I'm seriously not clear why anyone gives a damn about the MVP and the Cy Young.

I don't care about the result. I care about the process, the methods, and the arguments. I suspect I am far from alone in this regard.

I may have another thought - if Anaheim is playing like a pitcher's park this year for whatever reason (wind patterns, temperature, etc)

Stop saying that, because it's wrong. Anaheim 1 year park factors:

2012 - 92/91
2011 - 93/93
2010 - 94/94
2009 - 102/102
2008 - 100/100

It's been trending down for sure, but it's not a one year fluke.

Because they're using it as a strawman instead of engaging in an intelligent debate about Cabrera vs. Trout. Almost no one is arguing that Trout should be MVP simply because he has leads in WAR.

No, but they're arguing that Trout has such a big lead in WAR that Cabrera can't possibly be the MVP. That's basically tantamount to basing your entire conclusion on WAR.

EDIT: And the method of calculating the stat keeps changing, sometimes markedly - which is fine, and necessary - but it adds to the confusion. I just realized last night in reading it that the calculations have changed such that oWAR and dWAR no longer add up to WAR.

My biggest beef right here. Why can't I calculate WAR on my own? I've tried on several occasions over the years to create my own spreadsheet to calculate WAR on my own database of players, it just can't be done using the documentation available on line. The calculations are too transient and explanations too vague. For all Wins Shares failings at least Bill took the time to walk through, step by step, exactly how to calculate the thing so you could try it for yourself and make up your own mind as to how well it works (in Win Shares case, not that well). He used one of McGwire's monster seasons as an example, give me (I love demanding people do stuff for me for free) a Darwin Barney example so I can figure out what is a feature and what is a bug.

It's been trending down for sure, but it's not a one year fluke.

Why did it drop so precipitously in 2010?

I've heard people at ballparks talking about WAR and I assume most of them saw it on either MLB Network or ESPN. I can't imagine it's really explained well when it's shown on tv.

It's not that hard to explain. One sentence is all that is needed "a stat that takes all a players contributions, hitting, baserunning, defense and park adjustments, and show how much better than a freely available player he is"

You do not need to go into details about the definition of a replacement player, or talk about win value in explaining the stat. That is geek talk, basically it's a number that basically combines all the possible contributions a player brings to the table into one number. .

No kidding - was there a remodel in 2010? I don't think the rest of the league changed drastically.

"Here's the biggest problem with the wins against replacement argument: Even the people who invented the 'stat' can't agree on how to calculate it!" - Problem 1 is the sub-head of the article. Admittedly the fact that there are two stats titled the same thing is confusing but while it's a fair question, it's one a phone call or an e-mail probably could have helped clear up.

But you can be a perfectly competent baseball writer and commentator without making that phone call or sending that e-mail. It's very arrogant to think otherwise. The output of the model is often bizarre, and the methodology is frequently changing and controversial.

SugarBear has another valid point here. Why is it important that Madden didn't call or email Sean? If I have an issue with WAR after trying to figure it out, I don't call or email Sean - I discuss it with others in the community.

Also, Sean's explanation of WAR would be more useful if, when he explained that he's doing X, he also explained what the criticisms of doing X are or would be. It's very math-heavy, and so to expect Madden (or frankly most people) to have enough knowledge to figure out what the problems with doing X are is a bit much.

Why did it drop so precipitously in 2010?

#### happens?
Tropicana Field since 2005
100
100
100
101
97
94
92
93

From 2001-2003 Kauffman Stadium had a run of 107, 111, 108 before returning to it's roughly established 102-103 level.

2006-2008 Arlington played almost neutral amid a ten year run of being a hitter's park.

Park effects are pretty fluid I think, which is why a 3 year park effect is a good idea.

Even the people who invented the 'stat' can't agree on how to calculate it!"

Well Bill James came up with the concept and settled on the level that expansion teams have historically played at. Keith Woolner went to the bother of calculating how well backups actually played. Didn't change thing a lot. Then Keith went to the bother of calculating where replacement level should be assuming baseball talent is normally distributed (and only the extreme right segment even gets as far as A ball)

One thing I urge people who are distrustful of where replacement level is to just look at the Zips projections before the season starts. Eliminate the starters and the guys who you can send to the minors without having to clear waivers. (Plus the really good backups. David Ross is not freely available for instance) That's basically the pool that you're picking from if you want a "no real cost" replacement.

