Help B-R with Projections: User AQ Ratings are the Final Piece
Posted by Sean Forman on April 1, 2011
Help Determine a Player's Attractiveness Quotient
At B-R we are always trying to acquire new data to give you a fuller picture of the players and of team personnel. For example, a recent deal we signed with team equipment personnel will allow us to soon add shoe size, hat size, and glove and bat models for all current major league players (still working on the minor leagues).
One other area I'm excited about is the new projection system we are working on. Obviously, the system will start with the stats, but from there we are incorporating the whole range of player info: height, weight, comps, HS/College, etc and one additional item that is unique to our particular system.
It is well known that more attractive people earn more money in many fields, but recent research has shown that this may be not because of bias, but because better looking people perform at a higher level (they have more confidence due to their beauty).
In order to add this important and previously unused data point to our player projection system, we are now in the process of accumulating player appearance data, what we call a player's Attractiveness Quotient (AQ). Our initial research shows that 40 points of AQ correlates to 1 point of OPS for hitters (the effect is somewhat lessened for pitchers as hitters are often intimidated by homely pitchers).
This is where you come in. A few tweaks to our recently launched EloRater, and we now have an AqRater (Attractiveness Quotient Rater). Your rankings will drive the final, vital piece of our new Pujols™ projection system (that's its codename as it really will be that much better than systems like Pecota, Cairo, Chone, Zips and Marcel--whoever those last two marginal players are).
In case you were wondering about the implementation details for the AqRater, I can give you a small peek at the proprietary details. We have the same basic universe of players as the EloRater, and I've added some really good looking and not so good looking marginal players as controls.
We did do some initial ratings before launch to test the system. I made my wife (thanks, Honey!) and my mom (thanks, Mom!) spend about 12 hours each rating players. My wife has a thing for glasses, in case you were wondering why Kent Tekulve and Specs Torporcer were rated so high at the start. I expect those to wash out in the months ahead. Dave Davis provided the images. The link to the AqRater is below, have at it.
April 1st, 2011 at 1:39 am
I'd recommend that you use different photos for every year of a players career. 1989 Ken Griffey Jr. may well be more attractive than tubbier 2010 Ken Griffey Jr., and that might account for his dropoff in performance.
Maybe something by age, too. Is 27 the peak of attractiveness as well as of productivity? But when do we determine the player's age? Opening Day, I guess, but it's varied through the years. Use an opening or spring photo, and maybe just use April 1 as a default date for a player's age.
April 1st, 2011 at 2:39 am
"better looking people perform at a higher level (they have more confidence due to their beauty)."
Please explain Babe Ruth then. And Lou Gehrig. Or is this a new development that has become more pronounced as more and more visual media has become available? From better photographs to television to internet to HD TV...after all, JFK's success in the election over Nixon was credited to his appearance on television, while Nixon was clearly the better choice based on common sense.
April 1st, 2011 at 7:44 am
Remind me again; what's the date today?
April 1st, 2011 at 7:51 am
why do none of the player pages have a picture? the only way you can see their pictures are when your rating...its stupid.
April 1st, 2011 at 8:03 am
I don't see why this is necessary. Attractiveness quotient:
1. Don Mossi
2. Everyone else
April 1st, 2011 at 8:42 am
I'm really glad to see BR undertake this project, as it will ultimately supply me with the much-needed data for my work developing the linear weights for xAQ. This variable, xAQ, was the last piece needed for my x$ model used to predict career earnings ($).
Kudos to all a BR!!
April 1st, 2011 at 9:16 am
Oh, people who don't understand. Detroit Michael (@3) explained it already.
April 1st, 2011 at 9:37 am
"Our initial research shows that 40 points of AQ correlates to 1 point of OPS for hitters "
When researching the CHONE system, I initially got similar results, but my correlation completely went away when I added Clu Haywood* to the dataset.
*"Leads the league in most offensive categories, including nosehair".
April 1st, 2011 at 9:55 am
I particularly like the Hot 100.
Well done, Sean!
April 1st, 2011 at 10:09 am
http://www.baseball-reference.com/bullpen/Art_Quirk
April 1st, 2011 at 10:31 am
The ratings don't seem to be changing. I gave a vote to Vida Blue but it didn't get persisted. I know what day it is, but I was kinda hoping this would stay as a frivolous feature. Oh well. 🙂
One thing, though. We do know that post-1960-debut player pictures are here and ready to go if the copyright issues ever get sorted out.
April 1st, 2011 at 10:50 am
David,
They are definitely changing. My vote for Dolph Camili just bumped him up in the ratings by one.
April 1st, 2011 at 11:28 am
Oh OK. I see Ted Lilly has taken the lead in the hot 100 now. I'll resume voting. 🙂
April 1st, 2011 at 11:29 am
I am the standard upon which all attractiveness should begin and end. There lies the template.
April 1st, 2011 at 12:00 pm
Are we going to get the flipside of the Hot 100, i.e. the players deemed least attractive?
April 1st, 2011 at 12:02 pm
Gotta admit, I bit on this one for a few seconds...and then it hit me. LOL.
April 1st, 2011 at 12:04 pm
@14 Paul:
On the other hand, it appears that Now Magazine obtained a picture of me upon which they photoshopped the Mayor's head:
http://www.joeydevilla.com/2011/03/31/mayor-griffin-er-ford-issues-remove-and-dispose-order/
So I would anchor the other end of the spectrum.
April 1st, 2011 at 1:44 pm
I guess this means there are no BR projections coming?
April 1st, 2011 at 2:32 pm
Can't......stop......voting.....
April 1st, 2011 at 6:12 pm
"It is well known that more attractive people earn more money in many fields, but recent research has shown that this may be not because of bias, but because better looking people perform at a higher level (they have more confidence due to their beauty)."
While I'm sure that bias and confidence both play a role, I've always wondered if attractiveness is connected to overall "fitness". I don't mean physical fitness, but evolutionarily, wouldn't a more attractive person do better? Isn't it likely that they would also have better evolutionary traits, like strength, speed, physical acumen, etc?
April 1st, 2011 at 11:09 pm
Is this all a way to introduce having pictures for post-1960 players?
In any event, most of us here are going to be awful raters. Speaking as a heterosexual man, I can make a judgment when it's obvious (Barry Bonds vs. Don Mossi). I keep hitting Skip much more often than I do with the Elo Rater. I should drag my wife over here for a few.
April 1st, 2011 at 11:32 pm
Ah, I drafted my 5-year-old daughter.
April 2nd, 2011 at 10:29 am
Harold Baines always looked like he just woke up, during the pictures from his playing days. Yet he still went on to have a long career and a retired number.
April 3rd, 2011 at 11:08 pm
I'm just catching up on this on April 3, but I sort of had an idea of why this got posted on the date on which it was posted.
This is, of course, something that we female fans have been an expert in doing, although I have generally been interested in players based on other things than just their appearance.
Still, I think that the current starting rotation for the Phillies ranks as one of the best looking in baseball history, even though I am old enough to be the mother of any one of them. (This last item could not have been said if Jamie Moyer were among them, though.)