THE BOOK--Playing The Percentages In Baseball

Do you want the OPS as a decimal (i.e. .750 or just 750)?

Either way.  I’m going to parse them out, since people will put it in, in either format.

http://www.baseballthinkfactory.org/files/newsstand/discussion/the_book_blog/

I think we can assume that most Yankee fans are consumers of Yankee media - YES for the games, and some NY newspaper voice as well.

Therefore, a mass “opinion poll”, like the one proposed here, is flawed for this reason. If Michael Kay, for example, talks about Matsui’s good health every single day on TV, I’m reasonably sure that an opinion sample of fans will believe good things about Matsui’s health.

I’m a believer in Wikipedia, but one of the reasons I think it generally works is that people from many different backgrounds can and do participate. Leaving aside the self-selection evident in those who would read and respond to a statement on Tango’s site, I believe that sports fans generally share much of their information culture.

The flaw is the design!

It doesn’t matter if it’s true that Matsui is healthy or unhealthy according to the YES mouthpieces.  All that I care about is that you report your thoughts on the matter, even if you think you are slightly, or incredibly, influenced by the media or peers.

That’s the point!  I’m consolidating subjective opinions into a number so that we can figure out how much to weight that opinion.  For example, Marcel, in his limited wisdom, predicts Matsui for 371 PA.  It does that because it doesn’t know why Matsui only had 201 PA in 2006.  Now, I could have built a better model that looks at his 2100 PA in previous 3 seasons with an above-average level of performance, and reasoned that he was injured, and not because he lost the manager’s confidence or his skills.  But, Marcel still can’t figure out what kind of injury, or how that injury may or may not carryover.  So, Marcel lumps him in with all other players who were full-time players, and then had 200 PA.

The Wisdom of Crowds will give us an additional regression point, on top of that 371 that Marcel has figured out.  Because of the Crowds, maybe we’ll expect Matsui to have 550 PA or 450 or 150.  Who knows.  That’s why we are doing this.

Same for Papelbon.  Let the Crowd figure out how he’s going to be used, based on however they process the limited biased information.

One thing I’m trying to do for this season (but probably won’t get around to finishing for all 30 teams) is to come up with projection “adjustments” without having looked at the projections themselves.  See the link for an example, but I explained the gist of the idea in the introduction (http://triplesteal.blogspot.com/2007/03/2007-predictions-introduction.html):

“My normal method would be to take all the numbers, look at them for a few minutes, and then adjust them up or down, but there’s a problem: once I see the projections, my adjustments are biased based on what I’ve just seen. If I’m convinced ahead of time that Carlos Delgado is going to hit 40 home runs, then it doesn’t matter what the projection says, because I’m just going to pencil him in for 40 home runs. But if I’m going to do that, what’s the point of looking at the numbers in the first place?”

I think this method may do a good job of capturing some of the more subjective pieces of information (types of injury, scouting information) into a purely statistical projection while retaining the benefits of the objective numbers.  Allow humans to do what they do best, and let the computer do what it does best.

Already have over 100 people submitting their ballots (21 teams total), with 49 of them just for the RedSox!

Based on those responses, they give us this as the likely roster for Boston:
C Jason Varitek
IF Mike Lowell
IF Julio Lugo
IF Kevin Youkilis
IF Dustin Pedroia
OF Manny Ramirez
OF Coco Crisp
OF J.D. Drew
DH David Ortiz

C Doug Mirabelli
IF Alex Cora
IF Eric Hinske
OF Wily Mo Pena
OF David Murphy

SP Tim Wakefield
SP Curt Schilling
SP Josh Beckett
SP Daisuke Matsuzaka
SP Jonathan Papelbon

RP Joel Pineiro
RP Brendan Donnelly
RP Mike Timlin
RP Hideki Okajima
RP Julian Tavarez
RP J.C. Romero

Virtually all put Papelbon (46 of 49) puts Papelbon in the starting rotation, with two having him as the closer.  One person abstained.

Six has Timlin as the closer, 36 as the setup guy, and 2 as the mopup guy.
12 has Donnely as the closer, 33 as the setup guy.
22 has Pineiro as the closer, 14 as the setup guy (or spot starter), 6 as the mopup guy, and 1 expects him to not be on the roster.

Now, you know you are in trouble, when one of your three best relievers is considered either the front-runner for the ace job, or the front-runner to be released.

