Monday, January 21, 2013

2012-2013 NCAA FBS Conference Strength of Schedule

Last year I wrote about NCAA FBS Conference strength of schedule (SOS), so I thought that I would continue that for this year.  Now that all the NCAA bowl games have been completed - and I have some free time to work on this - here is the NCAA FBS Conference SOS's for the 2012-2013 season (including bowl games).

As a reminder, here is how I measure a team's strength of schedule for a season, (with an adjustment of including four new teams this year, making the total of 124 NCAA FBS schools, instead of 120 as when I originally wrote the blog post).  For conference strength of schedule, I take each team's individual strength of schedule within one conference and average of all of those strength of schedules to find the conference strength of schedule.  For example, let's suppose the four independent's were a "conference", then for Army, BYU, Navy and Notre Dame I am taking the average of each of the four schools individual strength of schedule measures to calculate their conference SOS.

The "league" as a whole has an average strength of schedule of 65.53, and a standard deviation of 9.55. The reason that the "league" average SOS is it not equal to 62.50 - which would be the average of 1 through 124, since there are currently 124 NCAA FBS schools - is because NCAA FBS teams play teams in the football championship subdivision (FCS) and I do not have a model to rank them, so each FCS school is given a rank of #125 for the season. Given the number of FCS schools on FBS teams schedule, the average for all FBS schools increases to 63.53 from 62.50, which is not that much of a difference.

Here are the results - with the SEC having the most difficult strength of schedule for the 2012-13 NCAA FBS season.


Conference SOS
SEC 55.53
Big12 57.29
Pac 12 59.32
Big10 61.99
Big East 62.70
Ind 64.89
ACC 65.47
CUSA 70.56
Sun Belt 72.49
MidAmerican 72.88
WAC 73.38
Mountain West 73.97

Wednesday, January 16, 2013

2012 NFL Pay and Performance

Now that I have some free time, I thought that I would also look at the relationship between payroll and performance in the NFL after having done the competitive balance calculation.  Here is a step-by-step guide to doing this type of analysis using MLB as an example.  So just for the 2012 NFL season (which is admittedly a rather small sample) I am going to estimate the relationship between payroll and performance.  Historically the relationship between payroll and performance in the NFL has been rather small over long periods of time, so we should not expect much of an effect for just one season.

After collecting the data for winning percentage in the 2012 NFL regular season and for payroll for the 2012 regular season I ran the following regression:  winning percent = f(payroll).  There is no need to adjust for relative payroll if only using one season as the average payroll for all the teams is the same for just one season - again this is a small sample size for this type of analysis.

The regression (ordinary least squares) results show that for the 2012 season, payroll has no statistically significant effect on wining percentage.  In other words the effect of team payroll on team performance in the 2012 NFL regular season was statistically zero, as the t test was less than two in absolute value.

Saturday, January 12, 2013

A BCS Model has Notre Dame Number 1

The USA Today has a story that one of the six BCS computer models (Colley Matrix) has Notre Dame as the number 1 team after their loss to Alabama.  My NCAA FBS Production Model has Notre Dame as the #9 most productive team in the nation at the end of the bowl season.

In the past my model has disagreed with the final ranking - say last season when the model had Wisconsin as the most productive team (with three losses) and not Alabama.  This to me is less of a concern for my model since I am interested in ranking teams based on their on-field production as measured by the marginal value of various on field statistics.  Hence a "computer model" can incorporate a number of factors that the human cannot or has an implicit bias as I blogged about over at the wagesofwins.

In an of itself, one model ranking Notre Dame as #1 does not mean that computer or statistical models are invalid, just like one human voting that is contrary to conventional wisdom means that humans should not rank NCAA FBS teams.  Rather, my argument is that there should be more computer NCAA FBS ranking models which would give a more accurate picture of how computer models rank team in the aggregate - as opposed to only six.

Friday, January 11, 2013

Competitive Balance in the NFL

With the NFL regular season over, let's take a look at how competitively balanced the NFL was this past season.  As a reminder, I am using the Noll-Scully measure of competitive balance which statistically measures the actual performance from the ideal performance.  An ideal level of competitive balance using the Noll-Scully measure would equal 1.000, and levels above 1.000 means that the league has some competitive imbalance, with higher Noll-Scully numbers meaning more competitive imbalance.

So here is the Noll-Scully measure of competitive balance for the 2012 NFL regular season.  Using the standard deviation of a population yields a Noll-Scully = 1.524 and using the standard deviation of a sample yields a Noll-Scully of 1.549.  So is this good or bad?  That depends on how competitive the NFL has been in prior years.  So I looked at the regular season standings from 1981 to 2011 and as you can see below the 2012 NFL regular season is very similar in terms of competitive balance to prior seasons.  For those interested, here is my step-by-step guide to calculating the Noll-Scully measure of competitive balance.

Here is the Noll-Scully competitive balance measures using both the population and sample standard deviations from the NFL's 1981 regular season to the 2012 regular season.  As you can see in the table below that the 2012 regular season is not much different than the average during this time period.  Average NS (population) from 1981 to 2012 = 1.505 and Average NS (sample) from 1981 to 2012 = 1.531.

Season
NS Population
NS Sample
1981
1.364
1.389
1982
1.677
1.708
1983
1.370
1.395
1984
1.651
1.681
1985
1.540
1.569
1986
1.639
1.669
1987
1.372
1.397
1988
1.320
1.344
1989
1.415
1.441
1990
1.591
1.620
1991
1.710
1.742
1992
1.625
1.655
1993
1.260
1.283
1994
1.375
1.400
1995
1.204
1.224
1996
1.448
1.473
1997
1.429
1.453
1998
1.682
1.711
1999
1.470
1.494
2000
1.549
1.575
2001
1.600
1.627
2002
1.300
1.320
2003
1.509
1.533
2004
1.515
1.539
2005
1.668
1.694
2006
1.425
1.447
2007
1.634
1.660
2008
1.631
1.657
2009
1.585
1.611
2010
1.474
1.497
2011
1.610
1.636
2012
1.524
1.549

Wednesday, January 9, 2013

2012 Final Top 25

Now that the NCAA FBS season is finally over, with Alabama defeating Notre Dame in the "Title" game, here is the final Top 25 for the season.  Note that the model and experts agree (at least on who is #1), with some very different rankings from #2 on down.  Here is the final Top 25 using the NCAA FBS Production Model.

Rank Team
1 Alabama
2 Florida State
3 Oregon
4 Georgia
5 Utah State
6 Northern Illinois
7 Texas A&M
8 Cincinnati
9 Notre Dame
10 Boise State
11 Fresno State
12 San Jose State
13 Florida
14 LSU
15 Ohio State
16 Rutgers
17 Kansas State
18 Clemson
19 UCF
20 South Carolina
21 Oklahoma State
22 BYU
23 Oklahoma
24 Arizona State
25 Tulsa

Here are this year's Top 25 for the previous weeks:
2012 NCAA FBS Top 25 for Week #15
2012 NCAA FBS Top 25 for Week #14
2012 NCAA FBS Top 25 for Week #13
2012 NCAA FBS Top 25 for Week #12
2012 NCAA FBS Top 25 for Week #11
2012 NCAA FBS Top 25 for Week #10
2012 NCAA FBS Top 25 for Week #9
2012 NCAA FBS Top 25 for Week #8
2012 NCAA FBS Top 25 for Week #7
2012 NCAA FBS Top 25 for Week #6
2012 NCAA FBS Top 25 for Week #5
2012 NCAA FBS Top 25 for Week #4
2012 NCAA FBS Top 25 for Week #3
2012 NCAA FBS Top 25 for Week #2

Monday, January 7, 2013

BCS Title Game 2012

Tonight we have the Alabama Crimson Tide of the Southeastern Conference against the Notre Dame Fighting Irish who are only one of four independent NCAA FBS teams playing for the BCS title.  As I have done for the other 34 bowl games this season, here is how each team looks in terms of productivity over this season using the NCAA FBS Production Model.

Alabama is currently 12-1 while playing against an average strength of schedule.   Alabama's strength of schedule equals 61.54 which is slightly less than the average overall (as of the end of the regular season) of 66.73, but is still within one standard deviation of the mean, and thus still similar in a statistical sense to most of the other teams in the football bowl subdivision.  As of the end of the regular season Alabama and 82 of the 124 FBS schools (including the four new teams that are playing basically an FBS schedule) played a similar schedule in terms of overall strength.  If you quote that Alabama played Georgia, Texas A&M and LSU (all top 40 teams), I will respond with Arkansas, Florida Atlantic, Missouri, Auburn and Western Carolina (all bottom 40 teams).  That said, Alabama's best win was against currently ranked #7 Georgia in the SEC championship game and their only loss was to currently ranked #13 Texas A&M.  Alabama is currently ranked as the #1 team overall in terms of productivity in the NCAA FBS Production Model with the #5 most productive offense and the #3 most productive defense.

Notre Dame is currently 12-0 while playing against a tougher strength of schedule as compared to the rest of the "league".  Notre Dame's best win was against currently ranked #18 Stanford.  Notre Dame is currently ranked as the #5 most productive team in the football bowl subdivision with the #50 ranked offense and the #1 ranked defense in the nation.

