In our book, The Wages of Wins, we show that historically MLB is more competitively balanced now than it was in its early history and that competitive balance in MLB has been improving over the last few decades. Yet, MLB is still not as competitively balanced as European football (soccer) or American football. In The Wages of Wins and in published peer-reviewed journal articles, Dave, Marty and/or I have said this is due to the changes in the population of players in that sport. As we state in our book, this is Steven Gould's argument for the disappearance of the 0.400 hitter in MLB.
So, just how competitively balanced was Major League Baseball in 2011? Using the Noll-Scully competitive balance metric, which calculates competitive balance by taking the standard deviation of actual winning percent and dividing it by the standard deviation of winning percent in an ideal league (i.e. in a league where wins and losses are randomly drawn in a statistical manner) and using the data from the end of the 2011 season, I calculate that the American League's Noll-Scully measure of competitive balance = 1.788 and for the National League the Noll-Scully measure of competitive balance = 1.858. Note if wins and losses are completely random then the Noll-Scully measure of competitive balance equals 1.000.
Thus the American League was about 79% more competitively unbalanced than what we would observe if wins and losses were determined randomly and that the National League was almost 86% more competitively unbalanced than if wins and losses were randomly assigned. So is this high or low? Compared to MLB historically, this year's measure is much more competitive and is much closer to the historical average of the NHL.
Friday, September 30, 2011
Thursday, September 29, 2011
Syracuse Should Vacate Toledo Win
Officials in the Syracuse-Toledo game missed a call on an extra point, that ended up resulting in the game being tied at the end of regulation. If the call would have been made correctly, Toledo would have won the game in regulation. Mid-American conference officials are upset that the call was made incorrectly. Rightly so. Syracuse should vacate this win and Toledo should be awarded the win. I will be watching this for the rest of this season. If Syracuse wins six games and becomes bowl eligible, or if Toledo wins five games and does not become bowl eligible, this will be another black eye on college football.
Here is a clip from the Big East Network.
Here is a clip from the Big East Network.
Wednesday, September 28, 2011
New Mexico Fires Head Football Coach
The USAToday reports that the University of New Mexico has fired their head football coach Mike Locksley. It was rumored during the end of the 2010 season that Locksley would be fired, but some feel that the $1.5 million buyout was holding up making a coaching change.
NCAA rules violation (academic fraud) by assistant coaches under the previous head coach limited the number of scholarships Locksley could offer in the 2009 and 2010 seasons. This has likely impacted the team's recent performance.
In 2008, New Mexico's head coach was Rocky Long, and the Lobos finished 4-8. In that season according to the NCAA FBS Production model, New Mexico ranked #82 in offensive production, #37 in defensive production and #66 in overall total production. In terms of strength of schedule, New Mexico had a 59.17, which is close to average. (An exactly average strength of schedule equals 60.5).
After the 2008 season, head coach Rocky Long left New Mexico, and the Lobos hired Mike Locksley for the 2009 season. New Mexico finished 1-11 that season, with their only win over Colorado State. In terms of production, New Mexico ranked #113 in offensive production, #109 in defensive production and #113 in overall total production. The Lobos strength of schedule was 61.17 which was slightly easier than the previous season.
Unfortunately New Mexico's productivity worsened on both sides of the ball in 2010 and hit rock bottom. New Mexico was the lowest ranked (#120) offense and defense, making them the lowest (worst) team in NCAA football bowl subdivision in 2010. Strength of schedule was 60.58, which again is very similar to the previous two seasons, and almost exactly average.
This season (up to 9/25/11) when Locksley was terminated, the New Mexico Lobos were 0-4. Notice they were competitive with Colordo State (at the time the #104 ranked team and with Sam Houston State - a non-FBS school - which I use the default rank of 121). The loss to a non-FBS school, seems to be the last straw at UNM. In terms of production, through Sept. 25th, the Lobos were the #116th ranked offense, the #119th ranked defense, resulting in the worst overall productive team at #120 in the model. New Mexico's Strength of Schedule was equal to 71, meaning that their overall schedule was easier than many of the other schools in the football bowl subdivision.