Pedroia:
- 6 see him playing 150+ games
- 26 have him playing 130-149
- 13 have him with 90-129 games

That works out to around 550 PA.  Marcel has him at 249, which shows how unreliable Marcel is with rookies.  http://www.fangraphs.com/statss.aspx?playerid=8370&position=2B James, Chone, and ZIPS have him between 530-590, with an average of… 554.  Damn, the Redsox fans know their stuff.

Not only will we be able to learn alot, but this serves as a repository for historical purposes.  We’ll now know exactly the outlook of hardcore Redsox fans in March 2007.

3.61: Dice-K (1 SD = 0.36)
3.78: Papelbon as a starter (1 SD = 0.48)
3.91: Schilling (1 SD = 0.22)
4.01: Beckett (1 SD = 0.38)
4.52: Wakefield (1 SD = 0.27)

That’s what the Redox fans think.

Tango, would you like people to input data on individual players rather than the entire team?

As a Mariner fan I’m wondering if I put in Shin-Soo Choo only for the Indians does that help or hurt the study, or have little effect?

It only helps if you base your judgement on information that I wouldn’t have.

Point of clarifications then...so only if I’m basing those projections on more than just other statistical models, but upon media reports, watching the player develop live, etc?

Just let me say that this is a great idea. I’ll post a link to this from my Braves blog later today.

Dave/9: Right.  I mean, if you are a Sox fan, I know you’ve already pored over all the numbers, and you’ve already done your forecast for them, but that you’ve also been exposed to the inner-goings of the team somehow, and so, you’ve clouded your analysis with alot of subjectivity (that’s GOOD).

If on the other hand, you also forecast the Jays, then you will base it almost all on stats, and that doesn’t help me.  I don’t really care how fans use stats, but I do care how they combine stats with subjectivity.  Doing it from the point-of-view of being in the eye of the storm is what I’m after.

300 forecasts so far (123 for the Redsox… thank you SOSH). 42 for the Jays (thank your BattersBox.ca), so their initial results are noted below. 

Four teams not represented at all: Marlins, Rox, Tigers, Pirates.

These teams have less than 5 fans: Angels, Cubs, Whitesox, Royals, Twins, Phillies, Dodgers, Mariners, Astros, Rangers.  I’ll be finding bloggers for these teams, and twisting arms.

========================================
C Gregg Zaun

IF Lyle Overbay
IF Aaron Hill
IF Troy Glaus
IF Royce Clayton

OF Reed Johnson
OF Vernon Wells
OF Alex Rios

DH Frank Thomas

C Jason Phillips
IF Jason Smith
IF John McDonald
OF Adam Lind

P Roy Halladay
P A.J. Burnett
P Gustavo Chacin
P Tomo Ohka

P B.J. Ryan

P Brandon League
P Jason Frasor
P John Thomson

P Shaun Marcum
P Jeremy Accardo
P Brian Tallet
P Francisco Rosario

The relievers are always the interesting ones.  39 of 39 put Ryan as the ace reliever, so no surprise there.

The first interesting case is John Thomson.  19 of you guys have him in the starting rotation, 2 as the spot starter, 1 as the mopup guy, 5 as a callup, and 13 didn’t mark him at all.

The young League has 16 expecting him as the ace reliever, 20 as the setup guy, 1 in mopup duty.  You guys obviously think that 2006 is much more real than his 2005.  A good forecasting engine would therefore accept your comments, and put much more weight on his 2006 data.  I guess I wasn’t clear with “ace reliever”, since I was asking for “role” not “performance”.  Unless those 16 people think that he’s going to share closing duties with Ryan?

Frasor: 5 as the ace reliever, 32 in the setup role, 2 provided no forecast.

Marcum: 3 in the starting rotation, 15 in the setup role, 3 as the mopup, 10 as a callup, 6 with no forecast.

Accardo: 1 as the ace, 20 in the setup role, 10 for mopup, 2 as a callup, 8 with no forecast.

Rosario: 16 as setup, 10 mopup, 5 callup, 9 no forecast.

Tallet: 7 as setup, 24 mopup, 2 callup, 7 no forecast.

3.13 Roy Halladay (1 sd = 0.25)
3.56 A.J. Burnett (0.39)
4.67 Gustavo Chacin (0.32)
4.73 Tomo Ohka (0.44)

I posted a link to this on Halos Heaven.  They are doing similar things asking for over/under on Angel players, so hopefully a few will check it out.

