Sagarin's FCS Top 25 - Week 5

[jmufan999]

All credit/criticism goes to Jeff Sagarin
Top 25 (actual Sagarin ranking w/ FBS teams in parentheses)

1. North Dakota State (57)
2. Youngstown State (75)
3. Delaware (86)
4. Northern Iowa (89)
5. Georgia Southern (90)
6. Illinois State (95)
7. James Madison (97)
8. Eastern Washington (98)
9. Cal Poly (101)
10. Wofford (103)
11. Old Dominion (110)
12. Central Arkansas (111)
13. South Dakota State (112)
14. Northern Arizona (115)
15. Montana (116)
16. Sam Houston State (117)
17. Harvard (118)
18. North Dakota (120)
19. Albany (121)
20. New Hampshire (122)
21. The Citadel (126)
22. Richmond (130)
23. Appalachian State (132)
24. Villanova (133)
25. McNeese State (136)

http://www.usatoday.com/sports/sagarin/fbt12.htm

[UAalum72]

Thru Week 4. Still influenced by starting ratings.

[The Eagle's Cliff]

Thru Week 4. Still influenced by starting ratings.

Always influenced by starting ratings in spite of what Mr. Sagarin claims. His ratings wouldn't work at all if all teams started with a value of 0 and were assigned a variable identifier. For developing an approximate picture, his system is great. Thankfully, a playoff system will be implemented and expanded over the years to diminish the significance of this guess work.

[andy7171]

Thru Week 4. Still influenced by starting ratings.

Are you complaining? At least Albany is ranked. Towson isn't.

[darell1976]

All credit/criticism goes to Jeff Sagarin
Top 25 (actual Sagarin ranking w/ FBS teams in parentheses)

1. North Dakota State (57)
2. Youngstown State (75)
3. Delaware (86)
4. Northern Iowa (89)
5. Georgia Southern (90)
6. Illinois State (95)
7. James Madison (97)
8. Eastern Washington (98)
9. Cal Poly (101)
10. Wofford (103)
11. Old Dominion (110)
12. Central Arkansas (111)
13. South Dakota State (112)
14. Northern Arizona (115)
15. Montana (116)
16. Sam Houston State (117)
17. Harvard (118)
18. North Dakota (120)
19. Albany (121)
20. New Hampshire (122)
21. The Citadel (126)
22. Richmond (130)
23. Appalachian State (132)
24. Villanova (133)
25. McNeese State (136)

http://www.usatoday.com/sports/sagarin/fbt12.htm

No Montana State? Isn't UNI a bit high? I don't buy his rankings.

[wapiti]

No Montana State? Isn't UNI a bit high? I don't buy his rankings.

MSU is right behind McNeese at 137, #26 in FCS.

[Dane96]

Are you complaining? At least Albany is ranked. Towson isn't.

Albany will drop like hotcakes after the Monmouth game is over...actually the game after that, win or lose this weekend.

[PantherRob82]

Pretty decent list of teams, obviously the order could be adjusted some.

[darell1976]

MSU is right behind McNeese at 137, #26 in FCS.

To me thats way to low. I would put MSU in the EWU-Cal Poly area around 8-10 at least. #26 that's below UND.

[Nova09]

The problem a lot of people seem to have with computer rankings is they are based entirely on what a team has done, not what we think a team is capable of doing. as more data is collected, the computers do an excellent job of predicting what a team is capable of based on what it has done. But right now, MSU has a strength of schedule rank of 230 so while they have taken care of business, it is not necessarily all that impressive. What amazes me is that uni has a strength of schedule of 12, which I have never seen from an FCS team other than after week 1 if an FCS played a top 10 FBS

[darell1976]

The problem a lot of people seem to have with computer rankings is they are based entirely on what a team has done, not what we think a team is capable of doing. as more data is collected, the computers do an excellent job of predicting what a team is capable of based on what it has done. But right now, MSU has a strength of schedule rank of 230 so while they have taken care of business, it is not necessarily all that impressive. What amazes me is that uni has a strength of schedule of 12, which I have never seen from an FCS team other than after week 1 if an FCS played a top 10 FBS

NDSU has a SOS of 242 worse than Montana State but yet still at #1. Again if its based on what they have done why is NDSU #1? Robert Morris, Colorado State and Prairie View isn't impressive.