Every now and then a guy who can flat play gets caught behind better players and they really shouldn't be seen as part of the pool, but it's not a bad place to start if you want your own starting point.

Stop saying that, because it's wrong.

I think it is definitively true that Angels Stadium is playing like a pitcher's park this year. For it to be wrong it would have to be playing like a hitter's park, which your data contradicts.

It's not that hard to explain. One sentence is all that is needed "a stat that takes all a players contributions, hitting, baserunning, defense and park adjustments, and show how much better than a freely available player he is"

It's probably useful to also mentioned the unit being used (wins via runs). I imagine the biggest problem with WAR to the average fan is the perceived lack of relation to what's happening on the field.

Hamilton just hit a two-run double, did he just accumulate .05 WAR?

a 3 year park effect is a good idea

In which case we can figure out the WAR title after next season.

In which case we can figure out the WAR title after next season.

But I've already contracted the engraver!

Okay, point number 1: Can people making the case for Trout stop referring to him as a "Gold-Glove center fielder?" He's only played 871 innings in CF (fewer than 100 games equivalent).

Re: Anaheim Park Factor:

Over the last 10 years, Anaheim has mostly played as a neutral park. From 2002-2010, it was between 96 and 102 every year. Despite no new parks in the AL west and no changes to the Angels' park during this time frame, it is now playing approximately 8% tougher to hit in for no apparent reason. That seems fishy. Why should we presume that the 2011-2012 data for Anaheim stadium is more accurate than the larger data series?

Re: Trout's performance in general:

While it may not be necessary to use WAR when discussing Trout v. Cabrera in 2012, Trout's performance does call the metric into question for me.

Trout in 2012 is at 10.5 WAR in 136 games. Bonds in 2001 was at 11.6 WAR in 153 games. WAR wants me to believe that Trout's 2012 is more valuable than Bonds' 2001.

Is there anyone who cares to make that argument without using WAR?

a 3 year park effect is a good idea



In which case we can figure out the WAR title after next season.

Well, what's the solution then? Just throw up your hands and say we can't calculate it perfectly so why bother?

It's not that hard to explain. One sentence is all that is needed "a stat that takes all a players contributions, hitting, baserunning, defense and park adjustments, and show how much better than a freely available player he is"

You do not need to go into details about the definition of a replacement player, or talk about win value in explaining the stat. That is geek talk, basically it's a number that basically combines all the possible contributions a player brings to the table into one number. .

Disagree. Nobody would have a problem with your overall definition about - dare I say, not even Madden. The entire argument is over the "how," not the what.

One thing I urge people who are distrustful of where replacement level is to just look at the Zips projections before the season starts. Eliminate the starters and the guys who you can send to the minors without having to clear waivers. (Plus the really good backups. David Ross is not freely available for instance) That's basically the pool that you're picking from if you want a "no real cost" replacement.

???? Is anyone on here complaining about the definition of replacement level? That is just a baselevel that can be changed willy nilly at any time, it shouldn't be holding people back from accepting or not accepting the stat. I mean even if they make replacement level so low that Trout is posting a 20 win vs Cabrera's 12, it shouldn't stop people from accepting the stat because of where replacement is at. I don't think anyone casual, cares if replacement is accurate, it's about the components being accurate that matter.

As for your argument about replacement level not being obtainable I fail to see how that's meaningful. There needs to be a baseline and if you'd prefer it be average there's a metric for that. Pick your baseline, Wins Above Francoeur? Wins Above Some Guy in Rookie Ball? Fine, the number winds up a bit different but it is the same end result.

How about Wins Above Steve Pearce? He's been the definition of freely available replacement talent this year. He was signed by the Twins as a free agent from the Pirates last winter (December 15). He was released at the end of spring training by the Minnesota Twins (March 27). He then signed with the Yankees and started the season with them (March 29). The Orioles purchased his contract from New York, acquiring him without losing a player in return (June 2). The Astros then claimed Pearce off of waivers from the Orioles (July 28), again acquiring him without losing any of their own resources. The Yankees proceeded to purchase his contract from the Astros (August 27). Just last week, the Orioles grabbed him off of the waiver wire from the Yankees (September 29). Including spring training, Pearce has changed teams six times in the past twelve months: Pirates, Twins, Yankees, Orioles, Astros, Yankees, Orioles. No team had to expend any resources to acquire him (roster players, compensation picks or prospects) or pay anything more than a nominal amount of cash.

