THE BOOK--Playing The Percentages In Baseball

Great stuff.

Just scanning the list, a couple things that popped out:

Everett is now equal to Hanley according to the fans.  If someone would have told me that a year ago I probably would have been speechless.

A little surprised Alexi Ramirez is there with Utley.  I’ve probably only seen Alexi in a couple handful of games though.

Honestly I was under the impression that Vernon Wells had been declining defensively for some years now and he gets the same rating as a guy who get to a ton of balls in Gomez.

Gomez has huge speed numbers.  In your case, you should compare trait by trait.

There is a halo effect that it may take a year or two for fans to realize that their loved ones aren’t as good as they used to be.

Everett’s arm has fallen off, and this report is a perfect example of how valuable it is.  MGL would regress his UZR the same as he normally would otherwise, and would likely still give him a UZR as best SS in the league, a ludicrous position based on the Twins fans who think he (now) throws horribly.

So in Everett’s case would you want to regress him to say how an average 2B would perform if he were at SS?

Ideally, you regress toward what his tools say he is, which in this case is a “63” for a SS, which is a shade above league average.

Furthermore, he went from someone like an 85 to a 63, and we will likely suspect that players whose toolset is viewed to have drastically been altered to have suffered some enormous fundamental change.  And so, we may even consider that we have a “new” player, and simply regress far more toward our “63” than we would someone else who didn’t go through such a life-changing event.

Victor, I’m working on my ZR-based defensive projection system, and here’s how I’m planning on doing it:

* Using the different tools in the FSR, translate that into a ZR score for that player.
* Then include that ZR score as the mean to which you regress.

Obviously this would affect players with little playing time at that position a lot more than players with a lot of playing time at that position - that means for some players, like Griffey, where the fans’ observation lag behind the underlying reality, the data would overwhelm the fans’ observation.

Not saying this is the only or best way, just how I would do it.

Tango, can point me in the direction of the weights you use for each position.  (Me just being curious to see how every player would do at each position.)

I’m not so sure Everett should be written off.  He may well have had the tools of a “63” at shortstop this year, but maybe that was due to lingering effects of injury, and Everett can regain some of that ability next year, maybe not back to an 85, but 77 or something.

Of course that depends on him finding a team willing to put up with his bat and give him a chance.

Rally is right.  We don’t know how persistent his transformation is.  Is it more like Global Warming (no turning back), or is it more like a 6-month monsoon (where the effect was devastating, but the storm will eventually pass)?  We don’t know.

***

As for the weights, I published them elsewhere in the blog.  Try checking the archives, as I’m a bit stuck at the moment.  If you have no luck, I’ll find it for you tomorrow.

However, I’ve got the “per position” numbers figured for each player.  I can post that file on Monday.

Here’s the link I found:
http://www.insidethebook.com/ee/index.php/site/comments/2007_fans_scouting_report_results/

Some popular players:

Pujols
C - 77
1B - 79
2B - 77
SS - 75
3B - 72
LF - 68
CF - 72
RF - 68

Wright
C - 63
1B - 71
2B - 71
SS - 69
3B - 65
LF - 70
CF - 71
RF - 68

Burrell
C - 50
1B - 37
2B - 36
SS - 38
3B - 48
LF - 30
CF - 33
RF - 32

Hanley
C - 56
1B - 61
2B - 63
SS - 63
3B - 62
LF - 69
CF - 71
RF - 68

Russell Martin
C - 67
1B - 72
2B - 72
SS - 70
3B - 67
LF - 71
CF - 73
RF - 70

Manny
C - 39
1B - 29
2B - 29
SS - 32
3B - 37
LF - 28
CF - 31
RF - 30

And the top 15 players at C and SS respectively:

Rolen, Scott 90
McDonald, John 88
Longoria, Evan 86
Markakis, Nick 86
Beltre, Adrian 85
Tulowitzki, Troy 85
Molina, Yadier 84
Mauer, Joe 84
Ellis, Mark 84
Suzuki, Ichiro 83
Phillips, Brandon 83
Beltran, Carlos 83
Chavez, Endy 82
Vizquel, Omar 82
Gutierrez, Franklin 82

McDonald, John 87
Beltran, Carlos 86
Suzuki, Ichiro 85
Chavez, Endy 85
Rolen, Scott 85
Ellis, Mark 84
Phillips, Brandon 83
Longoria, Evan 83
Vizquel, Omar 83
Beltre, Adrian 83
Gutierrez, Franklin 83
Markakis, Nick 82
Tulowitzki, Troy 81
Furcal, Rafael 81
Fukudome, Kosuke 81

Tango,
The links to previous years from the 2008 results page (http://tangotiger.net/scouting/scoutResults2008.html)
don’t seem to be working.  I know your main scouting page (http://tangotiger.net/scouting/) has previous year results, but they don’t have Player IDs listed with them.  I was wondering if you have published previous years’ data with STATS or (preferably) BIS or really any other common ID listed.

Thanks very much for your help, and for conducting this project year after year.  It’s always a lot of fun to go through the results.

Thanks for pointing that out.  I’ll fix that.

As for the playerid, until I can supply that for you, see if you can make due with VIEW/SOURCE.  The playerid is embedded in each player’s link.  Like here:
http://tangotiger.net/scouting/pos2007_2B.html

If you do VIEW/SOURCE, you will see a line like:
{a href=’sim2007_6899.html’>Ellis, Mark{/a>

So,
1. simply load the whole file into a text editor
2. sort all the data
3. delete everything that doesn’t start with a href
4. replace “{a href=’sim2007_” with a null string
5. replace “.html’>“ with a colon ( : )
6. replace “{/a>“ with a null string

(Change all the braces for the less-than sign.)

You will end up with
6899:Ellis, Mark
and all the other players on the page

Repeat for the other 7 positions.  (Or, if you don’t care about positions, in step 1, copy all the files together into one file first.)

Hope that helps if you are in a rush.  Otherwise, I’ll post the proper data at some point in the winter.

CSV files are available for 2004-2007:

http://www.tangotiger.net/scouting/files/

They’re not consistent with the IDs they use year-to-year, however. I do have a set of the FSR data remapped to use the same ID consistantly, and I really need to republish that at some point (I put a previous version of it up on EditGrid, but I’ve added a lot of “missing” player IDs to it as I rebuilt my BDB-to-STATS ID map table.)

Thanks very much guys.  That works just fine.

Someone emailed me about avgN, and asked about decimals. If this wasn’t clear:

avgN: number of ballots (which is really number of data points divided by 7)

just remember that not everyone necessarily evaluates all 7 traits for each player.  Most do 7, but plenty do 6 or 5 or even less.  So, my “count” for the number of ballots is the number of data points per player, divided by 7.

Just bumping.