[Nova09]

Always influenced by starting ratings in spite of what Mr. Sagarin claims. His ratings wouldn't work at all if all teams started with a value of 0 and were assigned a variable identifier. For developing an approximate picture, his system is great. Thankfully, a playoff system will be implemented and expanded over the years to diminish the significance of this guess work.

MATH ALERT: This is actually not true and no dummy variable is used.

The Predictor rating comes from a Markov Chain, meaning the current state is used to predict the very next future state. In this case, it means a team's record (and the teams it played to achieve that record) is used to predict the likelihood of winning the next game against a given opponent. Preconceived notions of how good a team would be before the season started are irrelevant once enough data is collected.

EloChess uses linear algebra organizing every team in a Markov Matrix and assigning a value for Team A against Team B in the AB entry of the matrix. Eigenvalues are then used to determine a numerical value for each row, which corresponds to the team represented by that row's rating. Ratings are then put in order to determine rankings.

[Nova09]

Valid point. At this point Sagarin's preseason rankings are still factoring in the equation. Eventually only the true results will be used. I guess he just doesn't have enough data yet, and he acknowledges it is biased at this point.

this was in response to darell

[darell1976]

Valid point. At this point Sagarin's preseason rankings are still factoring in the equation. Eventually only the true results will be used. I guess he just doesn't have enough data yet, and he acknowledges it is biased at this point.

this was in response to darell

True.

[UAalum72]

Are you complaining? At least Albany is ranked. Towson isn't.Not complaining. Expected complaints about Cit vs. GSU, but won't worry about it for another week or two.

[HailSzczur]

Valid point. At this point Sagarin's preseason rankings are still factoring in the equation. Eventually only the true results will be used. I guess he just doesn't have enough data yet, and he acknowledges it is biased at this point.

this was in response to darell

In another week or too this

The teams are NOT WELL-CONNECTED and so all three ratings are BAYESIAN.
will change to something like all the teams are now well-connected. I think thats basically what we're all waiting for.

[LEHIGH61]

All credit/criticism goes to Jeff Sagarin
Top 25 (actual Sagarin ranking w/ FBS teams in parentheses)

1. North Dakota State (57)
2. Youngstown State (75)
3. Delaware (86)
4. Northern Iowa (89)
5. Georgia Southern (90)
6. Illinois State (95)
7. James Madison (97)
8. Eastern Washington (98)
9. Cal Poly (101)
10. Wofford (103)
11. Old Dominion (110)
12. Central Arkansas (111)
13. South Dakota State (112)
14. Northern Arizona (115)
15. Montana (116)
16. Sam Houston State (117)
17. Harvard (118)
18. North Dakota (120)
19. Albany (121)
20. New Hampshire (122)
21. The Citadel (126)
22. Richmond (130)
23. Appalachian State (132)
24. Villanova (133)
25. McNeese State (136)

http://www.usatoday.com/sports/sagarin/fbt12.htm

What planet is he from?

[gotts]

This is funny reading these responses. If you click the link and see that the teams aren't well connected yet, maybe it's telling you that you shouldn't take this seriously yet. Anybody with a math background knows that sample size can skew everything.

[cpalum]

True.

Darell....just an off topic question your signature says "3-1 vs Cal Poly" isn't the all time series 2-2? Cal Poly has won the last two meetings haven't they?...

[BisonBacker]

Darell....just an off topic question your signature says "3-1 vs Cal Poly" isn't the all time series 2-2? Cal Poly has won the last two meetings haven't they?...

I think he's referring to their current season record and the next team they play is Cal-Poly. I don't know what their schedule is but I"m guessing that it's Cal-Poly.

[Jacked_Rabbit]

The lines the sportsbooks set are always very close what Sagarin's rankings predict they'd be. Since this is the case, it would appear Montana State should have great value since they are so low in his rankings. I'll be interested in seeing what their line at Southern Utah is released at this week.

Also, FWIW, 5 of the top 13 are from the Valley.

[cpalum]

I think he's referring to their current season record and the next team they play is Cal-Poly. I don't know what their schedule is but I"m guessing that it's Cal-Poly.

Thanks Backer....I figured I was missing something

[HensRock]

Also, FWIW, 5 of the top 13 are from the Valley.