Given that Alabama is more productive than Notre Dame, the model "predicts" that Alabama will be declared the national champion by winning this game.

Sunday, January 6, 2013

GoDaddy Bowl 2012

In the GoDaddy Bowl we have the Arkansas State Red Wolves of the Sun Belt Conference playing against the Kent State Golden Flashes of the Mid American Conference.  Here is how each team compares using the NCAA FBS Production Model.

Arkansas State is currently 9-3 playing against an average strength of schedule.  The Red Wolves best game of the season was 50-27 victory against currently ranked #42 Louisiana-Lafayette and their worst performance of the season was a loss to currently ranked #51 Western Kentucky.  Arkansas State is currently ranked as the #23 most productive team in the football bowl subdivision with the #21 most productive offense and the #39 most productive defense.

Kent State is currently 11-2 playing against an easier strength of schedule as compare to the rest of the league.  Kent State's best win this season was over currently ranked #15 Rutgers and their worst loss was to currently ranked #113 Kentucky.  Kent State is currently ranked as the #36 most productive team in the "league", with the #24 most productive offense and the #48 most productive defense.

Given that Arkansas State is more productive than Kent State, the model "predicts" that Arkansas State will be victorious in today's GoDaddy Bowl.

Saturday, January 5, 2013

BBVA Compass Bowl 2012

The BBVA Compass Bowl today has the University of Mississippi (Ole Miss) Rebels of the Southeastern Conference playing against the Pittsburgh Panthers of the Big East Conference.  Here is how each team compares using the NCAA FBS Production Model.

Mississippi is currently 6-6 playing against an average strength of schedule as compared to the rest of the "league".  Mississippi's best win of the season came against currently ranked #41 Mississippi State in order to become bowl eligible; all other five wins were against teams currently ranked 90 or higher.  Mississippi's worst loss was to currently ranked #55 Texas.  Mississippi is currently ranked as the #50 team in overall production with the #58 ranked offense and the #40 ranked defense.

Pittsburgh is also currently 6-6 playing against an average strength of schedule.  Pittsburgh's best win this season was against currently ranked #15 Rutgers and their worst loss was to Youngstown State (a non-FBS team), which in my measure of strength of schedule is equal to 125.  Pittsburgh is currently ranked as the #30 most productive team in the "league", with the #61 most productive offense and the #13 most productive defense.

Given that Pittsburgh is more productive than Mississippi, the model "predicts" that Pittsburgh will win the BBVA Compass Bowl today.

Friday, January 4, 2013

Cotton Bowl 2012

The Cotton Bowl has the Oklahoma Sooners of the Big 12 Conference against the Texas A&M Aggies of the SEC.  Here is how each team stacks up using the NCAA FBS Production Model.

Oklahoma is currently 10-2 playing an average strength of schedule.  Oklahoma's best victory of the season was against currently ranked #43 TCU.  Both of the Sooner's losses were to currently ranked top 20 teams in the NCAA FBS Production model (#5 Notre Dame and #16 Kansas State).  Oklahoma is currently the #26 most productive team in the nation with the #19 most productive offense and the #42 most productive defense.

Texas A&M is also 10-2 and has also played an average strength of schedule.  Texas A&M's best win was against currently ranked #1 Alabama and their two losses were to teams ranked in the top 20 (#9 Florida and #14 LSU).  Texas A&M is currently the #13 most productive team in the FBS with the #6 most productive offense and the #43 most productive defense.

Given that Texas A&M is more productive than Oklahoma, the model "predicts" that Texas A&M will win tonight's Cotton Bowl.

Thursday, January 3, 2013

Fiesta Bowl 2012

This year's Fiesta Bowl has the Kansas State Wildcats of the Big 12 Conference playing against the University of Oregon Ducks of the PAC 12 Conference.  Here is how each team looks using the NCAA FBS Production Model.

Kansas State is currently 11-1 playing against an average strength of schedule.  Kansas State's best win was over currently ranked #26 Oklahoma and was defeated only once by currently ranked #73 Baylor.  Kansas State is currently the 16th most productive team in the "league" with the #26 ranked offense and the #21 ranked defense.

Oregon is currently also 11-1 playing against an average strength of schedule.  Oregon's best win was over currently ranked #8 Fresno State and their only loss was to currently ranked #18 Stanford.  Oregon is currently the #2 team in the nation in terms of overall productivity, with the best offense (#1 in the nation) and the #27 ranked defense.

