WAR Upgraded on Baseball-Reference.com

[fenrir]

As many have noticed, we have not been updating the 2012 WAR data in-season. The reason for this is that we have been working on a major upgrade to our WAR framework. The link above gives a very detailed rundown including: how to compute each part of our WAR formula, our reasoning on making the changes, downloads for historical WAR data, charts of the various constants and factors we use, and a chart of differences between the new B-R WAR, the old B-R WAR and FanGraphs WAR.

...All in all, these are major improvements to the system and while there are dozens of details, here are the main points of difference between our old framework and the new one.

• 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.
• I’ve changed the oWAR and dWAR formulations. oWAR is now called ndWAR for (no-defense WAR), but is the same otherwise. dWAR now contains the position component of WAR, so the Career Leaderboard is now dominated by SS, C and other great defensive players.

http://www.sports-reference.com/blog/2012/05/war-upgraded-on-baseball-reference-com/

[brett]

There are two other factors I'd like to see. One is adjuting a players plate appearances to what they would have been on an average offensive team. A good hitter produces more value WHEN he is on a team that gets him to bat more times. This is not about ballpark neutralization which is already occurring, it is about neutralizing the effect of being on a team that turns the lineup over.

The other would be to add in extra bases taken on the hits, sacs and balls in play of other hitters versus average.

What does wRAA stand for?

I like putting the positional adjustment into defense because then all players offense is set at the same baseline.

I don't like eliminating pitchers batting from league rates. Having pitchers hit in a league does not affect the conversion of player events to runs and wins, but it SHOULD effect league quality and lower the LQ for a league where pitchers hit, but oh well.

And some people will despise the fact that these changes basically make war a system of putting a player into a neutralized setting, rather than actually accounting for his contribution to his team-a literal mathematical accounting.

[brett]

Crud. Brett lost 2 of his 3 league WAR leaderships for position players, losing 1976 to NETTLES! NO!!!!!! and also falling to #2 in 1979.

On the other hand he rises to 29th from I think 34 or 38 before probably because he had a DH in the league.

He also gains ground on Boggs, Kaline and Anson. Since he's one of the top 10 runners of the last 50 years on taking extra bases and also has a high level of post season value I think its a pretty good score.

Some people are going to love seeing Clemente up there.

[cbenson5]

Well, I guess that I will ask the obvious question. Is this new version more accurate than the previous one? I am just scanning through the list of position players and only see a few major changes. Both Mantle and Dimaggio fell significantly with the new calculations. Brooks Robinson got a nice bump, moving him into to just outside the top 40. It seems like everyone else, among the top 40 or so, remained in similar positions to the previous list.

[brett]

Well, I guess that I will ask the obvious question. Is this new version more accurate than the previous one? I am just scanning through the list of position players and only see a few major changes. Both Mantle and Dimaggio fell significantly with the new calculations. Brooks Robinson got a nice bump, moving him into to just outside the top 40. It seems like everyone else, among the top 40 or so, remained in similar positions to the previous list.

It looks to me like players in a DH league moved up slightly RELATIVE to the other players. That is probably good for rating players all time if you don't have an LQ adjustment because obviously a certain NL hitter will not be worth as much relatively speaking in a DH league, all else being equal. I am not sure what difference there is in the new defensive metric. Ripken still is +180 something. I think he's a great defender but basically the old and new metrics are the only 2 that have him anywhere near that level. Same for Boggs. In fact the defensive runs look just like before, and there are some case instances where those numbers are highly questionable. For example Ozzie Smith being +32 runs in 1989 when I've seen other systems which account for all kinds of tendancies say that it was only his 7th or 8th best defensive season.

Projected CS looks like it hurt some deadball base stealers. Cobb drops below Mays. Most players seemed to take a little drop. Cobb lost about 15 WAR and Anson about 10. George Davis drops about 10 as well. If you only lost 3-4 you probably gained ground.

