Jump to content
  • Welcome to AngelsWin.com

    AngelsWin.com - THE Internet Home for Angels fans! Unraveling Angels Baseball ... One Thread at a Time.

    Register today to join the most interactive online Angels community on the net!

    Once you're a member you'll see less advertisements. Become a Premium member and you won't see any ads! 

     

Jeff Passan: Why WAR doesn't always add up


jsnpritchett

Recommended Posts

Good article, I agree with Passan - limitations with defensive metrics make WAR rather dubious because of the way defense is calculated and included. I think Passan makes a good point, that defense should still be part of WAR, just much lower impact. I'd even go so far to say that halving the value of defense in WAR would give a more accurate read.

 

What GM would rather have Alex Gordon (.275/.353/.452, 6.1 fWAR) than Giancarlo Stanton (.291/.396/.557, 5.7 fWAR)? Or look at how much Victor Martinez--who has been the best hitter in baseball this year (.335/.405/.575, 30 HR)--has been penalized by being a DH (4.1 fWAR, -17.3 Def) compared to Josh Donaldson (.253/.343/.457, 5.6 fWAR, +14.2 Def).

 

Again, reducing defensive values - negative and positive - bridges the gap. If we halved Donaldson's and Martinez's defense, they'd meet in the middle around 4.8-5.0 WAR, which I think is more accurate. It still gives Donaldson props for being the great defensive third baseman that he is, but it also gives Martinez props for being the best hitter in baseball, and far better than Donaldson.

Link to comment
Share on other sites

​Pretty sure that's not true.

IT's definitely not true. An inside the park home run would never be assumed so the number of "runs saved" by the catch would, be definition, be less than for a home run robbing catch. 

 

It is true though that WAR is context neutral, meaning the score during teh game woudln't be considered.

Link to comment
Share on other sites

Good article, I agree with Passan - limitations with defensive metrics make WAR rather dubious because of the way defense is calculated and included. I think Passan makes a good point, that defense should still be part of WAR, just much lower impact. I'd even go so far to say that halving the value of defense in WAR would give a more accurate read.

 

What GM would rather have Alex Gordon (.275/.353/.452, 6.1 fWAR) than Giancarlo Stanton (.291/.396/.557, 5.7 fWAR)? Or look at how much Victor Martinez--who has been the best hitter in baseball this year (.335/.405/.575, 30 HR)--has been penalized by being a DH (4.1 fWAR, -17.3 Def) compared to Josh Donaldson (.253/.343/.457, 5.6 fWAR, +14.2 Def).

 

Again, reducing defensive values - negative and positive - bridges the gap. If we halved Donaldson's and Martinez's defense, they'd meet in the middle around 4.8-5.0 WAR, which I think is more accurate. It still gives Donaldson props for being the great defensive third baseman that he is, but it also gives Martinez props for being the best hitter in baseball, and far better than Donaldson.

 

There is a guy who is working on a WAR formula using a weighted three year defensive sample to try to negate some of the wildly up and down numbers that defensive metrics can see from yeah to year.   Something like that would still be iffy for young or old players but might be less volatile than the current versions of WAR.

Link to comment
Share on other sites

They need to include error bars in the stat for each player. Guys whose value is derived mostly from defense will have a higher +/- X wins next to their calculation.

 

I have a lot of issues with WAR but I think this article pointed out a couple of positives. It talked about the absurd defensive ratings for Gordon and Heyward, two guys who have seriously impressed me with their defense when I've watched them play, perhaps more than any others.

 

It pointed out the differences between Bumgarner and Vargas and how the difference in WAR all comes down to league and park adjustments. That's not a bad thing. The AL is tougher, KC is harder to pitch in. These things count.

 

In the same comparison it noted the difference in Vargas fWar and rWar and mentioned how Vargas is benefiting from Gordon's (and others) defense. The point missed is that fWAR is giving credit for the defense to Gordon and others, where as rWar is giving it to Vargas. This is an accounting thing, a matter of opinion. Each one takes a different side, and this is why there are two version of the stat. 

 

The article also goes on to erroneously mention that FIP derived fWAR does not factor in innings pitched. This could not be further from the truth. Innings pitched is one of the four components of FIP, strikeouts, walks, homers and innings pitched.

 

Passan talks a lot about shifts, and with how new they are we haven't really figure out how to take them into consideration, wishing for a future with accurate data on off the bat velocity, route efficiency and closing time. This misses the forest for the trees. The whole point of defense is to MAKE THE PLAY, nothing else matters. If it's an easy play or a hard play, when we are discussing value, the only thing we should care about is making the play. Positioning is important, if a fielder knows exactly where a ball is going to be hit before hand that is no less valuable than making a tremendous play. The question here becomes, who gets credit for the positioning, and how exactly do we calculate it?

 

One last thing, WAR really needs to fix it's DH calculation. There is something wrong with the way DH value is added. And while we are on the topic, why do we not see pitcher BATTING WAR included? The replacement level must be so low that it'd be easy for decent hitting pitchers to be adding significant value to their offensive and defensive WAR totals.