Given that Oregon is more productive than Kansas State, the model "predicts" that Oregon will win the Fiesta Bowl tonight.

Wednesday, January 2, 2013

Sugar Bowl 2012

The Sugar Bowl pits the Florida Gators of the Southeastern Conference against the Louisville Cardinals of the Big East Conference.  Here is how each team looks in terms of the NCAA FBS Production Model.

The Florida Gators are currently 11-1 playing against a tougher strength of schedule than the average for the "league".  Florida's most impressive wins was over currently ranked #4 Florida State and their only loss was to currently ranked #7 Georgia.  Florida is currently ranked as the #9 team in overall productivity with the #63 most productive offense and the #2 most productive defense - yes even better than one of the two teams playing for the "national championship" in a few days.  Who, you ask?  You will have to wait until I post the blog for the "national championship" game the day of the game.

Louisville is currently 10-2 playing against an average strength of schedule.  Louisville's best game was a win over currently ranked #11 Cincinnati and of their two losses, the worst of the two was to currently ranked #65 Connecticut.  Louisville is currently ranked as the #37 most productive team in the "league", with the #42 most productive offense and the #30 most productive defense.

Florida is more productive overall as compare to Louisville, thus the model would "predict" that Florida will win tonight's Sugar Bowl.

Tuesday, January 1, 2013

Orange Bowl 2012

The Orange Bowl hosts the Florida State Seminoles of the Atlantic Coast Conference and the Northern Illinois Huskies of the Mid American Conference.  Here is a look at each team using the NCAA FBS Production Model.

Florida State is currently 11-2 playing an easier strength of schedule as compared to the average strength of schedule for the "league".  Florida State's most impressive victory was against currently ranked #24 Clemson followed by their ACC Championship game victory over #59 Georgia Tech.  The Seminoles other nine wins were against teams currently ranked in the bottom half of the NCAA FBS.  The Seminoles worst loss (of only two) was to #70 North Carolina State.  Florida State is currently ranked as the #4 team in overall production with the #7 ranked offense and the #11 ranked defense.

Northern Illinois is currently 12-1 playing against a much easier strength of schedule than the average for the "league".  NIU is only one of three teams that have played a "much easier" strength of schedule meaning that their strength of schedule is greater than two standard deviations from the average league strength of schedule.  To put this into perspective, NIU only played two teams in the top half of the NCAA FBS production ranking (both wins) and the best of those two above average teams was against #36 Kent State.  NIU's only loss was by one point to currently ranked #78 Iowa.  NIU is currently ranked as the #3 team in overall production in the NCAA FBS, with the #3 ranked offense and the #18 ranked defense.

According to the model, these two teams are very evenly ranked, but given that NIU is slightly more productive than Florida State, the model "predicts" that Northern Illinois University will be the winner in tonight's Orange Bowl game.

Rose Bowl 2012

The "grand-daddy of them all" - the Rose Bowl - is being played by the Stanford Cardinal of the PAC 12 Conference and the Wisconsin Badgers of the Big 10 Conference.  Using the NCAA FBS Production Model, let's see how productive each team has been this season.

Stanford is currently 11-2 after defeating UCLA in the PAC 12 championship game, and playing a tougher strength of schedule than the league average.  Stanford's best win of the season was over currently ranked #2 University of Oregon and their worst loss was to currently ranked #52 University of Washington.  Stanford is currently ranked as the #18 most productive team overall with the #38 most productive offense and the #16 most productive defense.

Wisconsin is currently 8-5 after defeating Nebraska in the Big 10 championship game while playing against an average strength of schedule.  Wisconsin's best win of the season was against currently ranked #6 Utah State and their worst loss was to currently ranked #46 Michigan State.  Wisconsin is currently ranked as the #25 team in overall productivity with the #34 ranked offense and the #26 ranked defense.

Given Stanford is ranked higher in overall productivity than Wisconsin, the model "predicts" that Stanford will win the Rose Bowl today.

Capital One Bowl 2012

The Capital One Bowl has the Georgia Bulldogs of the Southeastern Conference playing the Nebraska Cornhuskers of the Big 10 Conference.  Here is how each looks using the NCAA FBS Production Model.

Georgia finished the regular season at 11-1, with a loss in the SEC championship game to currently ranked #1 Alabama, and the Bulldogs are currently 11-2 playing against an average strength of schedule.  Georgia's best win this season was against currently ranked #9 Florida and their worst (of their two losses) was to currently ranked #27 South Carolina.  Georgia is currently ranked in terms of overall production as the #7 team in the football bowl subdivision, with the #4 ranked offense and the #22 ranked defense.  Yes, one of the best teams in the SEC played a strength of schedule that is considered average in that is is within one standard deviation of the "leagues" average strength of schedule.  Later on this month, I will write a blog on the different conferences strength of schedule and investigate this in further detail.

