BeltwayHusker
Walk-on
Beware, this is a pretty nerdy post.
In the process of creating Big 10 divisions with "Competitive Balance", there has been some debate over whether Michigan and Penn State should be considered as competitively elite, due to their long histories of success, or whether the relative strength of Iowa and Wisconsin in recent years should be more heavily weighted. This question interests me, so I did a little bit of research...
Here's the method I used: I took the records of teams from 1990-1999, 1975-1999, and all time through 1999, and compared them to the teams' success in 2000 to 2009. I averaged the absolute deviation between each teams winning percentage across the compared periods. As a separate point of comparison, I ranked the winning percentages for all 100 teams (excluding teams that weren't in I-A long enough) and average the absolute deviations in the rankings. Now, I know there are more defensible ways of doing this, statistically (normalizing the data would be a start, since the set with the longer time-horizon naturally has less variance). If anyone for some reason wants to keep playing with this, the data sheet is linked below.
The results? In the year 2000, using 25 years of past history would have predicted success in the next ten years better than using data form only the previous 10 years. Interestingly, using rankings, 25 years was also superior to using all-time records, though this was pretty close.
On average, a team's winning percentage in the 2000's varied by 0.108 from their 25-year winning percentage prior to 2000. Their ranking (by winning percentage) in the 2000's varied on average by 21.8 spots (out of 100) from their ranking over 1975-1999. The team with the biggest variance? TCU: 5th best winning percentage in the 2000's, 90th in the 25 years prior.
So, this would suggest that it's reasonable (though of course no guarantee) to expect Michigan and Penn State to win more than Iowa and Wisconsin over the next 10 years, despite their better records in the last 10 years.
If you want to see the data, here it is: https://spreadsheets.google.com/ccc?key=0Avzywsc6VKg9dFRaYS1VR0kzakZVLU1Gb0RWREpFTnc&hl=en&authkey=CPea6uAF
On a separate note, why can't we have dynamic redistricting, with rivalries protected, every few years, based on a formula?
In the process of creating Big 10 divisions with "Competitive Balance", there has been some debate over whether Michigan and Penn State should be considered as competitively elite, due to their long histories of success, or whether the relative strength of Iowa and Wisconsin in recent years should be more heavily weighted. This question interests me, so I did a little bit of research...
Here's the method I used: I took the records of teams from 1990-1999, 1975-1999, and all time through 1999, and compared them to the teams' success in 2000 to 2009. I averaged the absolute deviation between each teams winning percentage across the compared periods. As a separate point of comparison, I ranked the winning percentages for all 100 teams (excluding teams that weren't in I-A long enough) and average the absolute deviations in the rankings. Now, I know there are more defensible ways of doing this, statistically (normalizing the data would be a start, since the set with the longer time-horizon naturally has less variance). If anyone for some reason wants to keep playing with this, the data sheet is linked below.
The results? In the year 2000, using 25 years of past history would have predicted success in the next ten years better than using data form only the previous 10 years. Interestingly, using rankings, 25 years was also superior to using all-time records, though this was pretty close.
On average, a team's winning percentage in the 2000's varied by 0.108 from their 25-year winning percentage prior to 2000. Their ranking (by winning percentage) in the 2000's varied on average by 21.8 spots (out of 100) from their ranking over 1975-1999. The team with the biggest variance? TCU: 5th best winning percentage in the 2000's, 90th in the 25 years prior.
So, this would suggest that it's reasonable (though of course no guarantee) to expect Michigan and Penn State to win more than Iowa and Wisconsin over the next 10 years, despite their better records in the last 10 years.
If you want to see the data, here it is: https://spreadsheets.google.com/ccc?key=0Avzywsc6VKg9dFRaYS1VR0kzakZVLU1Gb0RWREpFTnc&hl=en&authkey=CPea6uAF
On a separate note, why can't we have dynamic redistricting, with rivalries protected, every few years, based on a formula?
Last edited by a moderator:
