Statnerds Only: Which WAR do you prefer?

[Angelsjunky]

I've generally preferred Fangraphs, but this chart makes me reconsider. First of all, while I'm not a huge fan of WAR for pitchers, I like the fact that they base theirs on Runs Allowed rather than FIP. This takes into account the crafty types like Tom Glavine and Jered Weaver (when he was good) who manage to get outs without blowing everyone away.

Secondly, I like the fact that the hitting is based on the separate leagues and doesn't take into account pitchers. fWAR incorporates pitchers which would seem to give NL hitters a boost.

So what say you?

[Angelsjunky]

BR's WAR also seems to like Trout's defense better. Fangraphs WAR sees him as about average, above in 2012 and below in 2014. But this fails the eye-ball test, because I don't think you'll find a person who watches him field and sees him as less than good.

[tdawg87]

This ought to be a productive thread.

I prefer fWAR because defensive WAR is just stupid.

[Second Base]

Defensive metrics just don't match up with the eye test, and do for me, I still kick it old school and monitor throwing errors, FLD% and most of all, just what I see, 

In terms of pitching, I use FIP, oversll but I really like to monitor statcorner's top section that indicates hoe frequently a pitcher who is ahead or behind in the count. Needless to say, pitchers that are constantly ahead in counts are the most successful.

As far as pure offensive numbers, OPS+ is the standard right now. But if this is simply bbref vs FanGraphs, I give it to FG.

[Stradling]

I'm not a statnerd, but I'm willing to learn.  That being said BR is just easier to read and a better website.  FG seems busy and less user friendly, but I say that knowing its ignorant because I haven't tried to use it as much.  I have been looking at FG more but it's really just to compare the two.  

[AngelsLakersFan]

1 hour ago, Angelsjunky said:

I've generally preferred Fangraphs, but this chart makes me reconsider. First of all, while I'm not a huge fan of WAR for pitchers, I like the fact that they base theirs on Runs Allowed rather than FIP. This takes into account the crafty types like Tom Glavine and Jered Weaver (when he was good) who manage to get outs without blowing everyone away.

Secondly, I like the fact that the hitting is based on the separate leagues and doesn't take into account pitchers. fWAR incorporates pitchers which would seem to give NL hitters a boost.

So what say you?

One could argue that Weaver beating his FIP is value that should go to the park and defense. Run based WAR gives pitchers credit for for good / bad defense, while FIP based WAR filters out sequencing and good / bad luck. 

Also if you look closely, FG calculates hitting based on MLB average, with pitchers included, so NL hitters aren't seeing any boost. BR compares hitting to individual league average (without pitchers), so if the NL has a better batting line, then hitting is worth more in the AL. FG also makes a league adjustment.

[AngelsLakersFan]

49 minutes ago, ScottyA_MWAH said:

As far as pure offensive numbers, OPS+ is the standard right now. 

Use wRC+

OPS+ is just old school on base percentage and slugging percentage. wRC+ is the actual average run values of individual offensive events.

[AngelsLakersFan]

1 hour ago, Angelsjunky said:

BR's WAR also seems to like Trout's defense better. Fangraphs WAR sees him as about average, above in 2012 and below in 2014. But this fails the eye-ball test, because I don't think you'll find a person who watches him field and sees him as less than good.

I view him about average. He makes some SPECTACULAR plays, but he plays deep and I think he lets a lot of balls drop in, and clanks a few too many near the wall. His overall value really depends on the amount of spectacular plays he makes.

[Oz27]

Baseball Prospectus has the best pitcher WAR and oWAR as far as I am concerned. Their dWAR is inferior to FanGraphs but BP probably has the best complete package. Their oWAR is based on true average and pitcher WAR is centered around Deserved Run Average, which I think are the best indicators of a player's true talent level and are the best way to judge whether a player is likely to sustain their performance level. (Edit: Holy shit, I haven't checked Jered Weaver's BP WARP in awhile ... it is minus-5...he is five wins worse than a replacement player) They also build in some really important things to their overall WAR measurement which others ignore, such as framing. They're pioneering some fascinating game calling statistics for catchers as well which are fascinating and I'm sure we will see those built into their WAR calculation in the near future. I absolutely love their work overall, even if their non-catcher defensive metrics are probably inferior to Fangraphs and BR.

I've grown to hate BR's runs allowed/9 version of pitcher WAR. I know the Fangraphs FIP model isn't perfect because there are guys who constantly beat their FIP expectation (as Weaver did for a long time) but the BR version gives pitchers credit or blame for things which are completely beyond their control and that is just silly. BP is better than both of them but I trust Fangraphs way more for pitcher WAR than I do BR. I definitely believe we should be trying to take out factors beyond a pitcher's control, even if that leads to some problems.

BR's main positives are a really good website (if you don't subscribe to the Play Index, you 100% should) and their oWAR and dWAR are both pretty reliable. While wRC+ is better than OPS+, it's close enough where I trust what OPS+ tells me. I'm not here to bash BR at all, I love what they do and it is probably my most visited website. But my WAR power rankings are BP, then FanGraphs, then BR.

[Oz27]

Also, can we take bets on which anti-sabermetric poster is the first to chime into this thread with something really stupid?

[Troll Daddy]

1 hour ago, Oz27 said:

Also, can we take bets on which anti-sabermetric poster is the first to chime into this thread with something really stupid?

 

[Troll Daddy]

 

Will somebody please let me know when Alberts get to 1 game over replacement value. 

[Oz27]

Okay, really it was obvious who was going to be the first person to chime into this thread with something incredibly stupid. There was no need to take bets.

[tdawg87]

Yes but to be fair you baited him. And bated him.

[AngelsLakersFan]

14 minutes ago, tdawg87 said:

Yes but to be fair you baited him. And bated him.

I'm pretty sure he only baited him. He came on his own...

[Taylor]

Whichever one makes Trout look better.

[Lhalo]

2 hours ago, AngelsLakersFan said:

I view him about average. He makes some SPECTACULAR plays, but he plays deep and I think he lets a lot of balls drop in, and clanks a few too many near the wall. His overall value really depends on the amount of spectacular plays he makes.

He plays where he's told to play. It's hard to credit or discredit an outfielder on his positioning when he's getting constant instructions from the dugout. I agree with you about the misplays near the wall though. 

[kevinb]

9 hours ago, Oz27 said:

Baseball Prospectus has the best pitcher WAR and oWAR as far as I am concerned. Their dWAR is inferior to FanGraphs but BP probably has the best complete package. Their oWAR is based on true average and pitcher WAR is centered around Deserved Run Average, which I think are the best indicators of a player's true talent level and are the best way to judge whether a player is likely to sustain their performance level. (Edit: Holy shit, I haven't checked Jered Weaver's BP WARP in awhile ... it is minus-5...he is five wins worse than a replacement player) They also build in some really important things to their overall WAR measurement which others ignore, such as framing. They're pioneering some fascinating game calling statistics for catchers as well which are fascinating and I'm sure we will see those built into their WAR calculation in the near future. I absolutely love their work overall, even if their non-catcher defensive metrics are probably inferior to Fangraphs and BR.

I've grown to hate BR's runs allowed/9 version of pitcher WAR. I know the Fangraphs FIP model isn't perfect because there are guys who constantly beat their FIP expectation (as Weaver did for a long time) but the BR version gives pitchers credit or blame for things which are completely beyond their control and that is just silly. BP is better than both of them but I trust Fangraphs way more for pitcher WAR than I do BR. I definitely believe we should be trying to take out factors beyond a pitcher's control, even if that leads to some problems.

BR's main positives are a really good website (if you don't subscribe to the Play Index, you 100% should) and their oWAR and dWAR are both pretty reliable. While wRC+ is better than OPS+, it's close enough where I trust what OPS+ tells me. I'm not here to bash BR at all, I love what they do and it is probably my most visited website. But my WAR power rankings are BP, then FanGraphs, then BR.

This might sound dumb but what do you mean by out of a pitchers control? 

[Oz27]

16 minutes ago, Kevinb said:

This might sound dumb but what do you mean by out of a pitchers control? 

Primarily balls in play. Stats like ERA allow a pitcher to be hurt or hindered by the quality or lack of it of the defense behind them, which obviously they have no control over and therefore should not be judged by. There is also luck in whether balls in play fall for hits or outs and that means statistics like ERA can make a pitcher look much better or worse than they actually are. While a statistic like FIP has its flaws, it takes away most of the elements which a pitcher doesn't have control of.

[kevinb]

3 minutes ago, Oz27 said:

Primarily balls in play. Stats like ERA allow a pitcher to be hurt or hindered by the quality or lack of it of the defense behind them, which obviously they have no control over and therefore should not be judged by. There is also luck in whether balls in play fall for hits or outs and that means statistics like ERA can make a pitcher look much better or worse than they actually are. While a statistic like FIP has its flaws, it takes away most of the elements which a pitcher doesn't have control of.

