Please explain difference between rWAR and fWAR

[Angelsjunky]

Someone here has explained it to me before, but it evidently hasn't sunk in. What exactly is the difference between Baseball Reference's (rWAR) and Fangraphs' (fWAR) versions of WAR? I know Fangraphs is based upon Ultimate Zone Rating and B-R on something else, but what does all that mean? What are they measuring?

 

The reason I ask is that I'm having a hard time with what rWAR does with Trout's defense. As far as I can tell the hitting component is either the same or very similar.

 

Baseball Reference:

4.3 rWAR; 5.5 oWAR, -1.2 dWAR

 

Fangraphs:

5.4 fWAR; 0.3 Fielding Runs

 

Baseball Reference's version says that Trout has been a worse offensive player this year than Miguel Cabrera (-0.9 dWAR) and the same as Chris Davis, and overall tied for 6th worst in the major leagues.

 

Now I've probably watched 12-15 games this year. I live on the east coast so don't get the chance to watch the Angels as much as I'd like to. But what I've seen is not a bad defensive fielder as rWAR says he is, not even an average fielder as fWAR says, but a very good one. What gives? I mean even if we account for year-to-year fluctuation, I can accept Fangraphs but not B-R. Or am I being a homer? Is Trout a bad fielder and I'm completely deluded?

 

By the way, its the other way for Peter Bourjos. Fangraphs has him at -3.1 Fld and B-R has him at 0.0 dWAR. Both really liked him in previous years so for whatever reason he's not coming out very well this year, albeit in only 40 games.

[halomatt]

isn't rWAR just Baseball Reference's version and fWAR Fangraphs's version? That's what I thought. I read somewhere that Trout's fielding numbers are down because he is not getting those "over-the-fence takeaways" that he was getting last year. My understanding is that both systems give more defensive credit for those catches.

In Bourjos's case he probably has not played enough to escape SSS issues, but also he may not have made enough of "those" catches to I crease his value.

I like WAR as a catch all but the subjectivity of the defenseive measurement does give pause for thought and critique.

[Mudville]

If over the fence catches are a part of the factoring in of defensive capabilities then the entire stat needs to be abandoned because it is true fantasy baseball. The instances where a fielder even has a remote chance to bring one back is slim and the success rate is nothing to base a fielders capability on since each is a case by case basis as to whether the play had a 5% or 80% chance of being accomplished.

[Docwaukee]

bref uses Total Zone for the D component of WAR. 

 

fangraphs uses UZR. 

 

UZR is better.

[Docwaukee]

If over the fence catches are a part of the factoring in of defensive capabilities then the entire stat needs to be abandoned because it is true fantasy baseball. The instances where a fielder even has a remote chance to bring one back is slim and the success rate is nothing to base a fielders capability on since each is a case by case basis as to whether the play had a 5% or 80% chance of being accomplished.

but if the fielder accomplished it, then they should get credit for it. 

just like a hitter gets the same credit for a bunt single or a line drive that is hit so hard that it bounces of the top of the wall, right to a fielder and ends up a single. 

[eaterfan]

If over the fence catches are a part of the factoring in of defensive capabilities then the entire stat needs to be abandoned because it is true fantasy baseball. The instances where a fielder even has a remote chance to bring one back is slim and the success rate is nothing to base a fielders capability on since each is a case by case basis as to whether the play had a 5% or 80% chance of being accomplished.

 

They are certainly factored in, every defensive chance is factored in. They aren't weighted any more heavily than dropping a can of corn, but it's why defensive metrics aren't as reliable on a year to year basis as offensive metrics but they are improving each year. A 3 year sample is really a more accurate reading for defensive metrics because the number of chances is pretty small. That being said those metrics tell a heck of a lot more than just looking at errors.

[Angelsjunky]

bref uses Total Zone for the D component of WAR. 

 

fangraphs uses UZR. 

 

UZR is better.

 

OK, but why? How do TZ and UZR differ?

[Chimi]

If I remember correctly, TZ is calculated using score books that score keepers use while UZR is a bit more advanced and uses electronic score cards like Gameday that can give more precise data. Gameday will be more exact since a computer can plot something a lot better than someone watching the game and recording it manually in a book.

I might be mixed up a bit though so maybe someone can clarify.