How about 4 of the Top 6 !!

[The Eagle's Cliff]

MATH ALERT: This is actually not true and no dummy variable is used.

The Predictor rating comes from a Markov Chain, meaning the current state is used to predict the very next future state. In this case, it means a team's record (and the teams it played to achieve that record) is used to predict the likelihood of winning the next game against a given opponent. Preconceived notions of how good a team would be before the season started are irrelevant once enough data is collected.

EloChess uses linear algebra organizing every team in a Markov Matrix and assigning a value for Team A against Team B in the AB entry of the matrix. Eigenvalues are then used to determine a numerical value for each row, which corresponds to the team represented by that row's rating. Ratings are then put in order to determine rankings.

I have actually read his methodology on the website and I'm not going to pretend I understand it.

We all know the difference between Alabama and Savannah State, but how are schools like Toledo, Indiana, Northwestern, etc factored in to SOS? How is human judgement (assumption) avoided?

[Jacked_Rabbit]

How about 4 of the Top 6 !!

Yeah, that's probably not realistic. The MVFC is very strong this year, but I'm pretty sure 4 of the top 6 is not legitimate.

[frozennorth]

Always influenced by starting ratings in spite of what Mr. Sagarin claims. His ratings wouldn't work at all if all teams started with a value of 0 and were assigned a variable identifier. For developing an approximate picture, his system is great. Thankfully, a playoff system will be implemented and expanded over the years to diminish the significance of this guess work.

wrong. completely wrong.

edit: i don't know exactly how his predictor model is structured, since there are quite a few ways to go about it, but at it's basis is probably a process similar to either linear regression or some sort of iterative process, neither of which would need any data other than scores and schedules. elo_chess is similar to basic linear regression as far as i can tell.

[darell1976]

Thanks Backer....I figured I was missing something

Backer is right. After Saturday's game my sig will say 4-1 at Eastern Washington
:D
(Big Sky logo) 2-0

[ysubigred]

Geez!! Y-town ranks 75 but for a BCS bowl game we're 44 LOL!! xcrazyx

[Nova09]

I have actually read his methodology on the website and I'm not going to pretend I understand it.

We all know the difference between Alabama and Savannah State, but how are schools like Toledo, Indiana, Northwestern, etc factored in to SOS? How is human judgement (assumption) avoided?

I'll risk revealing too much about myself since not many nova fans are on here anyway, but I was a math major and wrote my senior thesis on computer rankings so I have a pretty good handle on it. That's not to say I'm an expert or sound like I know better than anyone else, just letting you know that I'm not being argumentative I actually do know where I'm coming from. Having no idea what anyone else's math background is I didn't want to get into it too much, but to give you a short answer usually the strength of schedules are actually computed after the rankings, which might seem counterintuitive. I can't speak to Sagarin specifically because his formulas are protected as intellectual property, but usually a matrix relates all variables which results in an ordering from highest to lowest. Strength of schedule as a concept is rolled into this output, but then based on the outcome a "strength of schedule score" can be calculated easily by looking at the rankings of the teams each team has played. So while it might seem like we use the strength of schedule to determine the strongest team, the computers actually determine which team performed best relative to their unique schedule, and only then do we determine whose schedule was actually hardest.

[Dane96]

Let me cut to the chase:

COMPUTER RANKINGS SUCK.

It's a hypothesis at best.

[ysubigred]

I'll risk revealing too much about myself since not many nova fans are on here anyway, but I was a math major and wrote my senior thesis on computer rankings so I have a pretty good handle on it. That's not to say I'm an expert or sound like I know better than anyone else, just letting you know that I'm not being argumentative I actually do know where I'm coming from. Having no idea what anyone else's math background is I didn't want to get into it too much, but to give you a short answer usually the strength of schedules are actually computed after the rankings, which might seem counterintuitive. I can't speak to Sagarin specifically because his formulas are protected as intellectual property, but usually a matrix relates all variables which results in an ordering from highest to lowest. Strength of schedule as a concept is rolled into this output, but then based on the outcome a "strength of schedule score" can be calculated easily by looking at the rankings of the teams each team has played. So while it might seem like we use the strength of schedule to determine the strongest team, the computers actually determine which team performed best relative to their unique schedule, and only then do we determine whose schedule was actually hardest.

xrulesx Plus 1 ^^ Thanks Nova for clearing that up for us xsalutex

[Jacked_Rabbit]

The lines the sportsbooks set are always very close what Sagarin's rankings predict they'd be. Since this is the case, it would appear Montana State should have great value since they are so low in his rankings. I'll be interested in seeing what their line at Southern Utah is released at this week.