Pearce's line this year is .239/.328/.377 for an OPS of .705 and an OPS+ of 92. So a corner infielder/first basemen who hits for an OPS+ of 92 and is an average defender (he's +2 by fielding runs, -0.1 by WAR) is freely available to any team who wants him and has been acquired for free six times in twelve months. The exact specifics may not be the same from one season to the next but the concept of freely available talent is not entirely hypothetical.

Pearce has 0.2 WAR on the season (0.4 for Baltimore, -0.1 each for Houston and New York).

Sean,

I just checked again, and couldn't find this, so:

Does b-r allow for sorting players by oWAR?

If not, why not?

Does b-r allow for sorting players by oWAR?

Yes.

Go down to the list of players, and click oWar

Trout in 2012 is at 10.5 WAR in 136 games. Bonds in 2001 was at 11.6 WAR in 153 games. WAR wants me to believe that Trout's 2012 is more valuable than Bonds' 2001.

Is there anyone who cares to make that argument without using WAR?

Sure.

First of all, the run environment in 2001 was higher than today. 4.7 runs/game in the 2001 NL vs 4.45 in the 2012 AL, a 5.5% decrease.
Second, a sizable chunk of Bond's value was in intentional and semi intentional walks, the value of which is overstated in OPS and OPS+. Bonds walked 177 times, and had 86 non-HR hits and HBP, and scored 56 runs, despite having good hitters batting after him (Kent, Snow, Rios, Galarraga).
Third, by 2001 Bonds wasn't particularly fast, nor a good defender, both of which Trout excels at.

Too lazy to RTFthread, just wanted to ask how many of these guys have unquestioningly incorporated Passer Rating into their articles...

I like WAR, but I do think this comment from the article above is a valid one:

Can you explain what revisions to the formula went into effect within the last year that retroactively reduced Albert Pujols' 2003 WAR from 10.9 to 8.4?

I understand there is some variation as the normalized standards evolve, but 2.5 wins is a massive drop. Both his defense and offense were dropped >1 run.
Incidentally, FanGraphs revised the same 2003 Pujols season from 9.0 to 10.1 WAR.

I'm generally quick to embrace new advanced metrics, but it's hard to put stock in WAR when there's a chance that it will be drastically revised many years after the fact.

I haven't actually looked up this specific example to know whether his numbers are correct or not, but I do remember surfing B-Ref this spring and noticing that a lot of the WAR numbers were different than they used to be, and I admit it is a little hard for me to put too much faith in a stat that fluctuates. Are we confident now that the current numbers are accurate, or are they going to change again in a year or so?

For the record though, I think it's an awesome concept and a great starting point.

Too lazy to RTFthread, just wanted to ask how many of these guys have unquestioningly incorporated Passer Rating into their articles...

That is usually one of the points I like to make. Heck the writers haven't even learned the flaws with the stat and will present the stat while mentioning it's possible flaws.

I think being able to pull up a list of players sorted by oWAR would help, but I don't think b-r allows that.

The P-I does give a pretty good proxy in Batting Wins, based on Palmer's linear weights. Offensive leaders in 2012:

Does b-r allow for sorting players by oWAR?


Yes.

Go down to the list of players, and click oWar

That's a specific 2012 page. I'm talking about doing something like pulling up a list of 2B since 1961 with >1000 games sorted by oWAR. I still can't seem to do it.

it is a little hard for me to put too much faith in a stat that fluctuates

I know you mean "faith" in a weak-colloquial sense, but of course it isn't a matter of faith; everything should be considered critically.

And on your actual point, in many ways I'd have less faith in a metric that never changed. Pancake Flops hasn't changed since it was introduced, and it's more useless all the time. Why won't anyone give us a Pancake Flops 2.0??

That's a specific 2012 page. I'm talking about doing something like pulling up a list of 2B since 1961 with >1000 games sorted by oWAR. I still can't seem to do it.

Not that I have been able to find.

It is the definition of hubris to denigrate WAR and then use RBI as your stat of choice for determining an MVP.

Third, by 2001 Bonds wasn't particularly fast, nor a good defender, both of which Trout excels at.

Bonds had fewer GDP's (5) than Trout (7) in 2001 in more opportunities and ran well (13 stolen bases at 81% success rate). Both of those numbers indicate good speed. Bonds was an average to slightly below average defender, with a weighted average of -5 in UZR.

