updated for 9/22

B12 + AAC = combined 5-27. That's hot.

(09-22-2013 09:12 AM)quo vadis Wrote:  updated for 9/22

If UCONN holds on and defeats Michigan, the AAC would have moved up to 5th, is this correct or is my math wrong?

(09-22-2013 06:26 PM)sfink16 Wrote:  

(09-22-2013 09:12 AM)quo vadis Wrote:  updated for 9/22

If UCONN holds on and defeats Michigan, the AAC would have moved up to 5th, is this correct or is my math wrong?

You are correct: if UConn wins the AAC would be 4-16, a .200 win %, higher than the Big 12.
Actually, we would be 4.5-16, because as a top 15 Michigan would qualify as a Quality Win, worth 1.5 wins.

Here is two lists taken from the MWC Board:

Conf vs Conf
1. Pac-12 .875
2. SEC .750
3. Big Ten .714
4. ACC .696
5. Big-12 .684
6. AAC .522
7. Sun Belt .375
8. Indy's .263
9. CUSA .257
10. MAC .194
11. MWC .174

Every conferences record verse the big 5
Obviously does not count conference games from big 5

PAC-12: 6-3
SEC: 6-5
Big ten: 7-6
AAC: 5-6
ACC: 4-5
Big 12: 1-4
Sun Belt: 2-9
C-USA: 2-16
MAC: 1-20
MWC: 0-17

Updated for 9/29.

http://www.colleyrankings.com


Colley's Bias Free Matrix Rankings
An Official Bowl Championship Series Ranking for 2001–2013

(10-01-2013 05:47 PM)HoustonRocks Wrote:  http://www.colleyrankings.com


Colley's Bias Free Matrix Rankings
An Official Bowl Championship Series Ranking for 2001–2013

Looks like Colley mirrors mine. :)

Updated

updated ...

Quo,

After reading your number 2 over and over again I've come to the conclusion that it contains multiple controversial math.

Quote:2) A conference gets credit for 1.5 wins if the win is against any OOC team ranked in the AP and Coaches top 15 at the time the game was played. This is a bonus for a “quality win”.

When discussing "at the time the game is played", it doesn't take into account that many preseason rankings are carry-over rankings from previous year(s). For example, Arkansas was ranked 8th preseason last year but quickly fell to a horrible record to complete the year.

It also doesn't take into account teams that work their way up to the higher rankings from being unranked early in the year when the rankings are so meaningless.

I'm curious as to how those numbers would change when looking at the current week.

For example, if Cincy beat an unranked Purdue team early in the year and Purdue got hot running off 5 consecutive victories to become ranked in the top 15, it would be unfair to Cincy to not pick up the 1.5 points because they did beat a team was highly ranked this year. Obviously in this example, it didn't happen this way but you get my point.

(10-13-2013 08:39 PM)sfink16 Wrote:  Quo,

After reading your number 2 over and over again I've come to the conclusion that it contains multiple controversial math.

Quote:2) A conference gets credit for 1.5 wins if the win is against any OOC team ranked in the AP and Coaches top 15 at the time the game was played. This is a bonus for a “quality win”.

When discussing "at the time the game is played", it doesn't take into account that many preseason rankings are carry-over rankings from previous year(s). For example, Arkansas was ranked 8th preseason last year but quickly fell to a horrible record to complete the year.

It also doesn't take into account teams that work their way up to the higher rankings from being unranked early in the year when the rankings are so meaningless.

I'm curious as to how those numbers would change when looking at the current week.

For example, if Cincy beat an unranked Purdue team early in the year and Purdue got hot running off 5 consecutive victories to become ranked in the top 15, it would be unfair to Cincy to not pick up the 1.5 points because they did beat a team was highly ranked this year. Obviously in this example, it didn't happen this way but you get my point.

I do get your point and you are correct: My method does not take into account the fact that polls are far from perfect and does not factor in a team that was overrated pre-season (and thus at the time the game was played) falling off or a team that was under-rated at the time the game was played later zooming up. I thought about that when I developed my approach, and concluded that it would be (a) too much trouble to keep going back retrospectively and revising rankings based on these kinds of shifts, and (b) philosophically, using current rankings has its own difficulties. E.g., what if Cincy plays a Purdue team ranked #13 at the time the game was played, and beats them. Purdue will surely fall out of the top 15 thanks to the loss, but is it reasonable to not credit Cincy with a big win when it was the fact that they beat Purdue that knocked them out of that category?

