Computer Rankings

[summitlion1]

I never believed any of the computer rankings were purely objective. Understandably, they do not want to release the criteria they use, and I understand that, as that is their secret sauce. Over the years, however, I have seen some strange rankings, that I just didn't understand.

With that being said, I noticed an interesting stat in the most recent Sagarin rankings. He has a rating for recent, which is:

The RECENT, is score-based and weights RECENT play more heavily than earlier games. Its effect will become
more pronounced the longer a season goes if a given team happens to have an upward or downward trend.

I figured that Penn State's Recent rating would be significantly higher than their Mean ranting and Michigan's would be much lower. WRONG

Based on a 100 point scale, Michigan's Recent rating is .09 lower than their mean, and Penn State's is .49 higher. Lose two of your last three games, and drop less than one tenth of one percent?!?!?!?!

http://www.sagarin.com/sports/cfsend.htm

Crazy!

 

[fairgambit]

I never believed any of the computer rankings were purely objective. Understandably, they do not want to release the criteria they use, and I understand that, as that is their secret sauce. Over the years, however, I have seen some strange rankings, that I just didn't understand.

With that being said, I noticed an interesting stat in the most recent Sagarin rankings. He has a rating for recent, which is:

The RECENT, is score-based and weights RECENT play more heavily than earlier games. Its effect will become
more pronounced the longer a season goes if a given team happens to have an upward or downward trend.

I figured that Penn State's Recent rating would be significantly higher than their Mean ranting and Michigan's would be much lower. WRONG

Based on a 100 point scale, Michigan's Recent rating is .09 lower than their mean, and Penn State's is .49 higher. Lose two of your last three games, and drop less than one tenth of one percent?!?!?!?!

http://www.sagarin.com/sports/cfsend.htm

Crazy!

I have slowly come to realize that college football rankings are all about agendas. If you want a certain team ranked higher, or in the playoffs, you can find a reason to make it happen. Integrity is gone, although maybe it was never there. The many times we were screwed in the past should have given me a clue, but I just didn't see the forest for the trees.

 

[AWS1022]

I have slowly come to realize that college football rankings are all about agendas. If you want a certain team ranked higher, or in the playoffs, you can find a reason to make it happen. Integrity is gone, although maybe it was never there. The many times we were screwed in the past should have given me a clue, but I just didn't see the forest for the trees.

And all these rankings are skewed in some way by the eye test and preseason rankings. If a computer model was completely unbiased how would it have Western Michigan ranked 22nd with a 12-0 record? How would a computer know that their wins weren't as "good" as someone else's wins? Computers only know what someone tells it to know.

 

[Erial_Lion]

Based on Michigan's performance last week, they would have risen in most computer ratings (at least ones that take margin of victory into account). Losing on the road by 3 to the #2 ranked team isn't a bad thing when it comes to power ratings or computer ratings.

 

[JJTopp99]

I have slowly come to realize that college football rankings are all about agendas. If you want a certain team ranked higher, or in the playoffs, you can find a reason to make it happen. Integrity is gone, although maybe it was never there. The many times we were screwed in the past should have given me a clue, but I just didn't see the forest for the trees.

Integrity in college football rankings? LOL, it was never there and likely never will be.

 

[Nitt1300]

It's been all about opinions for as long as I can remember. From Tricky Dick to the present day, it's about politics as much as it is about what happens on the field.

 

[Cpeplion]

any poll that has a 7-4 LSU team that is 1-4 vs the top 30 @ #8 is complete horsecrap

 

[CJsE]

I never believed any of the computer rankings were purely objective. Understandably, they do not want to release the criteria they use, and I understand that, as that is their secret sauce. Over the years, however, I have seen some strange rankings, that I just didn't understand.

With that being said, I noticed an interesting stat in the most recent Sagarin rankings. He has a rating for recent, which is:

The RECENT, is score-based and weights RECENT play more heavily than earlier games. Its effect will become
more pronounced the longer a season goes if a given team happens to have an upward or downward trend.

I figured that Penn State's Recent rating would be significantly higher than their Mean ranting and Michigan's would be much lower. WRONG

Based on a 100 point scale, Michigan's Recent rating is .09 lower than their mean, and Penn State's is .49 higher. Lose two of your last three games, and drop less than one tenth of one percent?!?!?!?!

http://www.sagarin.com/sports/cfsend.htm

Crazy!