Nebraska is currently 10-3 playing against a tougher than average strength of schedule.  In other words, Nebraska has played better opponents this season than Georgia in a statistically measurable way.  Nebraska's best win was over currently ranked #23 Arkansas State and their worst loss was in the Big 10 championship game to currently ranked #25 Wisconsin.  Nebraska is currently the #32 most productive team in the football bowl subdivision with the #14 most productive offense and the #59 most productive defense.

Given that Georgia is more productive on both offense and defense, the model "predicts" that Georgia will be the winner in today's Capital One Bowl.

Outback Bowl 2012

Today's Outback Bowl features the University of Michigan Wolverines of the Big 10 Conference against the University of South Carolina Gamecocks of the Southeastern Conference.  Let's take a look at how productive each team has been this year using the NCAA FBS Production Model.

Michigan is currently 8-4 playing against an average strength of schedule.  Michigan most impressive win was against currently #45 ranked Northwestern and their worst performance was to currently #32 ranked Nebraska.  Michigan's other three losses were to team's currently ranked in the top 10 in the NCAA FBS Production Model.  Michigan currently is the #33 ranked team in overall productivity, with the #68 ranked offense and the #15 ranked defense.

South Carolina finished the regular season at 10-2 playing against an average strength of schedule.  South Carolina's most impressive game was a victory over currently #7 ranked Georgia and their worst loss (they only had two) was to currently #14 LSU.  South Carolina is currently the #27 most productive team overall with the #55 ranked offense and the #14 ranked defense.

Given that South Carolina is superior to Michigan on both offense and defense, the model would "predict" that South Carolina will be the winner in today's Outback Bowl.  Yet this game may be closer than some think.

Gator Bowl 2012

Today's Gator Bowl features the Mississippi State Bulldogs from the Southeastern Conference and the Northwestern Wildcats from the Big 10 Conference - so it is a game between cats and dogs.  Here's a look using the teams on-field production this year using the NCAA FBS Production Model.

Mississippi State finished the regular season at 8-4 playing against an average strength of schedule.  The Bulldogs most impressive win (victory over highest ranked team using the NCAA FBS Production Model) was over currently ranked #54 Troy and all seven of their other victories were over teams in the bottom half of the NCAA FBS production model rankings.  All four of their losses were to teams in the top half of the production model's rankings.  Mississippi State is currently ranked as the #41 team in overall productivity, with the #45 ranked offense and the #33 ranked defense.

Northwestern finished the regular season at 9-3 playing against an average strength of schedule.  The Wildcats most impressive win was over currently ranked #39 Vanderbilt.  All three of Northwestern's losses were to teams currently ranked in the top quartile of the NCAA FBS Production model rankings.  Northwestern is currently the #45 most productive team in the nation with the #46 ranked offense and the #38 ranked defense.

This game (in terms of the model) is fairly even, with a slight nod to Mississippi State.  So the model would "predict" that Mississippi State will be victorious in today's Gator Bowl, but with not much confidence.

Heart of Dallas Bowl 2012

Happy New Year!

The Heart of Dallas Bowl pits the Oklahoma State Cowboys from the Big 12 Conference against the Purdue Boilermakers out of the Big 10 Conference.  Here's the tale of the tape in terms from the NCAA production model for these two teams.

The Oklahoma State Cowboys finished the regular season at 7-5 playing an average strength of schedule.  The Cowboys most impressive win was 65-24 stomping over currently ranked #42 Louisiana-Layfatte, and their most disappointing performance was a 34-41 loss to #?73 Baylor at the end of the season.  The Oklahoma State Cowboys are currently the #49 team overall in the league with the #16 ranked offense and the #86 ranked defense.

The Purdue Boilermakers finished the regular season at 6-6 also playing an average strength of schedule.  The Boilermakers have recently fired their head football coach Danny Hope and replaced him with Darrell Hazell (former head football coach at Kent State).  In terms of on field production, the Boilermakers most impressive win this season was against currently ranked #78 Iowa followed by a win over currently #87 ranked Marshall.  In fact the other four Boilermaker victories this season were against team ranked in the top 100.  Purdue currently is the #58 team overall with the #57 offense and the #58 ranked defense.  Thus they are about as average a team as there is in the football bowl subdivision.

Given the model has Oklahoma State ranked slightly higher than Purdue, look for Oklahoma State to be victorious today in the Heart of Dallas Bowl.