Second, a sizable chunk of Bond's value was in intentional and semi intentional walks, the value of which is overstated in OPS and OPS+. Bonds walked 177 times, and had 86 non-HR hits and HBP, and scored 56 runs, despite having good hitters batting after him (Kent, Snow, Rios, Galarraga).

Bonds was only intentionally walked 35 times in 2001, which actually didn't lead the league. Even if we changed all of Bonds' intentional walks to outs, he would still have had an OBP 70 points higher than Trout.

First of all, the run environment in 2001 was higher than today. 4.7 runs/game in the 2001 NL vs 4.45 in the 2012 AL, a 5.5% decrease.

Granted.

Bonds was only intentionally walked 35 times in 2001, which actually didn't lead the league. Even if we changed all of Bonds' intentional walks to outs, he would still have had an OBP 70 points higher than Trout.

You missed the point that even the unintentional walks, were frequently "intentional".

You missed the point that even the unintentional walks, were frequently "intentional".

Which is not something considered by WAR.

Pancake Flops hasn't changed since it was introduced, and it's more useless all the time.

Cite. I think it's better than ever.

I know you mean "faith" in a weak-colloquial sense, but of course it isn't a matter of faith; everything should be considered critically.

Yes, I didn't literally mean "faith." "Stock" would've been a better word, but the poster I quoted already used that one so I didn't see the need to repeat it.

And on your actual point, in many ways I'd have less faith in a metric that never changed.

I agree to a point. I think it's a good thing that the creators of the stat are constantly trying to improve it and make it as accurate as possible. But doesn't it also make it a little hard to take at face value when it isn't permanent and can change at any time?

WAR wants me to believe that Trout's 2012 is more valuable than Bonds' 2001.

Is there anyone who cares to make that argument without using WAR?

How about this one.

Babe Ruth 1930. 359/493/732, 211 OPS+, 205 wRC+. Only 10 bWAR and 10.3 fWAR.

Does someone want to claim Trout has been better than that this year?

Does someone want to claim Trout has been better than that this year?

The implication seems to be "just look at that batting line, how could anyone be better than that." But of course:

* it was 1930
* Ruth was not much of an outfielder anymore
* Ruth was 10-for-20 stealing bases
* for reasons probably unknown even to Bob Shawkey, Ruth made 21 sacrifice hits that year, which is a bunch of additional outs – I'm not sure how OPS+ and wRC+ deal with them, but his conventional slash line is thereby not quite as good as it looks
* Ruth still had a fabulous season, 10 WAR to Trout's 10.5. I don't know how much confidence the originating statisticians have in how WAR represents a range of probabilities (they can tell us, I reckon), but what if it's possible that Trout had 10 and Ruth 10.5? Suddenly the question is moot.

Does someone want to claim Trout has been better than that this year?

AL R/G in 1930 - 5.41
AL R/G in 2012 - 4.45, a, 18% decrease. So first reduce Ruth's hitting numbers by 18%, then we can talk speed and defense.

You missed the point that even the unintentional walks, were frequently "intentional".

That actually didn't happen that much in 2001. I watched just about every Bonds AB that year. It didn't really start happening at all until the end of the year and was pretty limited. In 2002, yes, that really happened a lot.

Here's a brief piece of evidence:

In 2001, Bonds batted with zero outs in the inning during 182 PA's. He had a 16.48% walk rate. He batted with two outs in 242 PA's. He had a 26.86% walk rate (a 62% increase).
In 2001, the league as a whole batted with zero outs during 34,321 PA's. The league had a 6.79% walk rate. During 32004 PA's with 2 outs, the league had a 8.99% walk rate (a 75.52% increase).


If Bonds was walked "intentionally" in 2001, it didn't happen that often.

Luckily there's this amazingly cool new stat called OPS+ that's already been posted that shows Ruth at 211 and Trout at 168

And on your actual point, in many ways I'd have less faith in a metric that never changed. Pancake Flops hasn't changed since it was introduced, and it's more useless all the time. Why won't anyone give us a Pancake Flops 2.0??

EqA hasn't changed much, to my knowledge (Ron?), other than the name. And it's still a damned good approximator of rate of offensive value. Including steals, though not non-SB baserunning.

The problem with WAR is that it tries to do too much, and the defensive systems, as improved as they are and getting better, simply are not there yet.

AL R/G in 1930 - 5.41
AL R/G in 2012 - 4.45, a, 18% decrease. So first reduce Ruth's hitting numbers by 18%, then we can talk speed and defense.

Huh? He cited OPS+.