It also takes into account the teams played recently, their SoS and your performance agaianst them vs their recent opponents and their SoS.

Beating a team at home that lost 9 out of last 10 isn't going to bump PSU.

Beating Rutgers is, well, beating Rutgers. Not going to give a bump.

Indiana is a 45-65 ranked team, beating them is minimal bump.

His recent is also at 73% on the year which is pretty damn good.

 

[LafayetteBear]

His recent is also at 73% on the year which is pretty damn good.

Please translate into plain English.

 

[CJsE]

Please translate into plain English.

His recent ratings prediction system has correctly picked 73% correctly out of 700+ games this year. This places it among the top 10 major predictors on the year.

His other 2 are also in the top 10.

 

[McFate]

And all these rankings are skewed in some way by the eye test and preseason rankings. If a computer model was completely unbiased how would it have Western Michigan ranked 22nd with a 12-0 record? How would a computer know that their wins weren't as "good" as someone else's wins?

Sagarin's ratings have some preseason seeds early in the year, but those are eliminated once a sufficient number of games have been played to produce ratings from the current season alone. Sagarin says at this point his ratings are computed from this year's game results and nothing else; therefore they cannot possibly be "skewed by the eye test and preseason rankings."

How do you know that Western Michigan's 12-0 isn't as impressive as Alabama's? WMU's schedule is: 9 MAC teams of which only one lost less than four games, one FCS team, and two teams from the bottom half of the B1G's weaker division. It's difficult to peg how good they are because they didn't play anyone decent, which might mean big error bars on their #22 ranking. But that doesn't mean they're indistinguishable from the New England Patriots or PSU (both of which would also go 12-0 against that schedule).

 

[CJsE]

Most computer rankings that I have seen eliminate preseason information after week 6-7.

 

[summitlion1]

His recent ratings prediction system has correctly picked 73% correctly out of 700+ games this year. This places it among the top 10 major predictors on the year.

His other 2 are also in the top 10.

Are you saying that he has predicted the winner 73% of the time straight up? That doesn't sound very impressive to me. Random guessing will get you to 50%. Throw in the lopsided match up, and I'm pretty sure I could achieve 75%, and I know I'm not that smart.

 

[Mister Magoo]

In 1986, Sagarin had Oklahoma 1st after we beat Miami even though OU lost to Miami that year. Even with a loss to a team we beat while we were perfect. I gave up on him after that year.

 

[summitlion1]

Based on Michigan's performance last week, they would have risen in most computer ratings (at least ones that take margin of victory into account). Losing on the road by 3 to the #2 ranked team isn't a bad thing when it comes to power ratings or computer ratings.

That doesn't explain how their "Recent" rating only went down .09%. By Sagarin's own predictors, Michigan should have beaten Iowa by 15, but they lost by 1. They should have beaten Indiana by 28, but they only won by 10. They should have lost to OSU by 3, which they did. PLEASE explain how their "Recent" ranking did not change?

 

[bxle]

I question if the assumptions made in these programs pass the smell test.

 

[CJsE]

Are you saying that he has predicted the winner 73% of the time straight up? That doesn't sound very impressive to me. Random guessing will get you to 50%. Throw in the lopsided match up, and I'm pretty sure I could achieve 75%, and I know I'm not that smart.

It's a lot harder than you think year after year. He has the best percentage out of 5 of the former BCS computer rankings, but they are all pretty close to each other.

Also it almost always means you are better than 50%+ against the spread as well which is not easy to do consistently, even if you cherry pick games.

http://www.thepredictiontracker.com/ncaaresults.php

 

[Hotshoe]

In 1986, Sagarin had Oklahoma 1st after we beat Miami even though OU lost to Miami that year. Even with a loss to a team we beat while we were perfect. I gave up on him after that year.

That shouldn't be possible, but I remember that happening. Never liked his program, still dont.

 

[kgilbert78]

Integrity in college football? LOL, it was never there and likely never will be.

FYP

 

[AWS1022]

Sagarin's ratings have some preseason seeds early in the year, but those are eliminated once a sufficient number of games have been played to produce ratings from the current season alone. Sagarin says at this point his ratings are computed from this year's game results and nothing else; therefore they cannot possibly be "skewed by the eye test and preseason rankings."

