rWAR seems to overrate the O’s position players

[Frobby]

The theory of WAR is that a team of replacement-level players can win 48 games.  A .500 team should generate about 33 WAR.   A 35-40 team like the Orioles should have produced about 13 WAR to this point in the season.  Call it 14, since their Pythagorean record is one game better than actual, at 36-39.

However, BB-ref has the O’s position players at 11.8 rWAR, and the pitchers at 7.8 rWAR for a total of 19.6 rWAR, about 5-6 WAR higher than what the total should be based on the team’s record and run differential.   By contrast, Fangraphs has our position players at 5.8 fWAR and pitchers at 7.5 fWAR, for a total of 13.3 fWAR, right in line with the team’s record and run differential.   

So where do BB-ref and Fangraphs differ with respect to individual players?   Here are the biggest gaps:

Hays 2.7/1.8 = 0.9 gap

Mullins 2.3/1.6 = 0.7 gap

Mancini 1.6/1.1 = 0.5 gap

Mateo 1.5/0.9 = 0.6 gap

Urias 1.3/0.5 = 0.8 gap

Santander 0.9/0.6 = 0.3 gap

Chirinos 0.3/-0.8 = 1.1 gap

Normally, I prefer rWAR to fWAR, but here, it’s hard to get around the fact that rWAR doesn’t line up with our actual record or our run differential.

 

[Can_of_corn]

Probably an issue with offense being down in the league.  I would be shocked if it was only the O's position players.

[BohKnowsBmore]

Pythag has them at 36-39 (1 game better than current record), so alternative explanation could be the reverse, that record is underperforming underlying level of production. 
 

I don’t necessarily think that’s the case, more likely that we’re seeing random variation in the individual value approximation model(s) relative to on-field cumulative results. 

[Frobby]

8 minutes ago, BohKnowsBmore said:

Pythag has them at 36-39 (1 game better than current record), so alternative explanation could be the reverse, that record is underperforming underlying level of production. 
 

I don’t necessarily think that’s the case, more likely that we’re seeing random variation in the individual value approximation model(s) relative to on-field cumulative results. 

I took the Pythag into account when I said a total of 14 WAR would be about right for the total of the team.  

I think what this mostly boils down to is that Rdrs (which is the fielding stat that feeds into rWAR) likes our defense better than most other metrics do.  
 

[DrungoHazewood]

2 hours ago, Can_of_corn said:

Probably an issue with offense being down in the league.  I would be shocked if it was only the O's position players.

My first thought was that they use 3-year park factors, which would assume the O's are playing in a much better hitter's environment than they are with the modified OPACY. But that's backwards, that would overrate pitchers and underrate non-pitchers.

I think Frobby is right, most of the difference is that DRS sees the Orioles as the 4th-best defensive team in baseball, while OAA has them as about average.  Interestingly, DRS centers at about +9 on a team level, the Majors are +273 right now.  So an average team is 9 runs above average.

[Tony-OH]

Hmm, interesting. I've always liked rWAR over fWAR but this is interesting. Did you go back to last year and see if rWAR or fWAR was closer to the record to see if this is an anomoly?

[Frobby]

2 minutes ago, Tony-OH said:

Hmm, interesting. I've always liked rWAR over fWAR but this is interesting. Did you go back to last year and see if rWAR or fWAR was closer to the record to see if this is an anomoly?

Last year our position players were +7.6 and the pitchers were +1.1, for a total of +8.7.   That suggests a 56-57 win team.  Our Pythag last year projected us for 54 wins, so rWAR was not too far off that.  There’s always going to a little random fluctuation to account for things like RISP hitting and other sequencing issues.   But to be 5-6 wins off less than halfway through the season is a very dramatic disparity.   

[Tony-OH]

29 minutes ago, Frobby said:

Last year our position players were +7.6 and the pitchers were +1.1, for a total of +8.7.   That suggests a 56-57 win team.  Our Pythag last year projected us for 54 wins, so rWAR was not too far off that.  There’s always going to a little random fluctuation to account for things like RISP hitting and other sequencing issues.   But to be 5-6 wins off less than halfway through the season is a very dramatic disparity.   

Interesting. Wonder what is causing the jump?

[eddie83]

I brought this up many years ago. 
 

83 Orioles position player rWAR 26.0

83 Orioles pitchers rWAR 18.4

Total 44.4

84 position player rWAR 25.4

84 pitchers WAR 10.5

Total 35.9

83 Orioles went 98-64, Pth was 96-66

84 Orioles went 85-77, Pth was 83-79

 

So a little bit off but that was not the issue. The issue is how the 84 offense tanked and the 84 pitching staff was consistent. 
 

Orioles scored 799 runs for an OPS+ of 111 in 83, ERA was 3.63 2nd, ERA+109

 

In 84 they scored 699, 9th OPS+102, ERA was 3.71, 2nd  105 ERA+
 

dWAR went from 0.2 to 4.9. 
 

Basically the Orioles offense went way back, the pitching remained constant and yet the position player WAR was steady and not the pitching WAR.