This makes sense - but I do hope they don't both use the same methodology. It's actually illuminating to have different values of WAR for the same player - like a sensitivity test to give us a sense of the uncertainty in the estimate.

I heard one of the Fangraphs guys on some podcast recently and he wasn't sure what RL Fangraphs was using, and thought that they were upping it to .320 WP. I'd be surprised if it doesn't land there.

edit: actually it might have been a different Fangraphs guy.

Terrible news! I just spent months adjusting my database to the new B-Ref WAR. Why does no one stop and consider how these things effect me before going ahead?

Repoznanski

I really wish they had used an acronym that wouldn't lead to the same pun being used over and over in article titles.

Terrible news! I just spent months adjusting my database to the new B-Ref WAR.

Explain this process to me. WAR is a calculated field. Unless it takes you months to write a calculated column or stored procedure or something, any change should be pretty quick.

I supposed using the same replacement level would be OK, but I certainly hope they don't agree on pitching value methodology or use the same fielding metrics - that would actually result in having less useful information.

Explain this process to me. WAR is a calculated field. Unless it takes you months to write a calculated column or stored procedure or something, any change should be pretty quick.

It's mostly that I factor in WAR in a player's best 3 season and best 5 seasons. So it takes a couple seconds to determine which seasons those are. Also, my tongue is a little bit in cheek in that I actually enjoy inputing all that manually...I'm done my update now and starting to feel a litte bored. Actually look forward to an excuse to tinker.

Somebody should hold, I dunno, like a roundtable or something.

Greg: I'm pretty confused by what you're doing. Why do you even have to input anything manually?

#10 I'm guessing it's because he hasn't read the BB-REF page that explains in gory details how to calculate WAR.

That said, Sean provided a complete WAR database (in CSV format) so calculating WAR strikes me as ... needless.

Greg: I'm pretty confused by what you're doing. Why do you even have to input anything manually?

It's probably the "have to" that's the confusing part.

I input the data manually because
A) I'm databasely illiterate
and
B) I enjoy spending a minute or two looking up each guy's b-ref page and getting to know him as a player. It's more about the process than the finished product for me.

My complaint was more sarcastic than anything else. In a year or two I'll probably relish the excuse to go back and revisit thousands of players.

EDIT: And just because you kind of asked. What I do is career WAR, WAR in the best 5 seasons, WAR in the best 3 seasons, WAR per PA, and give different arbitrarily determined weights to each of the 4 categories and then it spits out a number.

For the 2012 season I changed my rule of only including active players once they reach 4000 PA to, 1000 PA or a 2.0 WAR season, so for now I've got zany things like Yasmani Grandal being the 30th best catcher since 1990.

Nerd Summit!

I see. I get that you enjoy doing it the way you do, but it's beyond inefficient. With a proper database, this is just a query or two. The whole process should take minutes.

Repoznanski

Rzeposczanski

So, the new replacement level is a .294 winning percentage. No explanation for why, but its seems picked because that makes 1000 WAR in the league rather than based on any calculation.

[16] So a team's WAR should be around wins-47 then. And yet, 2012 Yankees: 47.9 WAR, 2012 Orioles: 33.3 WAR. The Orioles won only two fewer games. So there's an issue of a stat called wins above replacement that's not actually based on wins. (more like Pythagorean wins above replacement, heh)

So there's an issue of a stat called wins above replacement that's not actually based on wins. (more like Pythagorean wins above replacement, heh)

I prefer it that way. Seems like many others do, but you can't please everybody. If you prefer it to add to actual wins, use win shares.

They should (and I hope THEY are listening) have 3 or 4 WAR calculations (in addition to agreeing on a RL)

FIP-WAR (100% defense)
RA-WAR (0% defense)
ERA-WAR(~2% defense)

and whatever current bbREF partitioned defense WAR is (something between ERA-WAR and FIP-WAR)

Fangraphs already has the first two of those. FIP-WAR is their normal pitching WAR, and RA9-Wins is your "RA-WAR". You can find the latter in the table under the Value header at the bottom of a players stat page.

Seems like many others do, but you can't please everybody

Hard to see why. If the sum of a team's players' WAR doesn't equal the team's WAR, that seems like a pretty big matzo ball hanging out there.

Yes, but both sites should have all 4.

And yet, 2012 Yankees: 47.9 WAR, 2012 Orioles: 33.3 WAR. The Orioles won only two fewer games. So there's an issue of a stat called wins above replacement that's not actually based on wins. (more like Pythagorean wins above replacement, heh)

You might be interested in Player won-lost records. They're calculated two ways: tied to actual team wins/losses (pWins/pLosses) and based on expected wins/losses (eWins/eLosses). So, for example, the 2012 Yankees had 2.1 more pWORL (pWins over replacement level) than the 2012 Orioles, but 7.3 more eWORL.