Replying to myself on this to get your toughts on the above hypothesis.

Montana State is favored by 1.5 at Southern Utah.

That seems low to me, as I'd expect MSU to be favored by atleast 3. Anyone with better knowledge of the Big Sky or these two teams have any input?

[unigriff]

as for UNI...they have a goliath opening schedule...i think that influences their computer stats rating. They are a top 5 FCS program currently. Their strength of schedule post this weekend will drop them some spots potentially.

[Saint3333]

Really don't care about these things in September, but if this is a mathimatical calculation how is App that many spots behind Montana and why is Citadel only one spot ahead of App? Preseason rankings still in effect?

[citdog]

as for UNI...they have a goliath opening schedule...i think that influences their computer stats rating. They are a top 5 FCS program currently. Their strength of schedule post this weekend will drop them some spots potentially.

and after the loss on saturday uni will be dropped from all playoff discussion. sad

[asumike83]

Really don't care about these things in September, but if this is a mathimatical calculation how is App that many spots behind Montana and why is Citadel only one spot ahead of App? Preseason rankings still in effect?

Must be. I think GA Southern is a good team but I'm not sure how any computer system without preseason bias could have them at #5. After losing to the #21 team and beating two teams outside of the top 50 (one near the bottom of all DI), that doesn't quite add up.

[nmatsen]

and after the loss on saturday uni will be dropped from all playoff discussion. sad

actually, if, IF IF IF IF IF IF, UNI loses on Saturday they won't be eliminated from playoff discussions until a 5th loss. A 7-4 UNI team (even with a DII win) with losses on the road to top 5, the number 1, defending national champions, and two top half Big 10 schools is a lock for the playoffs. You heard it here first. The 7 win "rule" does not exist. Also, we are 2 years into the 20 team format. Precidents are still be set...

[UNH Fanboi]

actually, if, IF IF IF IF IF IF, UNI loses on Saturday they won't be eliminated from playoff discussions until a 5th loss. A 7-4 UNI team (even with a DII win) with losses on the road to top 5, the number 1, defending national champions, and two top half Big 10 schools is a lock for the playoffs. You heard it here first. The 7 win "rule" does not exist. Also, we are 2 years into the 20 team format. Precidents are still be set...

It doesn't exist, yet it has never been broken. What big wins will UNI have on their resume at 7-4? How will their resume be any better than 2010 JMU (win over top 25 FBS) or 2011 Delaware (wins over two playoff teams)?

[CopperCat]

actually, if, IF IF IF IF IF IF, UNI loses on Saturday they won't be eliminated from playoff discussions until a 5th loss. A 7-4 UNI team (even with a DII win) with losses on the road to top 5, the number 1, defending national champions, and two top half Big 10 schools is a lock for the playoffs. You heard it here first. The 7 win "rule" does not exist. Also, we are 2 years into the 20 team format. Precidents are still be set...

The 20 team format may do some good for UNI down the road. But as I recall the last time a 7-4 team made the field was when MSU got in after it had beaten Colorado and lost to Chadron the following week. While they were 7-4, they still had the 7 D-1 wins. Portland state also had a similar résumé, but did not make the field. If there were four teams at 7-4 i would think at least two or three of them would have the seven D-1 wins and would probably get in over UNI under the 20 team format. Not saying UNI would be out with that next loss, but it wouldn't help their case much. Putting them next to a team like the aforementioned PSU, I would bet the committee would give the nod to PSU nine times out of ten.

[MTfan4life]

The 20 team format may do some good for UNI down the road. But as I recall the last time a 7-4 team made the field was when MSU got in after it had beaten Colorado and lost to Chadron the following week.

Ummm...Actually there have been several 7-4 teams that have gotten in since MSU...Both James Madison and Eastern Kentucky were 7-4 last season and got in as at-large. I know there was a handful the year before. Even before the playoffs expanded from 16 to 20, there were usually a couple 7-4 at-large bids per season.