(deleted) pressed submit by mistake.

Luckily there's this amazingly cool new stat called OPS+ that's already been posted that shows Ruth at 211 and Trout at 168.

Ruth in 1930 created 183 runs. Reduce that by 18% and it's 150. Trout this year has created 132. Is Trout's baserunning and defense this year superior enough to a 35 year old Babe Ruth to overcome an 18 run deficit? Very possible. Certainly not a ridiculous notion.

The P-I does give a pretty good proxy in Batting Wins, based on Palmer's linear weights.

But that's still not oWAR, and when Sean said in #56 "If you don't like the defensive numbers we give you the numbers assuming everyone is average defensively" I took that to mean, in a discussion about WAR, that he was giving us oWAR. Which is true, on a player-by-player basis - or even in a specific page like the 2012 page cfb cited - but we can't sort by oWAR in Play Index. And if we can't sort by oWAR in Play Index then the utility of oWAR is hugely limited.

And we do pay for Play Index, so it's not like this is an unreasonable request.

I do think that the capability to sort by oWAR could lead people to focus on oWAR more than WAR, but that's not my problem.

oWAR is essentially a souped up version of VORP - one that includes non-SB baserunning - so there is probably a vested interest on the part of the Seans to push WAR per se and not oWAR. Because there's nothing all that special about oWAR.

What do Trout's numbers look like when you put them in the 1930 machine?

What do Trout's numbers look like when you put them in the 1930 machine?

.362/.438/.627 if playing in Yankee Stadium.

What do Trout's numbers look like when you put them in the 1930 machine?

Using bb-ref neutralized batting you get



That is 1930 as a New York Yankee.

.362/.438/.627 if playing in Yankee Stadium.

So that was like what Jimmie Foxx hit, except with crazy fielding and running. I think he's close enough to Ruth that it's not silly to compare them, which is all that WAR would ask of you anyway.

I give the nod to Ruth in deference to his one start that year, a complete game victory.

Can you explain what revisions to the formula went into effect within the last year that retroactively reduced Albert Pujols' 2003 WAR from 10.9 to 8.4?

I understand there is some variation as the normalized standards evolve, but 2.5 wins is a massive drop. Both his defense and offense were dropped >1 run.

One of the Sean's explained the changes that were made.

A) 2003 is the first year of BIS defensive numbers. TotalZone liked Albert's defense in LF (+10), BIS didn't (-4). I think it's likely that the BIS numbers are closer to the truth.

B) The runs per win calc was re-worked.

From the WAR explained link that Sean mentions. (Version 1.0 being the version that was carried by baseballprojection.com)

Version 2.0, May 2012

Previous to launch of 2012 WAR numbers we undertook a top to bottom evaluation of our WAR numbers and added a number of improvements.

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.
dWAR now contains the position component as we feel this better captures player defensive value. In our view, even a poor defensive catcher is likely equally valuable to a good defensive first baseman in terms of team defense.


Version 2.01, May 2012

We made a minor change converting oWAR to afWAR (or average fielding WAR). This has been rolled back and is not in place now.

Version 2.1, May 2012

After launching version 2.0 on May 4th, we immediately became aware of an issue that had concerned us, but we thought we still had right. Pitchers were being overvalued due to a runs to win estimate that broke down for extreme performances. A stingy pitcher drives down the runs per win, but not to the degree we were showing.

A major change to runs to win calculation. See our Runs to Wins Page for a full explanation. We now handle runs to wins calculations in an exact rather than an estimated way.
With the change in runs to wins calculation we can now display Wins Above Average, a related win-loss % and a related win-loss% for 162 games played.
Leverage Index adjustment is now only applied to relief pitchers.
Leverage Index used is now the LI at the time the reliever entered the game rather than the average LI for all of their plate appearances. This is weighted by number of batters faced.
Converted Offensive WAR from afWAR back to oWAR. Note that oWAR + dWAR now double counts position, so adding them will not give WAR.
The adjustment for the difference in league starter and reliever ERA has been moved to the calculation of league average rather than the league replacement level.
BUG: Fixed a park factor bug. Pitcher opponent strength was not neutralized, so a park factor was being applied to a non-park adjusted number. Now the pitcher opponent strength is converted to a neutral environment, averaged for all opponents and then park adjusted based on a custom park factor for each pitcher based on the exact parks they pitched in.
Some feedback of version 2.0 we incorporated: Inside the Book and its readers helped immensely with the runs to wins issue.