't think it is anywhere near as good as either DRS or UZR. Now it looks like rWAR is using DRS.

[TheCrew07]

FIP doesn't really strip defense out of the equation, rather it backwards calculates the what ERA should be largely off of the rate stats.

 

I hate fangraphs version of WAR for pitching but I do like fangraphs version of WAR more for hitters. I don't want to get into a debate about the best way to derive a true result but if we can't agree on quality measure of defense over 1 year span how accurate will any version of FIP be doing it game by game? Trying to derive a different result than what actually happened is the worst part of sabermetrics in my opinion.

 

logan's link will give you the base difference in the various versions out there and we've debated basically every metric used in all of the calculations at some point on this site. Older fielding metrics like TZ were still being used on the MiLB side up until Minor League Splits shutdown not that long ago simply because there isn't zone data available for minor league games.

 

Like most things in life it comes down to what you personally want to believe, WAR is just different ways of coming up with an opinion on a given players performance over a certain time frame. I think it's a useful comparison tool especially when talking about roster building concepts.

[Ennder]

Pitching WAR is completely worthless, I don't just mean flawed I mean worthless. Hitting WAR I feel fangraphs does a decent job with though.

[logan82]

Keep in mind WAR is a counting stat not a rate stat. That is probably part of the reason it doesn't work well for pitchers.

[Ennder]

There just isn't a way to attach a value to pitchers. FIP certainly doesn't do a good enough job of 'stripping out defense' and things like BABIP, LOB%, HR/FB stabilize to a pitchers norm, not some baseline. If you just use ERA then you end up counting defense twice, once for the fielders and once for the pitcher etc. Park factors are completely flawed since they treat every hitter the same way. I don't know the answer to how to do pitching WAR but I do know that fangraph's system does not work well at all.

[thebruce44]

FanGraphs' WAR usually seems to pass the smell or eye test better for me. But that's not exactly an opinion I can back with tons of raw data. I just feel like I scratch my head more often with BBRef's version.

 

Completely agree with this.