Great, thanks!

Here’s a Q&A with the author of Wisdom of the Crowds:

http://www.randomhouse.com/features/wisdomofcrowds/Q&A.html

481 ballots and counting.

The teams needing more Fans:
2 ballots - Marlins, Tigers, Rangers
3 - Pirates
4 - Dodgers
5 - Astros
6 - Brewers, Rays, Royals

Teams (and fans) with great support:
146 - Redsox - http://sonsofsamhorn.net/index.php?showtopic=16472

51 - Jays - http://www.battersbox.ca/article.php?story=20070319142758552

25 - Yanks - http://forums.nyyfans.com/showthread.php?t=102245

24 - Cubs - http://www.northsidebaseball.com/Forum/viewtopic.php?t=39199

22 - Cards - http://gatewayredbirds.com/forum/viewtopic.php?t=14874

19 - Padres - http://ducksnorts.com/blog/2007/03/tangos-community-forecasts.html

17 - Braves - http://chop-n-change.typepad.com/chopnchange/2007/03/community_forec.html

15 - Giants - http://www.mccoveychronicles.com/story/2007/3/19/33455/5596

14 - Rockies - http://www.purplerow.com/story/2007/3/20/191538/953

12 - Reds - http://www.redszone.com/forums/showthread.php?t=55629

When I complete the balloting, I’ll make sure to acknowledge every site that linked to this project.

Any information on the OPS figures for the hitters? I’m interested in the Blue Jays, especially.

I’m not setup to parse automatically yet.  There’s cleanups to do, since some people put the decimal and not.  I did the Sox and Jays by hand.

However, I will definitely be doing a full report, just like with the Fielding Report.

581 and counting.

Still looking for fans:
2 - Tigers
3 - Marlins
4 - Pirates
5 - Astros
6 - Royals, Rays

Two new teams got great support:
28 - Twins - http://www.aarongleeman.com/2007_03_18_baseballblog_archive.html#5314013502169787676

22 - Orioles - http://www.orioleshangout.com/forums/showthread.php?t=44251

673 and counting. 

Still looking for fans:
3 - Tigers
3 - Marlins
6 - Astros, Royals, Rays

Two new teams got great support:
18 - Mariners
http://www.lookoutlanding.com/story/2007/3/21/122612/071

18 - Angels
http://www.halosheaven.com/story/2007/3/20/152129/007

759 and counting.

No change for Marlins (3 ballots), Royals (6) and Rays (6).

Big showing for the Dodgers, 41 ballots:
http://dodgerthoughts.baseballtoaster.com/archives/608530.html

Finally support for the Tigers with 15 ballots:
http://motownsports.com/forums/showthread.php?p=1047735

796 and slowing down.  Here’s the complete count, along with top referring site for each team.

n tm top referrer
175 bos http://sonsofsamhorn.net/index.php?showtopic=16472
70 tor http://www.battersbox.ca/article.php?story=20070319142758552
58 bal http://www.orioleshangout.com/forums/showthread.php?t=44251
48 min http://www.aarongleeman.com/2007_03_18_baseballblog_archive.html#5314013502169787676
43 la.  http://dodgerthoughts.baseballtoaster.com/archives/608530.html
37 nyy http://forums.nyyfans.com/showthread.php?t=102245
30 chc http://www.northsidebaseball.com/Forum/viewtopic.php?t=39199
27 stl http://gatewayredbirds.com/forum/viewtopic.php?t=14874
26 atl http://chop-n-change.typepad.com/chopnchange/2007/03/community_forec.html
24 sd.  http://ducksnorts.com/blog/2007/03/tangos-community-forecasts.html
23 ana http://www.halosheaven.com/story/2007/3/20/152129/007
20 nym
19 sea http://www.lookoutlanding.com/story/2007/3/21/122612/071
19 det http://motownsports.com/forums/showthread.php?p=1047735
18 cin http://www.redszone.com/forums/showthread.php?t=55629
17 sf.  http://www.mccoveychronicles.com/story/2007/3/19/33455/5596
17 col http://www.purplerow.com/story/2007/3/20/191538/953
14 cle
13 kc.  http://www.royalsreview.com/story/2007/3/26/144345/017
12 ari http://forum.diamondbacksbullpen.org/viewtopic.php?t=1756
11 was http://dcbb.blogspot.com/2007/03/fouled-off-bunts-7-8-3-who-really-knows.html
11 mil
11 cws http://www.whitesoxinteractive.com/vbulletin/showthread.php?t=85541
10 oak
10 tex http://www.newbergreport.com/nrBB/viewtopic.php?t=695
9 pit http://www.bucsdugout.com/story/2007/3/26/144053/200
8 phi
7 hou
6 tb.  http://www.raysbaseball.com/phpBB2/viewtopic.php?t=6932
3 fla
http://www.rotoauthority.com/2007/03/community_proje.html
http://www.hardballtimes.com/main/article/tht-links-rounders/
http://forums.somethingawful.com/showthread.php?threadid=2385453
http://www.baseballmusings.com/archives/019609.php
http://www.baseballthinkfactory.org/files/royals/discussion/tangotigers_doing_community_forecasts/
http://www.ootpdevelopments.com/board/showthread.php?t=142111
http://www.projectprospect.com/display/ShowPost?moduleId=836841&discussionId=15392&postId=203803
http://www.baseball-fever.com/showthread.php?t=58959
http://www.soxaholix.com/tp/2007/03/the_extralingui.html