NCAA rules violation (academic fraud) by assistant coaches under the previous head coach limited the number of scholarships Locksley could offer in the 2009 and 2010 seasons. This has likely impacted the team's recent performance.
In 2008, New Mexico's head coach was Rocky Long, and the Lobos finished 4-8. In that season according to the NCAA FBS Production model, New Mexico ranked #82 in offensive production, #37 in defensive production and #66 in overall total production. In terms of strength of schedule, New Mexico had a 59.17, which is close to average. (An exactly average strength of schedule equals 60.5).
After the 2008 season, head coach Rocky Long left New Mexico, and the Lobos hired Mike Locksley for the 2009 season. New Mexico finished 1-11 that season, with their only win over Colorado State. In terms of production, New Mexico ranked #113 in offensive production, #109 in defensive production and #113 in overall total production. The Lobos strength of schedule was 61.17 which was slightly easier than the previous season.
Unfortunately New Mexico's productivity worsened on both sides of the ball in 2010 and hit rock bottom. New Mexico was the lowest ranked (#120) offense and defense, making them the lowest (worst) team in NCAA football bowl subdivision in 2010. Strength of schedule was 60.58, which again is very similar to the previous two seasons, and almost exactly average.
This season (up to 9/25/11) when Locksley was terminated, the New Mexico Lobos were 0-4. Notice they were competitive with Colordo State (at the time the #104 ranked team and with Sam Houston State - a non-FBS school - which I use the default rank of 121). The loss to a non-FBS school, seems to be the last straw at UNM. In terms of production, through Sept. 25th, the Lobos were the #116th ranked offense, the #119th ranked defense, resulting in the worst overall productive team at #120 in the model. New Mexico's Strength of Schedule was equal to 71, meaning that their overall schedule was easier than many of the other schools in the football bowl subdivision.
| Date | Opponent | PF | PA | SOS | Venue | |||||
| 9/3/2011 | Colorado St. | 10 | 14 | 104 | Home | |||||
| 9/10/2011 | Arkansas | 3 | 52 | 29 | Away | |||||
| 9/17/2011 | Texas Tech | 13 | 59 | 30 | Home | |||||
| 9/24/2011 | Sam Houston St. | 45 | 48 | 121 | Home | |||||
| 10/1/2011 | New Mexico St. | Home | ||||||||
| 10/15/2011 | Nevada | Away | ||||||||
| 10/22/2011 | TCU | Away | ||||||||
| 10/29/2011 | Air Force | Home | ||||||||
| 11/5/2011 | San Diego St. | Away | ||||||||
| 11/12/2011 | UNLV | Home | ||||||||
| 11/19/2011 | Wyoming | Away | ||||||||
| 12/3/2011 | Boise St. | Away |
Tuesday, September 27, 2011
NHL and NBA Popularity
Today the Daily Iowan has an article arguing that the NHL will be more popular than the NBA, primarily because the NBA lockout will lower fan interest. As demonstrated in our book, The Wages of Wins, we show that fan game attendance does not change in a statistically significant manner after a work stoppage (strike or lockout) in either the NBA or the NHL. So in terms of fan spectator attendance the article is on thin ice.
What about television ratings? In my opinion, this is an important question that has not been resolved in any systematic way (that I know of anyway). In other words I have not seen any research on how fans preferences for the NHL or the NBA is changing. What I have seen is an article from AdWeek on the Nielsen Ratings of the recent NBA and NHL finals showing that the NBA ratings are much greater than the NHL Stanley Cup finals in 2011. US Television ratings have been fairly stable for the NHL according to this data.
While I admit this is anecdotal evidence, it does not support the underlying argument by the Daily Iowan writer that the NHL is poised to overcome the NBA in popularity.