[Bothrops Atrox]

It looks to me like players in a DH league moved up slightly RELATIVE to the other players. That is probably good for rating players all time if you don't have an LQ adjustment because obviously a certain NL hitter will not be worth as much relatively speaking in a DH league, all else being equal. I am not sure what difference there is in the new defensive metric. Ripken still is +180 something. I think he's a great defender but basically the old and new metrics are the only 2 that have him anywhere near that level. Same for Boggs. In fact the defensive runs look just like before, and there are some case instances where those numbers are highly questionable. For example Ozzie Smith being +32 runs in 1989 when I've seen other systems which account for all kinds of tendancies say that it was only his 7th or 8th best defensive season.

Projected CS looks like it hurt some deadball base stealers. Cobb drops below Mays. Most players seemed to take a little drop. Cobb lost about 15 WAR and Anson about 10. George Davis drops about 10 as well. If you only lost 3-4 you probably gained ground.

The biggest difference on defense is that 2003-2012 data is now DSR (+/-) instead of TZR, which is a good thing.

I'd love to do a comparison between the new brWAR and the new bgWAR.

[Nerdlinger]

Catchers seem to be getting screwed more, at least offensively.

[JDanger]

The other would be to add in extra bases taken on the hits, sacs and balls in play of other hitters versus average.

This is included in the Baserunning Runs component: "Non-Basestealing baserunning which includes items like 1st to 3rd on singles, outs on the bases, tagging up on fly balls, scoring from third on a ground ball, etc. "

What does wRAA stand for?

Weighted runs above average.

[brett]

This is included in the Baserunning Runs component: "Non-Basestealing baserunning which includes items like 1st to 3rd on singles, outs on the bases, tagging up on fly balls, scoring from third on a ground ball, etc. "

Where is the source for that quote? BBRef lists: "SB, CS, PB, WP and defensive indifference" under the Rbaser tab.

[JDanger]

Where is the source for that quote? BBRef lists: "SB, CS, PB, WP and defensive indifference" under the Rbaser tab.

http://www.baseball-reference.com/about/war_explained_position.shtml

Maybe Sean hasn't gotten around to updating the roll over definitions yet.

[brett]

http://www.baseball-reference.com/ab...position.shtml

Maybe Sean hasn't gotten around to updating the roll over definitions yet.

That's fantastic. I would have to say that the offensive component of WAR is now undeniably accurate to within a small percent. That leaves our trust in the defensive metric and our LQ adjustment to be subjective about.

[Ubiquitous]

That's fantastic. I would have to say that the offensive component of WAR is now undeniably accurate to within a small percent. That leaves our trust in the defensive metric and our LQ adjustment to be subjective about.

Well, not really. He switched from baseruns to linear weights. LW is easier to calculate and program but isn't any more accurate or really as accurate as other metrics.

The thing is there really is no such thing as "accurate" when it comes to metrics. Everything is an opinion. This single here, that double there, the out from there, so on and so on equals 2 runs or whatever. That is really just an opinion and it someone might say it is worth 2 runs today but then say it is worth 2.3 runs tomorrow, followed by 1.7 runs the next day, and quite possibly back to 2 runs in a month.

There is no final complete factual product and there never will be.

[fenrir]

In the new WAR system, 29 of the last 50 NL MVP's are pitchers, and 27 of the last 50 in the AL. The new version of WAR seems incredibly pitcher friendly.

[Bothrops Atrox]

In the new WAR system, 29 of the last 50 NL MVP's are pitchers, and 27 of the last 50 in the AL. The new version of WAR seems incredibly pitcher friendly.

Sean has been made aware of this and is contemplating fixing it. There is a distortion in his runs-to-wins estimator that helps pitchers at extreme levels.

[brett]

Sean has been made aware of this and is contemplating fixing it. There is a distortion in his runs-to-wins estimator that helps pitchers at extreme levels.

I think there should be. A pitchers runs saved are concentrated into a few games. Plus runs "saved" are worth more than runs allowed. If a batter produces 50 extra runs spread out over 162 games, his team will win about 5.6 extra games. If a player "saved" 50 runs spread out over 162 games his team would win about 6.0 extra games. If a pitcher saves 50 runs concentrated in 35 games, his team would win 6.8 extra games. That is all at a 700 runs per season average setting. It also by the way helps the pitcher with the better ERA+ in fewer innings. The same runs saved in 340 innings will not produce as many wins as in 225 innings.