How do you know that Western Michigan's 12-0 isn't as impressive as Alabama's? WMU's schedule is: 9 MAC teams of which only one lost less than four games, one FCS team, and two teams from the bottom half of the B1G's weaker division. It's difficult to peg how good they are because they didn't play anyone decent, which might mean big error bars on their #22 ranking. But that doesn't mean they're indistinguishable from the New England Patriots or PSU (both of which would also go 12-0 against that schedule).

How does the computer know that an 8-4 MAC team isn't better than a 7-5 SEC team? It's because it's told by a human which conference is better, which teams are better, etc. where did Sagarin have OSU ranked in 2014 when they beat no one...not one ranked team? I'll bet they were ranked higher than 22nd.

 

[McFate]

How does the computer know that an 8-4 MAC team isn't better than a 7-5 SEC team? It's because it's told by a human which conference is better, which teams are better, etc.

That's simply false. It is easy to come up with rankings systems that do not require any such assumption. I've linked one for you (up until recently, Sagarin used to publish ELO-based ratings as well, as those were used by the BCS).

How does the computer know that an 8-4 MAC team isn't better than a 7-5 SEC team? In brief: because the SEC as a group has much stronger performances against the rest of college football than the MAC as a group.

Let's consider the only 8-4 MAC team, Ohio U., who happens to be a MAC division champ. Their record in conference suggests they're one of the top few MAC teams. What did they do in non-conference play that would lead one to assign strength to the MAC as a whole? They lost to 2-9 Texas State. They lost to Tennessee. They beat a lower-division Gardner-Webb team. Their sole victory against their own divisional level was over 2-10 Kansas. There is not much respect flowing into the MAC via Ohio U.'s accomplishments.

Where did Sagarin have OSU ranked in 2014 when they beat no one...not one ranked team? I'll bet they were ranked higher than 22nd.

Ohio State was 3-0 vs Sagarin's top ten and 4-0 vs Sagarin's top 30 in 2014. Not sure where you get "not one ranked team." If you mean his rankings as of before the B1G title game, OSU was #9 (ELO) to #15 (PREDICTOR) in his various rankings. Even then, Ohio State had a win (Michigan State) against a top-ten team. Are you just making things up?

 

[irncty]

Computers do what they are told to do...they're dumb. If you update the 'secret sauce' algorithm to add a few points to some 'preferred' schools, nobody will know because there's no transparency or accountability.

 

[summitlion1]

Computers do what they are told to do...they're dumb. If you update the 'secret sauce' algorithm to add a few points to some 'preferred' schools, nobody will know because there's no transparency or accountability.

Correct! My point exactly.

 

[CJsE]

What Fate said. At the beginning of the season they usually use rankings based on previous season interconference record, but those switch part-way through the year based on interconference performance and individual performance.

Without delving too deeply into the schedules the MAC has a 47.9% ooc win rate out of 48 games, the PAC-12 has a 72.2% ooc win rate out of 36 games, the Sun Belt has a 40.9% ooc win rate out of 44 games and the Big Ten has a 76.2% ooc win rate out of 42 games. You can clearly see a difference in conference quality. You look at the interconference records of the teams, then you look at the intraconference performance of the teams involved and you get a pretty clear power index.

 

[summitlion1]

What Fate said. At the beginning of the season they usually use rankings based on previous season interconference record, but those switch part-way through the year based on interconference performance and individual performance.

Without delving too deeply into the schedules the MAC has a 47.9% ooc win rate out of 48 games, the PAC-12 has a 72.2% ooc win rate out of 36 games, the Sun Belt has a 40.9% ooc win rate out of 44 games and the Big Ten has a 76.2% ooc win rate out of 42 games. You can clearly see a difference in conference quality. You look at the interconference records of the teams, then you look at the intraconference performance of the teams involved and you get a pretty clear power index.

With all of the stats you keep providing, you still haven't explained how Michigan lost 2 of 3 games and only dropped .09% on "recent" rating. What would have happened if they won all three games? They might be ranked #1 now.

 

[CJsE]

With all of the stats you keep providing, you still haven't explained how Michigan lost 2 of 3 games and only dropped .09% on "recent" rating. What would have happened if they won all three games? They might be ranked #1 now.

1st. I don't know because he doesn't release his full formula.
2nd. They didn't drop only 0.09% on "recent" rating. Their recent rating is 0.09% lower in relation to their mean rating. To see what they dropped, you'd have to look at last week's recent rating compared to this week.