What about television ratings? In my opinion, this is an important question that has not been resolved in any systematic way (that I know of anyway). In other words I have not seen any research on how fans preferences for the NHL or the NBA is changing. What I have seen is an article from AdWeek on the Nielsen Ratings of the recent NBA and NHL finals showing that the NBA ratings are much greater than the NHL Stanley Cup finals in 2011. US Television ratings have been fairly stable for the NHL according to this data.
While I admit this is anecdotal evidence, it does not support the underlying argument by the Daily Iowan writer that the NHL is poised to overcome the NBA in popularity.
Monday, September 26, 2011
NCAA FBS Top 25 Production - Week 4
With week four completed, here is the top 25 (in terms of production) using weights from three different time periods. Special thanks to cfbstats.com for the data. The first is using the estimated weights just from this season (2011) , the second is using the estimated weights from all of the 2010 season, and the third is using the estimated weights over the 2008-2010 seasons. As you can see the model is starting to converge in terms of the teams. While the 2011 model is still pre-liminary, it is getting better as there is more variation in the data with each additional week. My expectation is that by the time the BCS starts ranking the top 25 NCAA FBS teams, that I will only be using the weights from the 2011 production data. Here are the top 25 production results.
Week 3 Top 25 NCAA FBS Production Ranking
| 2011 Weights | 2010 Weights | 2008-2010 Weights | ||||||
| Rank | School | School | School | |||||
| 1 | Wisconsin | 1 | Wisconsin | 1 | Wisconsin | |||
| 2 | Georgia Tech | 2 | South Florida | 2 | South Florida | |||
| 3 | South Florida | 3 | Florida | 3 | Florida | |||
| 4 | Alabama | 4 | Illinois | 4 | Virginia Tech | |||
| 5 | Florida | 5 | Stanford | 5 | Georgia Tech | |||
| 6 | Houston | 6 | Texas | 6 | Oregon | |||
| 7 | Illinois | 7 | Alabama | 7 | Houston | |||
| 8 | Michigan State | 8 | Virginia Tech | 8 | Illinois | |||
| 9 | Virginia Tech | 9 | LSU | 9 | Michigan State | |||
| 10 | Baylor | 10 | Georgia Tech | 10 | Arizona State | |||
| 11 | Cincinnati | 11 | UCF | 11 | Texas | |||
| 12 | UCF | 12 | Michigan State | 12 | Cincinnati | |||
| 13 | Oklahoma State | 13 | Cincinnati | 13 | Alabama | |||
| 14 | Ohio | 14 | Penn State | 14 | North Carolina | |||
| 15 | Bowling Green | 15 | Baylor | 15 | Stanford | |||
| 16 | Florida State | 16 | Texas Tech | 16 | West Virginia | |||
| 17 | Temple | 17 | Houston | 17 | Baylor | |||
| 18 | SMU | 18 | Oklahoma State | 18 | Penn State | |||
| 19 | LSU | 19 | Temple | 19 | LSU | |||
| 20 | Georgia | 20 | Utah | 20 | Michigan | |||
| 21 | Missouri | 21 | Arizona State | 21 | Texas Tech | |||
| 22 | Texas | 22 | Florida State | 22 | UCF | |||
| 23 | West Virginia | 23 | Michigan | 23 | SMU | |||
| 24 | Nebraska | 24 | Ohio | 24 | California | |||
| 25 | Michigan | 25 | Oregon | 25 | Oklahoma State | |||
Week 3 Top 25 NCAA FBS Production Ranking
Monday, September 19, 2011
Preliminary NCAA Football Top 25 Week 3
OK, this might be a little premature, but I have been working on downloading the NCAA FBS data and now have the spreadsheet set up so that I should just have to open up the production data file and it automatically update with the newest data - hopefully. Given that the data importing is going quicker for me this year, it still may be preliminary to attempt to use the NCAA FBS production model for ranking the top 25 FBS (football bowl subdivision) programs, so I am going to give a few answers, since at this point in the season I am not convinced that there is one best time period to use. The reason for this is that there is still not enough information (data) and variation in the data to be completely reliable as to the results from the production model. Hence I am going to have three top 25 ranking and you will see that the rankings will be difference depending on the time period that I use.