Link to comment
Share on other sites

There is a guy who is working on a WAR formula using a weighted three year defensive sample to try to negate some of the wildly up and down numbers that defensive metrics can see from yeah to year.   Something like that would still be iffy for young or old players but might be less volatile than the current versions of WAR.

 

But what exactly is the point of a WAR calculation for 2014 that is using data from 2012 and 13?

 

One major issue with defensive stats is that they are regressed. Players are making plays in the field that count for real actual outs and the stats are hedging their bets on just how much credit they should be given for making a play.

Link to comment
Share on other sites

They are using scouts to track hits with stop watches. Does anyone believe this has any true accuracy for using as a weighted measurement?

 

Currently defensive WAR is simply a work in progress. One day it will become relevant but for the most part it is for testing purposes only and doesn't reveal any true value of a players performance in the field.

Link to comment
Share on other sites

They are using scouts to track hits with stop watches. Does anyone believe this has any true accuracy for using as a weighted measurement?

 

Currently defensive WAR is simply a work in progress. One day it will become relevant but for the most part it is for testing purposes only and doesn't reveal any true value of a players performance in the field.

 

I believe there is huge potential for serious over and under estimation, but I do not believe for a second that it is not getting most things right. We can argue till the cows come home about whether or not Heyward and Gordon deserve THAT much credit for their defense, but WAR tellings us that they are elite defenders is not an accident.

Link to comment
Share on other sites

I'm pretty sure UZR and DRS don't calculate home run saving catches the same way. Saving a home run has a lot to do with chance and opportunity. Sure, it requires timing and athletic ability, too... but you also need to have luck on your side. You can't save/rob a home run unless the opportunity presents itself to you. Remember all of Trout's catches? That had a lot to do with opportunity than just pure skill

Another caveat: some stadiums don't make it possible to save/rob a home run. A LF or CF in Boston, for example, has no chance to rob a HR. It wouldn't be fair.

Link to comment
Share on other sites

A better defensive WAR calculation would have to be obtained by considering defensive shifts. Brett Lawrie for example got a lot of defensive credit last year and this year but he was part of the defensive shift where he was essentially in the SS position.

 

I also think that a run prevented is almost as good as a run scored. I say almost because you have to score at least one run to win a game. I think changes do need to be made to obtain a more credible WAR stat but I'd be careful about devaluing defensive WAR too much.

Link to comment
Share on other sites

But what exactly is the point of a WAR calculation for 2014 that is using data from 2012 and 13?

 

One major issue with defensive stats is that they are regressed. Players are making plays in the field that count for real actual outs and the stats are hedging their bets on just how much credit they should be given for making a play.

 

Yeah -- that's the guy's own issue with the numbers and why he doesn't really know how to move forward.  IMO defensive metrics are still a work in progress.  My guess is that as technology advances and more and more cameras are set up to track batted balls we may eventually have a more consistent methodology.  Until then it's going to be a bit hit and miss but even as it stands, defensive metrics help identify elite defenders.

 

Really enjoyed your previous post in this thread. 

Edited by Inside Pitch
Link to comment
Share on other sites

The article about Gordon getting more defensive credit because his peers are worse nails it for me.  The year to year defensive metrics will always suffer from small samples because it's divided by 9.

 

The real question is what's the point?  

 

Is it the necessity of being able to use a comprehensive metric to compare individual players?  To some degree, but really, it's a context neutral event log that has a goal of trying to predict the future.  It assumes that all triples are equal (for now).  

 

While I think it makes sense to assume such neutrality for any given event, the only reason it's being done that way is because if you tried to compensate for the situation, you'd be adding a variable that is almost impossible to account for.  But isn't that what ultimately makes baseball so great?  No matter how much analysis and no matter how detailed your metrics are, there will always be things that can't be accounted for.  

 

So you wash out the one thing that matters most because you can't control it and you try to formulate a system whereby all things being equal you have some semblance of what should happen.  

 

WAR is one of the better ways developed so far to account for this.  On an individual level it's far from perfect and it never will be.  Thank God or why bother playing.  But the goal is to account for the events in such a way that in the end you can attempt to control the situation as much as possible.  

 

I thought Passan's article was well written and useful to some degree, but ultimately misses the forest for the trees.  Anyone from Fangraphs or BR doing a regression analysis on the Royals and what is accounting for their wins can look at Gordon's WAR and see holes.  Sure it is important in making WAR a better stat for the future, but I can guarantee they aren't anointing him the best player in baseball.  

 

In the end, the point is to control as much as you can statistically as possible while knowing that there are other factors involved that are subject to human nature and require gestalt.   

Link to comment
Share on other sites

I believe there is huge potential for serious over and under estimation, but I do not believe for a second that it is not getting most things right. We can argue till the cows come home about whether or not Heyward and Gordon deserve THAT much credit for their defense, but WAR tellings us that they are elite defenders is not an accident.