As I understand it, both are calculated heavily from actually score vs predicted score. They essentially performed as the system expected them to perform, which is to lose by 3-4 points. Losing by 3 points on the road to higher ranked team. Home team gets 2.32 points, so Ohio State was most likely more than a 3 point favorite. Most computer rating systems aren't going to give large rewards for beating teams ranked outside the top 40 or give a large penalty for losing a close game to a top 25 team.

 

[AWS1022]

That's simply false. It is easy to come up with rankings systems that do not require any such assumption. I've linked one for you (up until recently, Sagarin used to publish ELO-based ratings as well, as those were used by the BCS).

How does the computer know that an 8-4 MAC team isn't better than a 7-5 SEC team? In brief: because the SEC as a group has much stronger performances against the rest of college football than the MAC as a group.

Let's consider the only 8-4 MAC team, Ohio U., who happens to be a MAC division champ. Their record in conference suggests they're one of the top few MAC teams. What did they do in non-conference play that would lead one to assign strength to the MAC as a whole? They lost to 2-9 Texas State. They lost to Tennessee. They beat a lower-division Gardner-Webb team. Their sole victory against their own divisional level was over 2-10 Kansas. There is not much respect flowing into the MAC via Ohio U.'s accomplishments.



Ohio State was 3-0 vs Sagarin's top ten and 4-0 vs Sagarin's top 30 in 2014. Not sure where you get "not one ranked team." If you mean his rankings as of before the B1G title game, OSU was #9 (ELO) to #15 (PREDICTOR) in his various rankings. Even then, Ohio State had a win (Michigan State) against a top-ten team. Are you just making things up?

Of course I meant before the Big title game...and I did forget MSU. So they played one ranked team and the rest of their schedule was dreadful....so where did Sagarin have then ranked going into the championship game because their schedule wasn't much better than Western Michigan's this year and they weren't undefeated. I'll bet it wasn't 22nd.

 

[AWS1022]

1st. I don't know because he doesn't release his full formula.
2nd. They didn't drop only 0.09% on "recent" rating. Their recent rating is 0.09% lower in relation to their mean rating. To see what they dropped, you'd have to look at last week's recent rating compared to this week.

As I understand it, both are calculated heavily from actually score vs predicted score. They essentially performed as the system expected them to perform, which is to lose by 3-4 points. Losing by 3 points on the road to higher ranked team. Home team gets 2.32 points, so Ohio State was most likely more than a 3 point favorite. Most computer rating systems aren't going to give large rewards for beating teams ranked outside the top 40 or give a large penalty for losing a close game to a top 25 team.

So LSU #8, Louisville #10 and 6-5 TCU being ranked four spots higher than 10-2 Boise State and five spots higher than 9-3 Nebraska makes sense to you? TCU is 6-5 in the worst conference in college football.

 

[john4psu]

Definitely agree on the comments of a four-loss LSU at #8 is mind boggling.

I've closely followed college football and the cfb polls obviously since the '68 season. To lose two of your final three regular season games (one to an unranked team) and look unimpressive in winning against an inferior team and yet be ranked #5 in the country, #3 in Sagarin has to be unprecedented - because it's wrong.

 

[CJsE]

Definitely agree on the comments of a four-loss LSU at #8 is mind boggling.

I've closely followed college football and the cfb polls obviously since the '68 season. To lose two of your final three regular season games (one to an unranked team) and look unimpressive in winning against an inferior team and yet be ranked #5 in the country, #3 in Sagarin has to be unprecedented - because it's wrong.

You've been closely following college football since '68, but he's been getting paid and called a expert for 30 years and was one of the 6 predictors for the BCS. I never said his rankings are perfect, but it's ridiculous for people to act like he doesn't know what he's talking about.

Personally, I like the Massey Composite for rankings because it takes averages more than 50 predictors to get its results. If you look at the composite, the median and mean of over 65 different computer predictors place Michigan at #4 so there's obviously something to the argument. That will change after the championships are played.

 

[AWS1022]

You've been closely following college football since '68, but he's been getting paid and called a expert for 30 years and was one of the 6 predictors for the BCS. I never said his rankings are perfect, but it's ridiculous for people to act like he doesn't know what he's talking about.

Personally, I like the Massey Composite for rankings because it takes averages more than 50 predictors to get its results. If you look at the composite, the median and mean of over 65 different computer predictors place Michigan at #4 so there's obviously something to the argument. That will change after the championships are played.