Now at 928 and counting (average of 31 per team, median of 20).  Balloting will stop on Opening Pitch.

Still lacking support from:
Marlins (3 ballots)
DRays (6)
Astros (7)
Pirates (9)

More support from these sites:
http://www.baseballprospectus.com/unfiltered/?p=289

PHI: http://www.backshegoes.com/bsg/viewtopic.php?t=1180

STL: http://www.vivaelbirdos.com/story/2007/3/27/8436/21209

Mets: http://www.amazinavenue.com/story/2007/3/27/161530/721

A’s: http://www.athleticsnation.com/story/2007/3/27/105832/632

I’ll likely release complete results next week.  If you have any ideas as to what I should produce, do so asap.

This is my plan:
1 - Come up with a “depth” chart, based on the playing time answers
2 - Come up with OPS and ERA forecasts
3 - Combine the two to get team runs scored and allowed

Expand #3 to see how much homerism there is.  That is, are we going to get a total of all 30 teams runs scored to equal their runs allowed?  If not, how much bias is there?  (When I do the fielders, the average score, on a scale of 1 to 5 is 3.3, rather than 3.0)

Compare #2 to Marcel.

***

I’m also going to go through all the forecasting systems (Pecota, Zips, Shandler, Bill James, Chone, MGL, Pete Palmer), and compare them to Marcel and the community.  We’ll revisit them with results in October.

Should be fun…

Anything else you want to see?  If there’s more forecasting engines out there that you’d like to contribute, let me know…

How about Diamond Mind Baseball’s forecast?

FYI, ProTrade seems to have had a similar prediction feature for MLB players this past off-season. It might be worthwhile to see if they would make their data available to you:

I would add the Hardball Times forecast if that hasn’t been mentioned.

I’m working with David at Fangraphs, and this is what the two of us have gotten:
PECOTA, THT, Zips, Chone, DMB, Marcel, MGL, Pete Palmer, Bill James, Shandler.  Plus the community.

I’ll be presenting the nine (sans Marcel) as a group, Marcel, and the community as a group, and we’ll take it from there…

I was hoping to have it done for mid-April, but I’ve got work this weekend, plus it’s tax season.  So, I’ll try to commit for April 30.

Just a little update.  This is the kind of data I received for a hitter’s OPS: -
,290
,590
,6590
,720
,830
,880
.
..598
..650
..750
..840
.00
.000
.001
.015
.100
.125
.200
.222
.234
.241
.244
.250
.256
.265
.275
.290
.298
.300
...

As you can see, I made a terrible choice in not making the OPS a drop-down box.  Not only do I have weird formatting, but clearly virtually all of these refer to OBP, not OPS.  I’m afraid to show you what other kind of data I received. Anyway, back to validation…

Curious if the results were ever published?

Yeah have it ever published before?

I was hoping to do a mid-season report, but I will end up just doing the end-of-season one.  I’m now getting reading for the Fielding Report.

Have the results been published yet?

Yes.  Start at post 29, or jump to 63:
http://www.insidethebook.com/ee/index.php/site/comments/community_forecast_2007_preliminary_results/

good!

yes ,it is right

it is cool

buy giuldwars gold