[AZMike]

isn't rWAR just Baseball Reference's version and fWAR Fangraphs's version? That's what I thought. I read somewhere that Trout's fielding numbers are down because he is not getting those "over-the-fence takeaways" that he was getting last year. My understanding is that both systems give more defensive credit for those catches.

In Bourjos's case he probably has not played enough to escape SSS issues, but also he may not have made enough of "those" catches to I crease his value.

I like WAR as a catch all but the subjectivity of the defenseive measurement does give pause for thought and critique.

 

If over the fence catches are a part of the factoring in of defensive capabilities then the entire stat needs to be abandoned because it is true fantasy baseball. The instances where a fielder even has a remote chance to bring one back is slim and the success rate is nothing to base a fielders capability on since each is a case by case basis as to whether the play had a 5% or 80% chance of being accomplished.

 

The over the fence catch thing is a big deal. He also got a huge amount of negative credit for trying to rob extra bases and accidentally assisting the other team in getting a home run. Twice this season there were balls that would have hit off the top of the wall that just slipped out of Trout's glove and rolled over the wall. Trout also had a few misplays earlier in the season where balls fell in for doubles that should have been caught.

 

That said I still don't consider the stat legitimate. According to both fangraphs and baseball reference JB Shuck is a better defender than Trout. That is complete BS.

[Mudville]

They are certainly factored in, every defensive chance is factored in. They aren't weighted any more heavily than dropping a can of corn, but it's why defensive metrics aren't as reliable on a year to year basis as offensive metrics but they are improving each year. A 3 year sample is really a more accurate reading for defensive metrics because the number of chances is pretty small. That being said those metrics tell a heck of a lot more than just looking at errors.

Eater, just how does one factor in a ball over the fence? Is it by ball flight measured, height of wall, the players verticle leap, wind and whether a fan can get in  the way?

 

I think this is another made up stat without real facts to properly judge if a player is just making an attempt rather than having a real chance to make a play.

[HaloMagic]

rWAR uses DRS, not TZ.

DRS and UZR both use info from Baseball Info Solutions. Obviously the way they interpret that info differs. I don't know that you could make a claim that one is better than the other without knowing exactly how they work. There are many examples of players whose UZR and DRS numbers remain relatively close to each other every season throughout their entire career. Then there are examples of players whose numbers never come close to matching. And there are a ton of players whose numbers match one year and are wildly different the next. It's just not easy to use defensive statistics in any way and assume they are reliable.

[HaloMagic]

That said I still don't consider the stat legitimate. According to both fangraphs and baseball reference JB Shuck is a better defender than Trout. That is complete BS.

They claim that Shuck has fielded the ball better, not that he is a better fielder. I can't say I agree with that but I think it's an important distinction to make. Defensive stats don't try to tell you who is better, only who did better.

[eaterfan]

Eater, just how does one factor in a ball over the fence? Is it by ball flight measured, height of wall, the players verticle leap, wind and whether a fan can get in  the way?

 

I think this is another made up stat without real facts to properly judge if a player is just making an attempt rather than having a real chance to make a play.

Generally they measure trajectory of the ball and velocity off the bat and go back through the data and say x% of the balls hit like that were turned into outs. No one knows the exact formulas they use but obviously it is better if you are making plays that very few fielders are making. Robbing HRs is a play that few fielders make. This does lead to some problems in that the fielders starting position isn't taken into account so teams that shift a lot can lead to weird anomalies. But being in the right position is part of being a good fielder. Also outfield shifts aren't as variable generally as infield shifts.

[halomatt]

They claim that Shuck has fielded the ball better, not that he is a better fielder. I can't say I agree with that but I think it's an important distinction to make. Defensive stats don't try to tell you who is better, only who did bett

I agree with this.  It's important to know what limitations of a statistic exist and, therefore, what it's exact purpose is.  A poorly defined statistic will not pass the valid or reliable checks.  For example, ERA is becoming more and more of a useless statistic because of many, uncontrolled variables that "muddy the waters" of just what an "Earned Run" is and who "allows" that earned run to exist.

 

Also, I'm not sure anyone who wants to be taken seriously by the community they are speaking to would "make up" a statistic.  Any researcher worth his/her salt would take great pains to make sure their statistic is both reliable and valid.  To do otherwise would risk discrediting their work by the larger community. 

 

[Jason]

This thread is nerd approved.

[seenoevil]

^