[Squealofthepig]

actually, if, IF IF IF IF IF IF, UNI loses on Saturday they won't be eliminated from playoff discussions until a 5th loss. A 7-4 UNI team (even with a DII win) with losses on the road to top 5, the number 1, defending national champions, and two top half Big 10 schools is a lock for the playoffs. You heard it here first. The 7 win "rule" does not exist. Also, we are 2 years into the 20 team format. Precidents are still be set...

To support this, name any six D1 win team that has ever been included in the playoffs. Any. Ever. Just one. Simple challenge, right? Name just one.

[Screamin_Eagle174]

actually, if, IF IF IF IF IF IF, UNI loses on Saturday they won't be eliminated from playoff discussions until a 5th loss. A 7-4 UNI team (even with a DII win) with losses on the road to top 5, the number 1, defending national champions, and two top half Big 10 schools is a lock for the playoffs. You heard it here first. The 7 win "rule" does not exist. Also, we are 2 years into the 20 team format. Precidents are still be set...

UM was 7-4 (6 DI wins) in 2010, but had lost to the top 3 teams in the BSC + Cal Poly, and they didn't get in, despite their track record being greater than UNI's, and their AD being head of the playoff selection committee. Not a chance UNI gets in with only 6 DI wins.

[coover]

I think he's referring to their current season record and the next team they play is Cal-Poly. I don't know what their schedule is but I"m guessing that it's Cal-Poly.This is not a complaint about what you said, BisonBacker.

California Polytechnic State University, many years ago, requested the NCAA and any media reporting oon their sport teams use the term "Cal Poly", and only "Cal Poly" when writing about its sports teams. There is no "-" between the names. The school should not be called by its actual name (above). It should not be called "Cal Poly State" or "California Polytechnic". It should not be called "University of California Poly" or any variations of that. It should not be called "Cal Poly San Luis Obispo" or "Cal Poly SLO". And most importantly, it should not be called "Cal Poly Pomona", which is a different school entirely, 250 miles southeast of San Luis Obispo. Call the school "Cal Poly" only, and we shall call your school whatever its proper name is, "NDSU", "North Dakota State", or "North Dakota State University".

By the way ... congratulations on your #1 rating so far this year. I do believe your team deserves it.

[coover]

Backer is right. After Saturday's game my sig will say 4-1 at Eastern Washington
:D
(Big Sky logo) 2-0 or 3-2 at Eastern Washington
(Big Sky) logo 1-1

[JMU2K_DukeDawg]

Really don't care about these things in September, but if this is a mathimatical calculation how is App that many spots behind Montana and why is Citadel only one spot ahead of App? Preseason rankings still in effect?

As with any mathematical calculations, it is determined based on a set of assumptions and the weighted value you give to each assumption in determining your final calculation. I am sure Nova09 could give us more details, but the general answer is assumptions, and it seems that preseason rankings are weighing heavily relative to what has happened in the field of play to date. For example, Lehigh's omission supports this as well.

[coover]

Let me cut to the chase:

COMPUTER RANKINGS SUCK.

It's a hypothesis at best.I was also a math major and know something about Computer Rankings ... You may believe they "suck" and after 4 weeks of play, even Sagarin would tell you they do, but each and every week they get more and more accurate. However, even at the end of the season, variables that are not calculated, cannot be calculated are still present. Such are injuries, illness, mental attitude, improved performance, and others. All rankings cannot measure such. But the truth is that by the end of the season, the computer will be more accurate than a group of voters, most of whom have not even seen more than 20 games between (maybe) 30 teams and, at best, guess which team is the best, which team should be #19, and which team should be #20. But the real problem is the voter who is biased and cannot or will not lose that bias when voting. The computer may be biased at the beginning of the season, but it will never be biased at the end.

[URMite]

Originally Posted by nmatsen
actually, if, IF IF IF IF IF IF, UNI loses on Saturday they won't be eliminated from playoff discussions until a 5th loss. A 7-4 UNI team (even with a DII win) with losses on the road to top 5, the number 1, defending national champions, and two top half Big 10 schools is a lock for the playoffs. You heard it here first. The 7 win "rule" does not exist. Also, we are 2 years into the 20 team format. Precidents are still be set...