So I use "at the time the game was played", understanding that it too is imperfect but on balance simpler to use and arguably less imperfect that the "current ranking" approach.

(10-14-2013 10:58 AM)quo vadis Wrote:  I do get your point and you are correct: My method does not take into account the fact that polls are far from perfect and does not factor in a team that was overrated pre-season (and thus at the time the game was played) falling off or a team that was under-rated at the time the game was played later zooming up. I thought about that when I developed my approach, and concluded that it would be (a) too much trouble to keep going back retrospectively and revising rankings based on these kinds of shifts, and (b) philosophically, using current rankings has its own difficulties. E.g., what if Cincy plays a Purdue team ranked #13 at the time the game was played, and beats them. Purdue will surely fall out of the top 15 thanks to the loss, but is it reasonable to not credit Cincy with a big win when it was the fact that they beat Purdue that knocked them out of that category?

So I use "at the time the game was played", understanding that it too is imperfect but on balance simpler to use and arguably less imperfect that the "current ranking" approach.

Good reply as I expected it to be very difficult to do it the other way. I guess it's a moot point now anyway since BCS is disappearing.

My only other suggestion would have been to withhold step number two until about week 2 or 3 to let the season unfold at least a little bit. Of course the next argument would be what about two top teams playing each other in week one.

I know, there is no perfect system.

(10-14-2013 11:26 AM)sfink16 Wrote:  

(10-14-2013 10:58 AM)quo vadis Wrote:  I do get your point and you are correct: My method does not take into account the fact that polls are far from perfect and does not factor in a team that was overrated pre-season (and thus at the time the game was played) falling off or a team that was under-rated at the time the game was played later zooming up. I thought about that when I developed my approach, and concluded that it would be (a) too much trouble to keep going back retrospectively and revising rankings based on these kinds of shifts, and (b) philosophically, using current rankings has its own difficulties. E.g., what if Cincy plays a Purdue team ranked #13 at the time the game was played, and beats them. Purdue will surely fall out of the top 15 thanks to the loss, but is it reasonable to not credit Cincy with a big win when it was the fact that they beat Purdue that knocked them out of that category?

So I use "at the time the game was played", understanding that it too is imperfect but on balance simpler to use and arguably less imperfect that the "current ranking" approach.

Good reply as I expected it to be very difficult to do it the other way. I guess it's a moot point now anyway since BCS is disappearing.

My only other suggestion would have been to withhold step number two until about week 2 or 3 to let the season unfold at least a little bit. Of course the next argument would be what about two top teams playing each other in week one.

I know, there is no perfect system.

I will think about your suggestion. Suggestions are welcome and I have changed the system in the past based on other's ideas. E.g., the first year i did this i only counted wins against non-AQ teams if they were ranked in the top 25. But someone convinced me that was too restrictive so now i count it as a win if the non-AQ is in the "others receiving votes" category as well. Thanks!

Updated

So BYU and ND count as BCS and Army doesn't? Didn't know that.

(10-20-2013 06:57 AM)sfink16 Wrote:  So BYU and ND count as BCS and Army doesn't? Didn't know that.

In my system, OOC losses always count no matter if the team lost to is a BCS-AQ team or not. So if, e.g., USC loses to Michigan that tags the PAC with a loss, or if they lose to Northern Iowa that counts as a loss as well.

But OOC wins only count as a win for the conference if the team beaten is (a) from a BCS-AQ conference or (b) if the team beaten is not from an AQ conference, then this counts as a win for the conference ONLY if the team beaten is ranked in the coaches or AP poll that week, including the "others receiving votes" category. You can see this explanation in my original post.

So the reason the AAC does not get a win for Temple's victory over Army is that Army is not from an AQ conference and they are not currently ranked in either poll. On the other hand, had Army beaten Temple, the AAC would have gotten tagged with a loss.

So you see that it is tough to pick up wins in my system, that's why even the top conferences have records that don't look that great, e.g. the PAC is #2 with a 10-7 record. That's because the PAC probably has about 20 other OOC wins this year, but since those wins came against non-AQ/non-ranked teams they got no credit for them in my system.

Updated for 10/27

updated

updated