So just a really fast Google search but FIP calculate home runs walks and strike outs? It doesn't account for base hits? I'm confused? I can see how that a good starting point but shouldn't base hits/ wild pitches/ pitchers errors be in that as well? I don't know how you'd calculate any of that but ya haha. Sorry for the random questions. I definitely am not a advanced stat guy but trying to learn

[Oz27]

2 minutes ago, Kevinb said:

So just a really fast Google search but FIP calculate home runs walks and strike outs? It doesn't account for base hits? I'm confused? I can see how that a good starting point but shouldn't base hits/ wild pitches/ pitchers errors be in that as well? I don't know how you'd calculate any of that but ya haha. Sorry for the random questions. I definitely am not a advanced stat guy but trying to learn

FIP is fielder independent pitching, so it factors in every outcome which does not involve a fielder (walks, home runs and strikeouts). Much of the thinking behind it is that pitchers have very little control over the outcome on balls in play. Batting average on balls in play is almost always around .300 and a significant variation from that is often luck (you might see people talking about a high BABIP for a hitter, which would often indicate that hitter has been lucky and will soon regress to the mean). But yeah, the thinking is that to judge a pitcher's true talent level we should take out factors he has no control over and that is especially possible due to there being little variation in BABIPs over time.

Having said that, there are some pitchers who constantly outperform their FIP. Jered Weaver has always been one of them. If someone does that over several seasons, it's impossible to not acknowledge that person has a skill which is beneficial in that regard. In Weaver's case, he has always generated fly balls (and pop ups) to a higher than average rate and has spent his whole career playing in a park where it is hard to hit a home run - so a lot of those fly balls stay in the park and end up as outs. There are others who beat their FIP and it is a stat which has some flaws, but there aren't many better indicators of a pitcher's true talent level.

[kevinb]

29 minutes ago, Oz27 said:

FIP is fielder independent pitching, so it factors in every outcome which does not involve a fielder (walks, home runs and strikeouts). Much of the thinking behind it is that pitchers have very little control over the outcome on balls in play. Batting average on balls in play is almost always around .300 and a significant variation from that is often luck (you might see people talking about a high BABIP for a hitter, which would often indicate that hitter has been lucky and will soon regress to the mean). But yeah, the thinking is that to judge a pitcher's true talent level we should take out factors he has no control over and that is especially possible due to there being little variation in BABIPs over time.

Having said that, there are some pitchers who constantly outperform their FIP. Jered Weaver has always been one of them. If someone does that over several seasons, it's impossible to not acknowledge that person has a skill which is beneficial in that regard. In Weaver's case, he has always generated fly balls (and pop ups) to a higher than average rate and has spent his whole career playing in a park where it is hard to hit a home run - so a lot of those fly balls stay in the park and end up as outs. There are others who beat their FIP and it is a stat which has some flaws, but there aren't many better indicators of a pitcher's true talent level.

Hmm ok cool thanks. Makes sense. 

[tdawg87]

FIP is nice and everything but then Blanton throws his sweaty, soggy, meat-wrench into it.

[Docwaukee]

the purpose of WAR isn't necessarily to tell us how good or bad a player is, but to accurately depict the value of events while drowning out the noise.  the ultimate goal, of course, is to predict the future.  the correlation of WAR to actual wins by regression analysis is pretty similar across all versions.  Close enough anyway that all of them have their merits and flaws.  But the correlation is never 1.0, and it never will be.  So I really have no preference.  What I find interesting is trying to figure out what accounts for those fractions following the first decimal point.  Sometimes it makes sense.  Other times it doesn't.  What might be noise for one player, is a repeatable skill for another.  The thing that I try to keep in mind is that regardless of how much info we think we have, each of the individual teams have access to far more robust data sets.  

At some point, it levels off to where the win correlation you can obtain by any given stat is enough decimal places to the right that it ceases to become useful or an undervalued piece of information that you can act on.  In essence, every team is going to use their proprietary method to hone it down as far as it can go.  Then it's going to boil down to the ability to make up those decimal points via something that isn't measurable.  Or at least not yet.  And then, it becomes a great deal about luck, timing, and playing the man.  

Those things are why they play and it's the most interesting part.  

 

[Oz27]

I heard a fun exercise the other day, involving players with wildly different WAR values by the three main formulas (Baseball-Reference, Fangraphs and Baseball Prospectus). This is the least scientific thing ever because it plays into all of our unsubstantiated biases. But I'll give three WAR values for the same player and it might give you some idea of which WAR model you most closely identify with.

Jered Weaver: -0.1 (Fangraphs), -0.9 (BR), -5.2 (BP)

Andrelton Simmons: 1.3 (BP), 1.9 (Fangraphs), 2.7 (BR)

Matt Shoemaker: 2.0 (BR), 3.0 (BP), 3.3 (FG)

Noah Syndergaard: 4.4 (BR), 4.7 (BP), 5.7 (Fangraphs)

Buster Posey: 3.4 (FG), 3.8 (BR), 7 (BP)

Yasmani Grandal: 2.3 (BR), 2.8 (FG), 5.8 (BP)

Bryce Harper: 2.3 (BR), 3.8 (FG), 4.2 (BP)

Jason Heyward: -0.4 (BP), 1.0 (BP) 1.3 (FG)

Sonny Gray: -0.3 (BR), 0.7 (FG), 1.9 (BP)