The first time period covers the 2008-2010 NCAA FBS seasons for the NCAA FBS production model. Using this longer time period, we see that Georgia Tech is currently as the most production football team, primarily because of the strength of their offense. Now I suspect things will be much different next week after they play North Carolina given the Tar Heels strong defense. But the model does not rank teams as to how I expect them to perform in the future, but rather on their actual performance. What is surprising is that Oklahoma is not ranked in the top 25 at all. Again, before you throw out the model, it is very early (i.e. preliminary) to feel all that confident in the exactness of the rankings. One of the advantages of using a longer time period is that the statistical reliability of the model is superior, but the disadvantage is that we are ranking 2011 team's using a model that includes team performance results from three seasons ago. Anyway, here is the top 25 using the 2008-2010 time period for estimating offense and defense productivity.
In order to give you a feel as to how the Top 25 rankings can change by using different time periods, below is the top 25 using just the data from the 2010 season. While just using the 2010 season is not as academically appealing, this is the time period that I used for the Top 25 rankings for last year. The 2010 time period is statistically valid (which is a plus) but again is a ranking of current teams on past performance. Ignoring this criticism, we see that LSU is the highest ranked team in terms of production, primarily due to the strength of their defense.
Notice that Oklahoma is ranked #18 using the 2010 time period, yet Florida State (which just lost to Oklahoma) is ranked #8. Again, the model is only taking into account actual on-field performance and not how teams fared in the wins and loss column. I suspect as teams play more games there will be a rise in highly productive teams and thus a fall is lower productive teams. This will be a point I will make repeatedly.
Finally, let's take a look at the top 25 using the same NCAA FBS production model but only using the data so far from the 2011 season. Remember when looking at the ranking below that there has not been enough variation in the data to get a statistically reliable model just using the data from the 2011 season. That is why I included the prior two rankings. As the season progresses, I will only use the data from the 2011 season, and when that happens I will drop the other top 25 rankings. Well, here is the Top 25 using just the 2011 season below. Take a look and I will make a few comments below in regard to some surprise teams. Of course I have already stated that this type of rank is premature or preliminary.
Some surprises: Texas Tech at #7. That can be explained by the fact that Texas Tech has played Texas State (a non-FBS team) and New Mexico (one of the worst in the FBS last year and so far this year as well - sorry Lobos fans). Wyoming at #10. Similar story, Wyoming played two non-FBS teams and then played Bowling Green (a below average FBS team). My guess is that Wyoming will fall this week as they play Nebraska.
The first time period covers the 2008-2010 NCAA FBS seasons for the NCAA FBS production model. Using this longer time period, we see that Georgia Tech is currently as the most production football team, primarily because of the strength of their offense. Now I suspect things will be much different next week after they play North Carolina given the Tar Heels strong defense. But the model does not rank teams as to how I expect them to perform in the future, but rather on their actual performance. What is surprising is that Oklahoma is not ranked in the top 25 at all. Again, before you throw out the model, it is very early (i.e. preliminary) to feel all that confident in the exactness of the rankings. One of the advantages of using a longer time period is that the statistical reliability of the model is superior, but the disadvantage is that we are ranking 2011 team's using a model that includes team performance results from three seasons ago. Anyway, here is the top 25 using the 2008-2010 time period for estimating offense and defense productivity.
| Rank | School | |
| 1 | Georgia Tech | |
| 2 | Alabama | |
| 3 | Wisconsin | |
| 4 | Illinois | |
| 5 | LSU | |
| 6 | South Florida | |
| 7 | Arkansas | |
| 8 | Cincinnati | |
| 9 | Stanford | |
| 10 | Vanderbilt | |
| 11 | UCF | |
| 12 | Michigan State | |
| 13 | Florida | |
| 14 | West Virginia | |
| 15 | Michigan | |
| 16 | Virginia Tech | |
| 17 | Purdue | |
| 18 | Rutgers | |
| 19 | Georgia | |
| 20 | Oregon | |
| 21 | Ohio | |
| 22 | Missouri | |
| 23 | Nebraska | |
| 24 | Ohio State | |
| 25 | Texas Tech |
In order to give you a feel as to how the Top 25 rankings can change by using different time periods, below is the top 25 using just the data from the 2010 season. While just using the 2010 season is not as academically appealing, this is the time period that I used for the Top 25 rankings for last year. The 2010 time period is statistically valid (which is a plus) but again is a ranking of current teams on past performance. Ignoring this criticism, we see that LSU is the highest ranked team in terms of production, primarily due to the strength of their defense.
