Twins and other AL Central WAR Projections

I've posted projections for the Twins and the rest of the AL Central (using Sky's spreadsheet) over at Twinkie Town. I plan to collaborate with the Twins community over there to refine Minnesota's projections. The other divisional teams are included for comparison. If anyone has a suggestion or points out a major error with the projections for any team, please let me know so I can fix. I bet a bunch of you follow the other AL Central teams much more closely than this Twins fan...

10 months ago Adam Peterson 18 comments 0 recs |

Story-email Email Printer Print

Comments

I would love to see this same thing for the NL Central

You overall tab shows what I have been wondering, do you think the AL central can average 85.36 wins. A lot of people a lot smarter than me have developed the numbers behind WAR, but (just going from my gut) maybe the 2.5 wins for replacement level is too high for the AL.

by JBrew on Jan 8, 2009 5:21 PM EST reply actions 0 recs

No

As I say in the TT post, taken as a whole these projections are not neutral when summed over all players. There’s an optimistic bias, likely on the order of 3-4 games (just a guess) because the projections do not account for the multiple 100s of PA that will have to come from folks not on the sheets. For example, when Punto goes down as the Twins SS, Harris can move over and Buscher takes over 3B. Tolbert could still backfill, but someone else like Machado, Plouffe, etc would come up and get AB. These projections don’t account for inevitable major injuries.

I think one way to handle this is to finish projections for the entire league and then calibrate so that overall win % is 50%.

by Adam Peterson on Jan 8, 2009 6:43 PM EST up reply actions 0 recs

This is my plan.

I also do not think I’m currently including the optimal number for replacement-level wins for teams. I thin it’s a bit high.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 8, 2009 6:58 PM EST up reply actions 0 recs

Could be

That’s effectively the same thing as my calibration, remove a WAR or three.

by Adam Peterson on Jan 8, 2009 9:40 PM EST up reply actions 0 recs

Replacement Level Offense

Looks like you’re using 2.5 for AL and 2.0 for NL. Since these WAR are added to the player’s total offensive WAR, isn’t this saying the replacement AL hitter is worse than the replacement NL hitter? I thought it was the other way around…

by Adam Peterson on Jan 9, 2009 9:23 AM EST up reply actions 0 recs

Nope.

Given the same stat line in the NL and AL, the AL performance is more impressive and more valuable. So you need to add more wins to the raw WAA value of it in order to make it look better.

On a spreadsheet like this, it’s probably ideal to use perfectly league-adjusted numbers and use 2.25 WAR for replacement for everyone. But it’s much easier in application not to league-adjust projections, since most players have spent the past few years all in the same league.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 9, 2009 10:21 AM EST up reply actions 0 recs

Replacement relative to average

Tangotiger uses 2.5 for AL and 2.0 for NL because the average AL hitter is better than the average NL hitter.
How to calculate WAR
It would seem that in order to use the different replacement win values, you would need to apply different average wOBA to calculate hitting WAR in column F of the spreadsheet.
In 2008, average OBP (which equals average wOBA) was .335/.331 for AL/NL. In previous years, the averages were .338/.334, .339/.334, .330/.330, .337/.333.
Other than 2005, there was a pretty constant .004 difference in average wOBA. Over 700 PA, this equates to a 2.8 run difference, or about a quarter point between the leagues.

I recommend you use 2.25 as the replacement WAR for AL teams. This will take about 2 wins off the AL totals, which feels about right given what I’ve seen so far.

by Adam Peterson on Jan 9, 2009 10:08 AM EST up reply actions 0 recs

Initial feedback shows win projections averaging well over 81 wins. I'll take a look.

It IS nice to see reasonable people still thinking the Indians and Tigers have good teams, even if they didn’t win a ton of games last year. I bet our community projections put them much higher than Vegas.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 8, 2009 6:21 PM EST reply actions 0 recs

pitching for the Tigers

yucky. Indians… got a lot of injuries and personnel shortcomings to overcome. I think the Central is the Twins’ to lose as of now.

by Daniel Berlyn on Jan 8, 2009 9:46 PM EST up reply actions 0 recs

These projections can't be right (or the Dodgers isn't)

All of these offenses are at least six wins better than the Dodgers? The Royals offense?? I just don’t see how that is possible.

Also, every single rotation in the AL Central is projected as better than the Dodgers, and Xeifrank’s Dodger projection assumes the Dodgers sign a league average starting pitcher. I know I’ve brought this up before but the projections posted here do not match up.

And the projections just don’t make intuitive sense either. I’m sorry, I like the Indians, but there is no way they can be predicted to win 91.5 games next year. And their hitting WAR is 28.6, basically double the Dodgers 14.74!! Obviously projections will vary but that is beyond ridiculous, their hitters alone cannot be as valuable as the whole Dodger team!

by Brendan Scolari on Jan 9, 2009 3:20 AM EST reply actions 0 recs

Dodgers-Indians gaps

Simply changing the A/N cell from AL to NL reduces a team’s WAR by about 4.5. The Indians go from 28.6 to 24.0 WAR offensively due to the different replacement player value. This explains some of the difference compared to the Dodgers. Removing the DH plate appearances reduces the Indians WAR down to 21.7 over 3753 outs, just a bit above the target value. That leaves about a 7 WAR gap between the two offenses as currently constructed using the separate projections.

I just ran the numbers (quick and dirty) for LA using the same spreadsheet and projection method. Offensively, I have La with a WAR of +19.2, and that’s not including Manny Ramirez or Andruw Jones. Sounds like there’s a pretty sizable gap between the projections, and I suspect it’s not due to large differences in wOBA and defense…

by Adam Peterson on Jan 9, 2009 9:14 AM EST up reply actions 0 recs

Adding in a DH will increase the offensive WAR total by a bunch.

But that doesn’t create more wins for AL teams, because they compete against themselves. That’s part of why the minimum win totals are different for AL and NL teams.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 9, 2009 10:24 AM EST up reply actions 0 recs

Hm.

Well I don’t know but the Dodgers projections posted here earlier had them at 79.83 wins and I just don’t see how the Indians would be that much better. But Adam Peterson’s numbers make it look like maybe the Dodgers projections are too low (or Adam’s projections are too high, which may very well be the case).

It’d be nice if we could get the formulas to match up though, because right now the projections are pretty meaningless outside of the division projected.

by Brendan Scolari on Jan 9, 2009 9:22 PM EST up reply actions 0 recs

It would be nice if someone took those Dodgers' projections and plugged them into my spreadsheet....

… just to see how different the calculations are.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jan 10, 2009 5:56 PM EST up reply actions 0 recs

My Dodger projections are now in my spreadsheet

I’ve now added Baltimore and the Dodgers to the sheet. Will slowly build this up as we go. For LA, I see them with a solid offense (18.9 WAR), with Ethier, Kemp and Martin all at .366 wOBA or higher. I assumed DeWitt at 2B and Hu backing up 2B and SS. I put Pierre in LF and did not include Andruw Jones. The pitchers I was able to project for LA were very good, but I came up 125 IP short for starters and 150 IP short for relievers, projecting all as replacement.

Anyway, Dodgers experts, what did I get right and wrong?

by Adam Peterson on Jan 11, 2009 5:18 PM EST up reply actions 0 recs