Sunday, February 1, 2009

Nationals Wins Above Replacement Updated

I've made some updates to my Nationals Wins Above Replacement calculations. First, I've updated the playing time and innings pitched to reflect the latest developments, including the acquisitions of bullpen arms like Wil Ledezma, Terrell Young, and Jesus Colome. Second, from here on out I'm going to use CHONE projections for hitting and PECOTA for pitching. In my experience PECOTA's the best for projecting pitching, while CHONE has been a bit better for hitters. I've also re-scaled the wins projection to reflect the lower run-scoring environment we saw in 2008. I'm now assuming 10.1 runs per win, down from 10.5.

You can as always view the full spreadsheet here to see all the dirty math. My contribution is to simply estimate playing time and then run the numbers. For more detail on the method, check out my original WAR post here.
Also, I have my calculation for fielding runs above replacement posted here and explained at the very bottom of this post.

But if you just want the bottom line results, here you go:

Hitters WAR
Pos Name PT wOBA FRAA Pos adj WAR
C J. Flores 70% .296 0 1.25 0.80
C W. Nieves 30% .288 0 1.25 0.20
1B N. Johnson 50% .399 7.38 -1.25 2.79
D. Young
1B R. Belliard 30% .331 0 -1.25 0.02
2B R. Belliard 40% .331 0.06 0.25 0.90
2B A. Hernandez 60% .288 0 0.25 -0.20
SS C. Guzman 85% .326 -3.60 0.75 1.78
SS A. Hernandez 15% .288 0 0.75 0.02
3B R. Zimmerman 85% .371 15.11 0.25 5.23
3B R. Belliard 15% .331 -8.94
0.25 0.20
RF E. Dukes 65% .369 6.85 -0.75 2.74
RF A. Kearns 35% .341 15.16 -0.75 1.17
CF L. Milledge 75% .348 -8.47 0.25 1.83
CF W. Harris 25% .330 11.85 0.25 0.84
LF J. Willingham 80% .362 -15.03 -0.75 1.30
LF W. Harris 20% .330 23.49 -0.75 0.70



Pitchers WAR

SP John Lannan 180 4.38 2.22
SP Scott Olsen 190 4.42 2.26
SP Daniel Cabrera 190 4.31 2.49
SP Collin Balester 150 5.49 0.02
SP Jordan Zimmermann
SP Shawn Hill 60 4.65 0.56
SP Jason Bergmann 50 4.75 0.41
CL Joel Hanrahan 70 4.28 0.17
SU Saul Rivera 70 4.09 0.32
RP Jesus Colome
50 4.70 -0.11
RP Jason Bergmann 30 4.75 -0.08
RP Steven Shell 60 4.35 0.10
RP Mike Hinckley 40 5.71 -0.53
RP Marco Estrada 40 5.89 -0.61
RP Shairon Martis 63 4.4 0.09
RP Wil Ledezma 50 4.71 -0.12
Terrell Young
Garrett Mock
Gary Glover

Total 1450

Total Team WAR
Total Offense WAR 19.74
Total Defense WAR 0.76
Total SP WAR 7.74
Total RP WAR -1.09
Total Wins 74.15


James Bjork said...

I'm still trying to get my brain around what "replacement level" is in terms of general player performance. Your other link still doesn't get at the main gist. Is it league-AVERAGE performance at the postition? I can't see how it could be, since the Nats score in the overall positive range, yet only have 74 wins.

e poc said...

Thanks, Steven. This is really cool.

Do you know why the runs above average are different on fangraphs than they are on the chone website? Also, how did you figure out offensive runs above average? a conversion from wOBA? You have the hitters a lot higher than the chone website does. Comparisons with fangraphs are a mixed bag: you've got Belliard, Harris, Kearns, and AHern lower; Milledge about even; Nick Johnson higher. Do you know why that's happening? (I'm particularly interested to know why the fangraphs numbers disagree with chone's website's numbers, if you happen to know.)

e poc said...

Okay: I'm wrong. You have Nick Johnson a little low compared to fangraphs as well. So I guess they're just using a slightly different run environment/league average than you? No big deal. So the big question I have is just why chone's website is different than fangraphs - any clue?

JN said...

Replacement level is basically what could be produced by a AAAA player, someone who's available on a Triple A team or on waivers.

A league average player is something like 20 runs above replacement level, or at least has been recently.

Steven said...

Right--what JN said. If I remember right, it's calculated by BP as the level of production of players who get less than 20% of the PT at a position, or something like that. So it's sometimes described as the value of "freely available talent on the street." That's not right. Many teams don't have "replacement level" talent available in their systems (see: Nationals 1B in 2008). But the actual average level of performance of the guys who get called upon as backups and call-ups.

Steven said...

Re: epoc--I dunno. I should double-check it. it's possible CHONE updated his projections. I pulled those numbers weeks ago. I don't remember if I got it from his site or Fangraphs either.

James Bjork said...

OK, I get it. So replacement level is almost an economic metric in that you have a benchmark for whether it is worth paying a guy a penny more than the $400k minimum. So how many games in a season does a team of entirely replacement-level players project to win? Might it be around 59?

Steven said...

Very timely--Fangraphs just did a post on this exact topic.