Notice that Oklahoma is ranked #18 using the 2010 time period, yet Florida State (which just lost to Oklahoma) is ranked #8. Again, the model is only taking into account actual on-field performance and not how teams fared in the wins and loss column. I suspect as teams play more games there will be a rise in highly productive teams and thus a fall is lower productive teams. This will be a point I will make repeatedly.
| Rank | School | |
| 1 | LSU | |
| 2 | Texas Tech | |
| 3 | UCF | |
| 4 | Stanford | |
| 5 | Illinois | |
| 6 | Texas | |
| 7 | Alabama | |
| 8 | Florida State | |
| 9 | Georgia Tech | |
| 10 | Florida | |
| 11 | Wisconsin | |
| 12 | West Virginia | |
| 13 | South Florida | |
| 14 | Ohio | |
| 15 | Virginia Tech | |
| 16 | Cincinnati | |
| 17 | Michigan | |
| 18 | Oklahoma | |
| 19 | North Carolina | |
| 20 | Utah | |
| 21 | Temple | |
| 22 | Baylor | |
| 23 | Oklahoma State | |
| 24 | Missouri | |
| 25 | Vanderbilt |
Finally, let's take a look at the top 25 using the same NCAA FBS production model but only using the data so far from the 2011 season. Remember when looking at the ranking below that there has not been enough variation in the data to get a statistically reliable model just using the data from the 2011 season. That is why I included the prior two rankings. As the season progresses, I will only use the data from the 2011 season, and when that happens I will drop the other top 25 rankings. Well, here is the Top 25 using just the 2011 season below. Take a look and I will make a few comments below in regard to some surprise teams. Of course I have already stated that this type of rank is premature or preliminary.
| Rank | School | |
| 1 | Georgia Tech | |
| 2 | North Carolina | |
| 3 | Texas | |
| 4 | Florida | |
| 5 | Virginia Tech | |
| 6 | West Virginia | |
| 7 | Texas Tech | |
| 8 | Stanford | |
| 9 | Illinois | |
| 10 | Wyoming | |
| 11 | South Florida | |
| 12 | LSU | |
| 13 | Florida State | |
| 14 | Wisconsin | |
| 15 | UCF | |
| 16 | Oregon | |
| 17 | Alabama | |
| 18 | Utah | |
| 19 | California | |
| 20 | Michigan | |
| 21 | Cincinnati | |
| 22 | Missouri | |
| 23 | Ohio | |
| 24 | Oklahoma State | |
| 25 | Miami (Florida) |
Some surprises: Texas Tech at #7. That can be explained by the fact that Texas Tech has played Texas State (a non-FBS team) and New Mexico (one of the worst in the FBS last year and so far this year as well - sorry Lobos fans). Wyoming at #10. Similar story, Wyoming played two non-FBS teams and then played Bowling Green (a below average FBS team). My guess is that Wyoming will fall this week as they play Nebraska.
Saturday, September 17, 2011
Monday, September 5, 2011
NCAA Football Model for 2011
I have been recently asked if I will be doing the NCAA football ranking this year, and the answer is yes. I will have to wait a few weeks before I can use only this seasons data, as there is not enough variation in the data for the model to be statistically valid. In the mean time, I am planning on using last year's model (and a model over the last three years to rank the NCAA football teams). I will be unable to do this for the first week, as two teams (Nevada and UAB) have not play yet.
Subscribe to:
Posts (Atom)