Friday, February 28, 2014

The Encroachment of Sports Analytics

Grantland has a good article on the encroachment of sports analysis in professional sports.

MLB Salary Inequality

With the start of MLB Spring Training underway, I wanted to turn my attention to analyzing baseball over the next few blogs.  For the first, I decided to look at MLB player salary inequality as measure by the Gini coefficient.  So I downloaded Sean Lahman's MLB database and after unzipping the files I used his Salary spreadsheet to calculate the Gini Coefficient for the data that he has from 1985 to 2013.  Here are the results since the 1985 season along with the number of players per season.

Season Gini n
1985 0.3843 550
1986 0.4888 738
1987 0.5104 627
1988 0.5073 663
1989 0.5328 711
1990 0.5332 867
1991 0.5432 685
1992 0.5728 769
1993 0.6257 923
1994 0.6221 884
1995 0.6787 986
1996 0.6623 931
1997 0.6402 925
1998 0.6353 998
1999 0.6290 1006
2000 0.6066 836
2001 0.6092 860
2002 0.6123 846
2003 0.6251 827
2004 0.6335 831
2005 0.6241 831
2006 0.6189 819
2007 0.6099 842
2008 0.6240 856
2009 0.6185 813
2010 0.6225 830
2011 0.6215 839
2012 0.6196 848
2013 0.6168 815

As you can see since the early to mid-90's MLB player salary inequality has remained fairly stable and fairly unequal - similar to South Africa in 2009, but less equal than the US income distribution, which was 0.477 in 2011.

Thursday, February 27, 2014

UMass Fires Head Football Coach

Charley Molnar was fired as the University of Massachusetts head football coach after the 2013 season, after leading the Minutemen to a combined 2-22 record for the last two seasons they competed in the Football Bowl Subdivision.  So I thought that I would take a look at the Minutemen over the last two seasons in terms of their production rankings using the Complex Invasion College Football Production Model, starting with the 2012 season - the first season that I have UMass data analyzed by the model.

Two seasons ago UMass finished at 1-11 overall and 1-7 in the Mid American Conference, while playing against an average strength of schedule as compared to the "league" average.  The Minutemen's only victory was a 22-14 decision against #101 ranked Akron and their worst loss of the season was to #114 ranked Miami (OH).  In terms of on-field performance, this was the worst team in the Football Bowl Subdivision given their overall rank was #124 out of 124 teams.  Why?  One reason was their pathetic offense - ranked #124 or worst in the Football Bowl Subdivision and their defense was not much better, ranked #121 out of 124 teams.  Surely a season to forget.

Last season UMass finished at 1-11 overall and 1-7 in the Mid American Conference again playing against an average strength of schedule.  This season the Minutemen's only victory was over #124 ranked Miami (OH) and their worst loss was to FCS Maine.  In terms of the team's on-field performance UMass finished as the #119 ranked team overall with the #121 ranked offense and the #88 ranked defense.  While UMass is improving (not much room to decline from last season), this was not enough for the school to retain their head football coach.

Analysis of 2013 NCAA FBS Head Coach Changes
Boise State and Chris Petersen
Texas and Mack Brown
Washington and Steve Sarkisian
Wake Forest and Jim Grobe
Wyoming and Dave Christensen
Eastern Michigan and Ron English
Florida Atlantic and Carl Pelini
Miami of Ohio and Don Treadwell
UConn and Paul Pasqualoni
USC and Lane Kiffen

Monday, February 17, 2014

NHL Goalie Performance at the Olympic Break

Well, I am back.  To mark my re-emergence I have decided to look at how NHL goalies have performed up to the Olympic break.  Below are the top 20 NHL goalies using our Wins Above Average measure.

Rank Player Team WAA GS SA SV%
1 Ben Bishop TBL 3.968 44 1229 0.933
2 Jonathan Bernier TOR 3.187 38 1444 0.927
3 Tuukka Rask BOS 2.877 43 1210 0.928
4 Carey Price MTL 2.724 48 1459 0.925
5 Ben Scrivens LAK, EDM 2.717 21 695 0.937
6 Semyon Varlamov COL 2.436 43 1436 0.924
7 Josh Harding MIN 2.228 26 690 0.933
8 Ryan Miller BUF 2.096 39 1373 0.923
9 Cory Schneider NJD 1.630 31 800 0.926
10 Cam Talbot NYR 1.602 15 449 0.935
11 Martin Jones LAK 1.547 12 350 0.94
12 Alex Stalock SJS 1.440 11 353 0.938
13 Anton Khudobin CAR 1.245 19 564 0.927
14 Frederik Andersen ANA 1.151 16 484 0.928
15 Marc-Andre Fleury PIT 1.081 47 1278 0.919
16 Thomas Greiss PHX 1.077 10 352 0.932
17 Philipp Grubauer WSH 0.933 14 458 0.926
18 Al Montoya WPG 0.924 15 495 0.925
19 Steve Mason PHI 0.856 43 1267 0.918
20 Henrik Lundqvist NYR 0.839 43 1243 0.918

Friday, February 7, 2014

NBA Officials and Racial Bias

In my Sports Economics class, the last topic I go over is the Price & Wolfers paper on racial bias by NBA referees.  The authors looked at the 1999 - 2002 NBA seasons and controlling for everything that anyone could think of as to why this would not be true, find that it is true.  This is an astounding result - which is why I save the best for last in Sports Economics.

Subsequently, the authors have looked at the 2003-2006 time period and find the same result, but then once the "cat was out of the bag" meaning once that this result was made public, the racial bias effect has disappeared using 2007-2010.  This is found in the recent issue of Business Week (h/t Will).

Wednesday, February 5, 2014

Olympics and Incentives

At the start of the 2012 Olympics, four women's badminton doubles teams were disqualified for attempting to lose a round-robin game in order to play an easier opponent in the tournament.  As the Winter Olympics start tomorrow, I thought it would be a good time to think about how one would determine if Olympians are shirking or tanking, such as to gain an advantage in later Olympic play.  I naturally thought about using data to make this determination (please note I do not have this data, so I am unable to do this type of analysis).  So I am going to make some assumptions to walk through this type of analysis.

The first area to look at would be serves.  This is an area that should give a clean test of shirking/tanking.  If the player has a history of international play, then they would have a series of data on their serves, especially in games that are important (winning a tournament, escaping elimination, winning a game, etc.) where they have an incentive to perform at their best, and then look at how the variation in their performance under instances where they have an incentive to perform their best and compare that to their serves if there is speculation of shirking/tanking.  Thus as the article mentions, serving into the net (so that their opponent cannot try to unsuccessfully return the shot) would be a very clean test of tanking.

Second would be in service returns, whether they are able to return the serve and the ability to return the serve in play.  Again, one could compare the data series (especially if they have played each other before) and see if there is a statistically significant difference between the two.

Third, one could look at joint tests (given the example above is doubles) and test to see how statistically likely it would be for both players to serve poorly.

Finally (OK, maybe not finally - but finally for this blog), one could look at volleys during the match between the player(s) involved to determine if there is some statistically significant differences in performance during questionable play.

More importantly, designing a tournament such that these types of incentives do not exist is also important; and penalizing players - such as international bans by sport federations might also reduce the need for this type of analysis.