Certain things computers can't take into account are a team having a difference maker, heart and team cohesion. A good team in a bad conference gets penalized (Boise State a few years ago is a good example) and an average team in a good conference gets inflated (LSU this year). There is no perfect system that's why they need to come up with a way to decide it on the field. The politics have to be taken out of it or just worry about each conference championship and not even have a national champion.

 

[Psychedelic_MackDaddy]

I never believed any of the computer rankings were purely objective. Understandably, they do not want to release the criteria they use, and I understand that, as that is their secret sauce. Over the years, however, I have seen some strange rankings, that I just didn't understand.

With that being said, I noticed an interesting stat in the most recent Sagarin rankings. He has a rating for recent, which is:

The RECENT, is score-based and weights RECENT play more heavily than earlier games. Its effect will become
more pronounced the longer a season goes if a given team happens to have an upward or downward trend.

I figured that Penn State's Recent rating would be significantly higher than their Mean ranting and Michigan's would be much lower. WRONG

Based on a 100 point scale, Michigan's Recent rating is .09 lower than their mean, and Penn State's is .49 higher. Lose two of your last three games, and drop less than one tenth of one percent?!?!?!?!

http://www.sagarin.com/sports/cfsend.htm

Crazy!

1. Apple
2. Microsoft
3. IBM

 

[Marylovesthelions]

I have slowly come to realize that college football rankings are all about agendas. If you want a certain team ranked higher, or in the playoffs, you can find a reason to make it happen. Integrity is gone, although maybe it was never there. The many times we were screwed in the past should have given me a clue, but I just didn't see the forest for the trees.

It is all bullspit. We smoke Iowa and they best Michigan the next game. Michigan loses another and still gets number four. There is no logic to the committee's thinking. If you lose two games in the last month, you should drop out of the top ten. We have played better football over the last six games that UM or OSU. Just beat Wisconsin.

 

[McFate]

Of course I meant before the Big title game...and I did forget MSU. So they played one ranked team and the rest of their schedule was dreadful....so where did Sagarin have then ranked going into the championship game because their schedule wasn't much better than Western Michigan's this year and they weren't undefeated. I'll bet it wasn't 22nd.

That question was already answered: #9 ~ #15 depending on which ranking you chose. I prefer PREDICTOR because it's the best estimate of team strength, but the ranking in USA Today would have been #13.

"Schedule wasn't much better" is simply false. WMU 2016 has played 12 games, of which only one was against a top-50 team (#41 Northwestern), and only three are even in the top 100. Ohio State 2014 played five top-50 teams and only one out of the top 100, using ratings as of the week prior to the conference championship games in both cases. If you sorted the two teams' opponents by rank, you'd hit 2014 Ohio State's next-to-worst opponent (#88 Indiana) before you'd get to 2016 WMU's third-best (#96 Illinois).

 

[psugisher]

I think the computer make a better argument for an expanded playoff. Teams can be rewarded for losses against good teams. Doesn't mean a team isn't good but to deserve a spot in a playoff, you have to win games, win your division, etc.. doesn't matter what Michigan's computer ranking is, they were 3rd place in their division. Expand the playoffs to 12 or 16 and include them. I want to see how Western Michigan matches up with the top teams in the country (not against PSU though). If teams outside the 5 power conferences can't get in the playoff, then why the heck are they on the schedule? Why do any of those games count? I'm pretty sure Western Michigan would give most top ten teams all they could handle and probably beat a few of them. I'd love to see a Western Michigan vs Michigan Cotton Bowl.. Anyway, computers have their place.. anything to help take away the human bias that gets in the way. The team name on the jersey does matter. It's why Notre Dame has been overrated year after year..

 

[AWS1022]

That question was already answered: #9 ~ #15 depending on which ranking you chose. I prefer PREDICTOR because it's the best estimate of team strength, but the ranking in USA Today would have been #13.

"Schedule wasn't much better" is simply false. WMU 2016 has played 12 games, of which only one was against a top-50 team (#41 Northwestern), and only three are even in the top 100. Ohio State 2014 played five top-50 teams and only one out of the top 100, using ratings as of the week prior to the conference championship games in both cases. If you sorted the two teams' opponents by rank, you'd hit 2014 Ohio State's next-to-worst opponent (#88 Indiana) before you'd get to 2016 WMU's third-best (#96 Illinois).