It doesn't exist, yet it has never been broken. What big wins will UNI have on their resume at 7-4? How will their resume be any better than 2010 JMU (win over top 25 FBS) or 2011 Delaware (wins over two playoff teams)?

I agree, but there may come a time when it is looked at as a "black swan".

i.e. All have been therefore all will be.

[MTfan4life]

UM was 7-4 (6 DI wins) in 2010, but had lost to the top 3 teams in the BSC + Cal Poly, and they didn't get in, despite their track record being greater than UNI's, and their AD being head of the playoff selection committee. Not a chance UNI gets in with only 6 DI wins.

One thing to add to your UM argument from that year. This got spoken about a lot last year when referring to Delaware's potential berth. We all know the NCAA is very heavily influenced by the $$$. The rich non-seeds get the home games. Knowing the NCAA is all about money, if there was any time the NCAA would have let in a 6 DI win team, it would have been the 2010 Montana team. The average first round attendance was about 8,000 that year. Montana had averaged about 19,000 the previous three seasons for first round games. They simply don't do that. They have enough quality teams over that unwritten rule win level. Maybe once we expand to 24.

[asumike83]

To my knowledge, the only team to ever make the playoffs with less than 7 DI wins was Coastal Carolina in 2010. However, they were not an at-large. They were 0-4 in non-conference play but went 5-1 in the Big South to get the auto bid.

[Nova09]

For those of you who are interested, Kenneth Massey does a much better job of explaining the system in an understandable way than anyone else I've seen, and he's willing to go a lot more in depth than most who do that professionally.

http://masseyratings.com/

Also, I remember when I was studying this stuff one of the well known guys got into it because he was a meteorologist and realized the math that relates complex weather systems in different regions to predict future conditions could be applied to diverse teams who do not necessarily play each other or many common opponents. I have in my head that it was Massey but haven't been able to find anything about that in his bio so probably actually someone else. Just a little anecdote for all.

[Dane96]

I was also a math major and know something about Computer Rankings ... You may believe they "suck" and after 4 weeks of play, even Sagarin would tell you they do, but each and every week they get more and more accurate. However, even at the end of the season, variables that are not calculated, cannot be calculated are still present. Such are injuries, illness, mental attitude, improved performance, and others. All rankings cannot measure such. But the truth is that by the end of the season, the computer will be more accurate than a group of voters, most of whom have not even seen more than 20 games between (maybe) 30 teams and, at best, guess which team is the best, which team should be #19, and which team should be #20. But the real problem is the voter who is biased and cannot or will not lose that bias when voting. The computer may be biased at the beginning of the season, but it will never be biased at the end.


Damn math majors. :) Let me tell you what lawyers calculate: Facts ;)

Show me the last game when an algorithm beat an FCS or FBS team in "Dell Precission M4500" Matrix Stadium.

[Nova09]

Damn math majors. :) Let me tell you what lawyers calculate: Facts ;)

Show me the last game when an algorithm beat an FCS or FBS team in "Dell Precission M4500" Matrix Stadium.

Lol, lawyers don't win anything on the strength of their facts, they win by appealing to emotion. The same way certain teams are given too much/little credit by human voters due to emotional responses.

Anyway I suppose you must make all your investments yourself, or at least only trust people with your money if they swear not to use computers in their analysis. After all, the computers diminish huge networks of people down to key financial data points. Computers can't see the true inner workings of a corporation! Computers don't know how emotions flare in the market! There are so many variables that we see in front of us that computers can't account for! (sound at all similar to the reasons people give for why computers can't assess a team?)

[Nova09]

Ok, Dane, I know your response was mostly in jest and I hope you know mine was, too.

But that brings us to an inherent conundrum those of us who support computer rankings must face (and fwiw, I support human polls, too): In order to have any utility, computer rankings must reveal to us something we don't already know. In order to be trusted, computer rankings must match our preconceived notions.

[Walkon79]

Replying to myself on this to get your toughts on the above hypothesis.

Montana State is favored by 1.5 at Southern Utah.

That seems low to me, as I'd expect MSU to be favored by atleast 3. Anyone with better knowledge of the Big Sky or these two teams have any input?

Seems low to me but I'm a homer!! But according to many we haven't played ANYBODY yet :)