@Fangraphs and @Baseball_ref unify their replacement level values

In an announcement on fangraphs.com, it’s been revealed that they and Baseball-Reference.com have unified the replacement level used for WAR calculations, meaning that bWAR and fWAR will now be the same.

I just chatted with our old friend @Neil_Paine about this and got some info:

  • Previously, B-R used a win level of 51-52 per 162 games as the replacement basis.
  • The new level is 47-48 wins per 162 games, which will be used by both sites.
  • Thus fWAR levels for all players are going down, and bWAR levels for all players are going up.
  • Previously, the longer a player’s career was, the larger the discrepancy existed between the two values.
  • This means that players with long careers (like Jack Morris….sigh) will get larger bumps up in rWAR or bumps down in fWAR.
0 0 votes
Article Rating
Subscribe
Notify of
guest

27 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Adam Darowski
Admin
10 years ago

This is good news, since at least that minor part of WAR should be standardized. I don’t care if they calculate WAR differently, but I’d love to see things like equal replacement values, equal park factors (are they? I don’t even know) and equal run values for different events (that may be standard, too… I don’t use fWAR much). I prefer the differences to be fundamental rather than minor things like that. Just makes it easier to explain to the folks who are trying to understand it.

Ed
Ed
10 years ago
Reply to  Andy

Andy – do you have any sense as to why they’re not friendly? Of course, maybe you don’t want to say even if you do know. I realize in some ways they’re competitors but it seems like they have a common interest in advancing statistical analysis of baseball. And to me, that should override other issues.

Ed
Ed
10 years ago
Reply to  Andy

Ok thanks Andy. I appreciate the explanation.

Adam Darowski
Admin
10 years ago
Reply to  Andy

I certainly haven’t gotten a sense that they’re UNFRIENDLY, just that they have different approaches and like to point out those differences. They both believe in the methodologies behind their calculations. I honestly think they’re telling different stories (particularly in the case of pitchers), so it’s okay.

birtelcom
Editor
10 years ago

Aside from the fact that b-ref and fg are competing businesses, they do seem to me to have slightly different emphases in approach. Fangraphs has a somewhat greater emphasis on player evaluation and prediction, which to some extent may derive from a slightly greater gearing toward the fantasy/gaming market. B-ref is more historical and archival in emphasis it seems to me, following more in the Baseball Encyclopedia/Who’s Who in Baseball tradition as compared to the Street & Smith Guide annuals tradition that seems to inspire fangraphs. I’m not entirely enthusiastic that they would collaborate on a compromise position on some… Read more »

birtelcom
Editor
10 years ago
Reply to  Andy

Old blog? Never heard of it.

Ed
Ed
10 years ago
Reply to  Andy

But with double the fun! And twice the flavor!!!

Insert Name Here
Insert Name Here
10 years ago

So when will this be rolled out by the sites? I don’t use Fangraphs, but I see B-R has yet to roll out the new WAR.

bstar
10 years ago

This is great news. Gives the anti-WAR mongers one less thing to complain about.

Apparently fWAR’s new numbers are now live. Tom Glavine now has 64 fWAR. Didn’t it used to be in the fifties? I think it did.

bstar
10 years ago
Reply to  Andy

You’re right-it was 69 fWAR for Glavine w/ the old replacement level. I was thinking 59.

Back when we were discussing Schilling/Mussina/Glavine, Bryan O’Connor listed the fWAR of these pitchers and also for Kevin Brown and John Smoltz later on down in the thread:

http://www.highheatstats.com/2013/01/circle-of-greats-1965-ballot/#comment-46797

old fWAR / new fWAR

Schilling 86.1 / 83.5
M Mussina 85.6 / 82.3
J Smoltz 82.5 / 78.4
Kevin Brown 77.2 / 73.8
T Glavine 68.5 / 63.9

The stepdown looks fairly uniform for all five pitchers.

Ed
Ed
10 years ago
Reply to  bstar

As a percent they range from 3.0% for Schilling to 6.7% for Glavine. So some definite differences which I believe are based on playing time.

Hank G.
Hank G.
10 years ago

“In an announcement on fangraphs.com, it’s been revealed that they and Baseball-Reference.com have unified the replacement level used for WAR calculations, meaning that bWAR and fWAR will now be the same.”

No, they won’t be the same, since they still calculate WAR differently. They’ve just agreed to use a common replacement level, since critics often denigrated WAR by pointing out that different sites couldn’t even agree on what replacement level was. I’m sure though, that they will find other reasons not to accept WAR as a valid measurement.

RJ
RJ
10 years ago

Woah, woah, woah, have they added another decimal place?! Looks like they have, for position players at least. Now I can finally argue that player A was 1/100 of a WAR better than player B. 🙂

RJ
RJ
10 years ago
Reply to  RJ

For example, Carlos Beltran (65.05) is clearly superior to Adrian Beltre (65.03). It’s the extra letter in his name what does it.

mosc
mosc
10 years ago
Reply to  RJ

I hate even having a decimal place to begin with. The number has too much variation and inaccuracy to give more than 2 significant figures.

RJ
RJ
10 years ago
Reply to  mosc

I can see both arguments. The calculation of WAR clearly gives you a number with multiple decimal places, so why not show an extra digit in the interest of preventing rounding errors? If you’re going to calculate something with that much precision, you might as well be as transparent as possible.

On the other hand, it implies a degree of accuracy that probably isn’t justified. Can we really say with a straight face that Beltran is worth two hundredths of a win more than Beltre over his career?

John Autin
Editor
10 years ago
Reply to  mosc

RJ & mosc, I agree that WAR should be shown with 1 decimal place. And it now seems that those 2-decimal figures we saw on B-R were just temporary, while the site was being updated. They’re back to 1 decimal place everywhere that I’ve looked.

RJ
RJ
10 years ago
Reply to  John Autin

It’s funny to see Sean posted the article I have linked below shortly before my comment @29, because GDP was exactly what I was thinking of when reading mosc’s comments about how WAR should be presented with less precision. Part of the problem with GDP estimates is that people take them as gospel, partly because they publish them to within a decimal place. In truth there’s no way in heck that they can be completely sure that that is the true value of a country’s GDP, but the decimal places implies certainty. I guess if one treats GDP estimates with… Read more »

Fireworks
Fireworks
10 years ago

Hank G. everyone knows they’re still calculated differently. This post is about the replacement level irrespective of the ambiguity of the sentence you quote. Personally I like this move. Replacement level should be standardized and is not an appropriate point of contention. As for the statements about FG and BR as it relates to their approach, it’s accurate to say that FG is about using the underlying elements of repeatable and predictive success as components in their WAR while BR focuses on the actual events that occurred. I greatly prefer BR overall and think rWAR’s advantage over fWAR becomes greater… Read more »

mosc
mosc
10 years ago

I always thought this number needed fixing… but not by making it the same. I view the talent pool of baseball as a variable thing. Some years, some eras, it’s deeper than others. Expansion teams make waves in the talent pool generally reducing the quality of the league “replacement” level quite a bit for a few seasons until their farm system develops and there is a proportional restoration of talent in the league. Perhaps an even bigger change was free agency. I think these shifts can be see in the extremes of OPS+ and ERA+. Not having the data it’s… Read more »