 

This is where I have to say, anyone can spot an elite player. Where advanced metrics really come into play is trying to parse the numbers to determine how much you can squeeze out of second and third tier players to fill out a roster. Is Aybar really better than Elvis Andrus. Neither are Derek Jeter in his prime so which do you go with for the money? It doesn't take any brains at all to say take Jeter in his prime over the both of them.

Link to comment
Share on other sites

I've been always saying here, that defensive WAR is way too fragile and subjective of a stat to be considered seriously. Saving a HR at the wall is mostly luck. It hugely depends on how many flyballs to the wall your pitcher allows. And what if it was a catch that probably 70% of the OFs make? How much credit should the player get?

 

Also, who do you compare it to?

Usually, center fielders are the best fielding OFs, and short stops are the best fielding IFs.

Why does the CF have to get compared with the best OFs?

Why do LF get compared to the worst fielders? Same issue with the IF.

 

If a CF or a middle IF is getting credit for their defense, I could understand a little.

But a guy that's not good enough to play CF getting credit for their defense because they are better than the worst fielders in the OF?

 

Give me a break.

Same with Donaldson. Andrelton Simmons would get a 7 WAR hitting .200 if he played 3rd base at the spacious A's park.

Link to comment
Share on other sites

I love the idea that lots of different people are looking for ways to best measure the effectiveness of a player.  I just have a hard time wrapping my head around a lot of it because I can't calculate this stuff on my own.  To make it worse - there seems to be different schools of though on how to calculate things like WAR.  It all makes my head explode.  I sincerely wish I understood it better. 

 

When someone like Ben Zobrist leads the league in WAR one year and people try to tell me he's the best player in baseball, I go.. "Huh?"

 

Again, I don't have any issue with what the sabermetrics community is trying to do... but I find it amusing when they argue amongst themselves in trying to determine how to do some of this stuff.

 

I guess it's a work in progress.  Someone wake me when they finally figure it out.

Link to comment
Share on other sites

I've been always saying here, that defensive WAR is way too fragile and subjective of a stat to be considered seriously. Saving a HR at the wall is mostly luck. It hugely depends on how many flyballs to the wall your pitcher allows. And what if it was a catch that probably 70% of the OFs make? How much credit should the player get?

 

Also, who do you compare it to?

Usually, center fielders are the best fielding OFs, and short stops are the best fielding IFs.

Why does the CF have to get compared with the best OFs?

Why do LF get compared to the worst fielders? Same issue with the IF.

 

If a CF or a middle IF is getting credit for their defense, I could understand a little.

But a guy that's not good enough to play CF getting credit for their defense because they are better than the worst fielders in the OF?

 

Give me a break.

Same with Donaldson. Andrelton Simmons would get a 7 WAR hitting .200 if he played 3rd base at the spacious A's park.

 

That's not the whole story though. Yes a left fielder gets compared to other left fielders defensively, but they also get compared to other left fielders offensively (through the replacement level calculation). 

 

So while it is easier to rack up the defensive numbers cited in the article, the full calculation looks a lot different. If you go to fan graphs and look at the WAR page you can see how Gordon and Heyward rate out "defensively" compared to center fielders and other defenders once the replacement level is brought in. (SPOILER ALERT: Heyward still ranks #1, Gordon #9, Simmons #6, Donaldson #10 and notable Angel Erick Aybar #17).

Edited by AngelsLakersFan
Link to comment
Share on other sites

I have a big problem with stats that tell past performance and stats that are supposedly predictive and when they are jumbled up.

For example, I believe that DRS is used in one of the WAR calculations, and I'm told that is contextual. In other words, if you make a diving stop with the bases loaded it saves more than the same play with the bases empty. That has nothing to do with skill. It's about opportunity.

On the other hand, FIP is used for pitchers WAR, and that is essentially a predictive number. It counts a strikeout as better than a groundout because the strikeout is more likely predictive of skill and luck.

So they are mixing up two different types of things and trying to compare them.

Link to comment
Share on other sites

I have a big problem with stats that tell past performance and stats that are supposedly predictive and when they are jumbled up.

For example, I believe that DRS is used in one of the WAR calculations, and I'm told that is contextual. In other words, if you make a diving stop with the bases loaded it saves more than the same play with the bases empty. That has nothing to do with skill. It's about opportunity.

On the other hand, FIP is used for pitchers WAR, and that is essentially a predictive number. It counts a strikeout as better than a groundout because the strikeout is more likely predictive of skill than luck.

So they are mixing up two different types of things and trying to compare them.

By the way, I don't think either kind of stat is better. You just need to use them for the right things. For award voting, for example, I'm more interested in what really happened than how much skill was involved. For player acquisition, the opposite.

Edited by Jeff Fletcher
Link to comment
Share on other sites

By the way, I don't think either kind of stat is better. You just need to use them for the right things. For award voting, for example, I'm more interested in what really happened than how much skill was involved. For player acquisition, the opposite.

 

So did you think that Cabrera deserved the MVP over Trout the past two seasons?

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...