2006 Prediction using Win Shares

[RME JICO]

Using reasonable numbers going off past seasons I came up with the following:

 

2006 Win Share Projections

SP

1. Buehrle - 21

2. Contreras - 17

3. Garcia - 17

4. Garland - 16

5. Vazquez - 15

RP

6. McCarthy - 8

7. Cotts - 10

8. Politte - 9

9. Jenks - 14

10. Hermanson - 7

11. Pitcher - 4

C

12. Pierzynski - 15

13. Widger - 2

IF

14. Crede - 14

15. Iguchi - 15

16. Konerko - 22

17. Thome - 18

18. Uribe - 17

19. Ozuna - 4

20. Mackowiak - 8

21. Player or 12th Pitcher - 3

OF

22. Dye - 16

23. Podsednik - 14

24. Anderson - 10

25. 4th Outfielder - 3

 

299 Win Shares

99.67 Wins

 

I got this on my first try just using projections compared to the last 3 years. I even have some conservative numbers and I actually think a lot will be higher.

 

Do any of these look ridiculous or not realistic? I only see one or two players that would do worse, but there are about 8 that should do better.

 

The only change would be if Count is moved and B-Mac moves into the rotation, where I would give B-Mac 15 WS. Then the other relievers would account for 8 more WS if B-Mac was in the rotation. That would be a difference of -2 if Count (17 WS) was moved, still equaling 99 wins oddly enough.

 

If Win Shares are that accurate, even if a couple of players under perform or are injured, the Sox should automatically be looking at 94+ wins.

Edited January 2, 2006 by RME JICO

[Greg Hibbard]

can you explain what the numbers mean?

[RME JICO]

can you explain what the numbers mean?

Here is the definition:

 

Bill James invented Win Shares as a simple way to compare baseball players. The idea was to develop a statistic that allows you to compare shortstops and outfielders, starters and relievers, relievers and shortstops, etc. Since the win is the ultimate measure of success, James developed a stat that measures each player's contribution to his team's wins, or Win Shares.

 

 

You can check out the 2005 Win Shares here:

 

http://www.hardballtimes.com/winshares/index.php

[thedoctor]

very interesting. i'm not a big sabermetrics guy, but that is an interesting projection. it will be interesting to see if those numbers mirror reality.

[santo=dorf]

QUOTE(RME JICO @ Jan 2, 2006 -> 02:31 PM)

Since the win is the ultimate measure of success,

<{POST_SNAPBACK}>

Whoa, did James actually say that?

 

Then why do SABR-heads discount wins by a pitchers and compare lame stats like BABIP or DiPS?

[RME JICO]

very interesting. i'm not a big sabermetrics guy, but that is an interesting projection. it will be interesting to see if those numbers mirror reality.

Even using worse case scenarios it is hard to project anything less than 92 wins.

[TLAK]

Even using worse case scenarios it is hard to project anything less than 92 wins.

Well done, and you are being conservative. I see you are including Paulie's bad season from a couple years ago and Thome's injury year in the averages, no cherry picking here.

 

Like all statistics, winshares measure past performance, and there is no guarentee they will repeat, or that a bunch of guys won't get hurt. There is also no predicting a breakout year -say Crede plays the whole season like he did in October. So they still will need to play the games.

 

We can be berry, berry hopeful on the upcoming season, I still like last year's team better but thats just me looking through my champaign stained glasses.

[Cerbaho-WG]

Whoa, did James actually say that?

 

Then why do SABR-heads discount wins by a pitchers and compare lame stats like BABIP or DiPS?

 

Because wins aren't an accurate value of a pitcher's performance, and BABIP has been shown to predict luck better than any other stat. If you're saying pitcher X is a good pitcher because of wins, you're not very bright. Wins cannot place a value on a pitcher, BABIP can predict dropoffs or better performances, and DiPS, VORP, and MLVr are better ways of valuing pitchers than merely eyeballing them.

[SSH2005]

^^^^^ Yep.

 

Shawn Estes was a stud in 2004. He had 15 wins. Oh wait, his ERA was 5.84.

[Dam8610]

Then why do SABR-heads discount wins by a pitchers and compare lame stats like BABIP or DiPS?

 

Because a pitcher can be average and get 20 wins (see: Andy Pettitte in 2003), or they can be great and get fewer than 15 wins (see: Roger Clemens in 2005). In other words, the win stat for pitchers is far too dependent on factors that the pitcher can't control.

Edited January 3, 2006 by Dam8610

[santo=dorf]

QUOTE(Dam8610 @ Jan 3, 2006 -> 03:45 PM)

Because a pitcher can be average and get 20 wins (see: Andy Pettitte in 2003), or they can be great and get fewer than 15 wins (see: Roger Clemens in 2005). In other words, the win stat for pitchers is far too dependent on factors that the pitcher can't control.

<{POST_SNAPBACK}>

I understand that, but my other question was "Did James really say that?"

[RME JICO]

I understand that, but my other question was "Did James really say that?"

Yes James said that. It is the basis for Win Shares. Team wins not pitcher wins.

[jackie hayes]

I understand that, but my other question was "Did James really say that?"

He's quoting Studes, not James. Looking at a WS article that James wrote, I can't find that quote verbatim, but I think it's a fair summary anyway.

 

I'm not sure where the "Whoa" comes in, I don't know why that statement would be controversial...

[RME JICO]

He's quoting Studes, not James.  Looking at a WS article that James wrote, I can't find that quote verbatim, but I think it's a fair summary anyway.

 

I'm not sure where the "Whoa" comes in, I don't know why that statement would be controversial...

Thats true, I got the info from here:

 

http://www.hardballtimes.com/main/article/...es-have-arrived

[chitownsportsfan]

Studes, in the article you linked says the following:

 

Conversely, they're (win shares) not well-suited for evaluating a player's true talent, so they're also not good for forecasting a player's or team's future performance.

[RME JICO]

Studes, in the article you linked says the following:

Yep, that is correct, but that goes for ALL past performances. Win Shares just puts all the stats into an easy rating scheme for comparisons. There is no way to predict what each individual player will do, and when they will do it, but you can get a ball park figure as long as the player plays close to his past performances.

[jackie hayes]

Yep, that is correct, but that goes for ALL past performances.  Win Shares just puts all the stats into an easy rating scheme for comparisons.  There is no way to predict what each individual player will do, and when they will do it, but you can get a ball park figure as long as the player plays close to his past performances.

How are the projections done? Using past actual WS, or projections of the 'pieces'? I thought it was the latter, which would help answer the question. WS may not be very good at forecasting, b/c they reflect a lot of luck as well as talent, but by being smart with the projections of the component stats, you can try to minimize the contributions of past luck and get a better forecast based on ability.

 

Btw, does anyone publish the other win shares-ish stat, based on the contribution of each player to the probability of the team winning? Anyone know?

[chitownsportsfan]

I believe that is correct Jackie. The recent Bill James book on win shares devoted over 73 pages to explanation. The only way to obtain his explanation (in my understanding the most complete) is to purchase the book. I'd really like to get my hands on it.

 

To your second question, I think that the Hardball Times has it in their annual. Not sure however.

[jackie hayes]

Cool, thanks, I'll check out the HBT annual and see if they have it. Personally, I think that stat is a lot more interesting and intuitive, anyway.