Top 50 based on what? The same computer system or human input into that computer system. Until you can convince me or anyone else that #65 can't beat #50, it's still just opinion. All these arguments can be skewed to fit whichever side someone is trying to justify. I guarantee if the roles of PSU and OSU were reversed, we would be hearing the exact same arguments for OSU that are being used for PSU. And all of it takes away from proving it on the field.

 

[CJsE]

Top 50 based on what? The same computer system or human input into that computer system. Until you can convince me or anyone else that #65 can't beat #50, it's still just opinion. All these arguments can be skewed to fit whichever side someone is trying to justify. I guarantee if the roles of PSU and OSU were reversed, we would be hearing the exact same arguments for OSU that are being used for PSU. And all of it takes away from proving it on the field.

Here is the closest you are going to get. This analyzed 10 years of the Massey composite and the games between the ranked teams.

Highlights
- Teams ranked 1-5 beat teams ranked 6-15 77% of the time and 16-30 90.5% of the time.
- Teams ranked 1-5 only lost to teams ranked outside the top 31 teams 6 times out of 397 games (3 of those losses to teams ranked 31-50) for a 98.5% win rate.
- Teams ranked 1-5 never lost to a team ranked 91-128.
- Teams 6-15 beat teams ranked 16-30 76.3% of the time and teams ranked 31-50 87% of the time.
- Teams ranked 31-50 beat teams ranked 1-5 only 2.5% of the time, 16-30 27% of the time, teams 51-70 71% of the time and teams 71-90 86% of the time.

Edit: forgot the link http://www.bcftoys.com/massey

 

[McFate]

Top 50 based on what? The same computer system or human input into that computer system. Until you can convince me or anyone else that #65 can't beat #50, it's still just opinion.

If you don't think it's accurate to say that WMU's only non-awful opponents are Northwestern, Toledo, and Illinois (and we're being somewhat generous in not counting 3-9 Illinois as "awful")... perhaps you can tell me which other solid opponents they've played. In your view, which of 2016 WMU's opponents are on par with 2014 OSU's opponents Michigan State or Virginia Tech or even Minnesota (8-5)?

All these arguments can be skewed to fit whichever side someone is trying to justify. I guarantee if the roles of PSU and OSU were reversed, we would be hearing the exact same arguments for OSU that are being used for PSU. And all of it takes away from proving it on the field.

I've not said one word about the 2016 playoffs or the relative quality of any of the playoff contenders. I only participated to correct outright misinformation ("Sagarin's ratings rely on eye test," "there's no data from the results on the field to suggest the MAC is weaker than the SEC so it must be the programmer's bias," "Ohio State didn't play anyone ranked in 2014," "WMU's schedule in 2016 is comparable to OSU 2014").

 

[AWS1022]

If you don't think it's accurate to say that WMU's only non-awful opponents are Northwestern, Toledo, and Illinois (and we're being somewhat generous in not counting 3-9 Illinois as "awful")... perhaps you can tell me which other solid opponents they've played. In your view, which of 2016 WMU's opponents are on par with 2014 OSU's opponents Michigan State or Virginia Tech or even Minnesota (8-5)?



I've not said one word about the 2016 playoffs or the relative quality of any of the playoff contenders. I only participated to correct outright misinformation ("Sagarin's ratings rely on eye test," "there's no data from the results on the field to suggest the MAC is weaker than the SEC so it must be the programmer's bias," "Ohio State didn't play anyone ranked in 2014," "WMU's schedule in 2016 is comparable to OSU 2014").

You're obviously Sagarin's brother or cousin or something. Any system that has LSU #8 and Louisville #10 is seriously flawed. Even the human polls aren't that far off (and never are).

 

[McFate]

You're obviously Sagarin's brother or cousin or something.

Ad hominem.

You've been adamant about a couple of other folks in this thread answering certain questions of yours. Why do you refuse to provide the courtesy that you expect of others? Again:

If you don't think it's accurate to say that WMU's only non-awful opponents are Northwestern, Toledo, and Illinois (and we're being somewhat generous in not counting 3-9 Illinois as "awful")... perhaps you can tell me which other solid opponents they've played. In your view, which of 2016 WMU's opponents are on par with 2014 OSU's opponents Michigan State or Virginia Tech or even Minnesota (8-5)?