SportsHoopla Top 25 Week 6 discussion

[4down20]

In what ways is the VERN not objective?

In that it's extremely limited and ignores the majority of the data available.

" based only on wins per game of the team, of their opponents, and of the opponents' opponents."

It's basically the equivilant of the crappy SoS formula that the NCAA uses and everyone knows if flawed because it says all "4-1" teams are equal. With only a slight influence from the opponents record, which is in itself also limited.

And that's why Toledo ends up #7. Because it's completely unable to accurately weigh the true value of the teams. 90% of my formula was all towards that 1 goal. It is the most important factor of computer rankings for which all other things are based on.

 

[ericd7633]

Maybe? Right now it's not but by year's end you may be right. Right now it's just a neutral site win over a 3-2 team that has a quality win and a blowout loss

Which is better than anything Wisconsin or Iowa State has done. Who the hell have they beaten?

 

[occupant]

In what ways is the VERN not objective?

Ah so he's talking shit on the VERN?

It is the most objective, it isn't really a poll.

It simply measures a teams opponents' records and the records of their opponents' opponents in determining schedule strength.

Pure. No previous seasons to dirty it up, no preconceptions about who is better before games are even played.

It's a useful tool in helping to determine predictive ratings.

 

[rmilia1]

Which is better than anything Wisconsin or Iowa State has done. Who the hell have they beaten?

No one. I just value a road win more,than a neutral site win when there isn't a great distinction between teams. Thats all

 

[ericd7633]

No one. I just value a road win more,than a neutral site win when there isn't a great distinction between teams. Thats all

Fucking dumb.

 

[4down20]

Pure. No previous seasons to dirty it up, no preconceptions about who is better before games are even played.

It's a useful tool in helping to determine predictive ratings.

Those things are true of almost all computer rankings at this point, or close to being so. They only use the previous years data until it's no longer needed for valid results.

 

[dennis580]

I actually don't agree with that. I think the schedules are very similar but I gave the edge to iowa because they have 2 true road wins

The playoff committee definately takes road wins into consideration. Jeff Long talked multiple times about roads wins last year during interviews.

 

[rmilia1]

The playoff committee definately takes road wins into consideration. Jeff Long talked multiple times about roads wins last year during interviews.

Makes 100 percent sense. Road wins are just better wins

 

[rmilia1]

Fucking dumb.

Nope. Road wins are tougher to get thus better wins

 

[4down20]

Nope. Road wins are tougher to get thus better wins

Homefield advantage is 3.5 pts. It's like 3.500003430 over the past 30 years or something like that. So, 3.5 is good enough.

That's it.

Unless there was a difference in the score by more than 3.5pts for the home team, it's unlikely that homefield advantage mattered.

At best you have a 7 pt swing in terms of which team is home. As you either get a 3.5 pt bonus, or you get a -3.5 pt penalty. But when it comes to rewarding teams, you either give them 3.5, or take 3.5 for the advantage.

 

[nddulac]

Homefield advantage is 3.5 pts. It's like 3.500003430 over the past 30 years or something like that. So, 3.5 is good enough.

That's it.

Unless there was a difference in the score by more than 3.5pts for the home team, it's unlikely that homefield advantage mattered.

At best you have a 7 pt swing in terms of which team is home. As you either get a 3.5 pt bonus, or you get a -3.5 pt penalty. But when it comes to rewarding teams, you either give them 3.5, or take 3.5 for the advantage.

Given the way that distributions work, this argument is complete bullshit.

Home field advantage may average 3.5 points, but the variance is huge.

Fail.

 

[nddulac]

In that it's extremely limited and ignores the majority of the data available.

" based only on wins per game of the team, of their opponents, and of the opponents' opponents."

It's basically the equivilant of the crappy SoS formula that the NCAA uses and everyone knows if flawed because it says all "4-1" teams are equal. With only a slight influence from the opponents record, which is in itself also limited.

Ah so he's talking shit on the VERN?

It is the most objective, it isn't really a poll.

I haven't heard of VERN before, but I am sure that like any system it has its limitations.

But production of results with which you disagree does not mean it is not objective. But then again, any computerized system has its subjective aspects. For example, how the system computes how much better one team is than another is always a subjective choice - irrespective of the system. The advantage of a computer system is that it can apply that subjective criteria set over the entire data set uniformly.

So while consistency does not imply objectivity. But neither does producing results with which one disagrees imply subjectivity.

 

[4down20]

Given the way that distributions work, this argument is complete bullshit.

Home field advantage may average 3.5 points, but the variance is huge.

Fail.

This guy tried to measure teams individually, and the results were rather shocking.

College Football by the Numbers: The Myth of Home Field Advantage

But yeah, I guess Hawaii doing about 19 pts better at home than on the road could be seen as a big difference. Yet, somehow Penn St and the majority of good teams are actually lower. Penn St came in with only a .2 homefield advantage.

And since he isn't using specific ratings, 3.5 is actually a good general rule and is widely used.

 

[4down20]

I haven't heard of VERN before, but I am sure that like any system it has its limitations.

But production of results with which you disagree does not mean it is not objective. But then again, any computerized system has its subjective aspects. For example, how the system computes how much better one team is than another is always a subjective choice - irrespective of the system. The advantage of a computer system is that it can apply that subjective criteria set over the entire data set uniformly.

So while consistency does not imply objectivity. But neither does producing results with which one disagrees imply subjectivity.

So if I were to rank teams based solely on the number of yards they get passing in their games, would you call it objective?

I'd say while you can call it objective for using the data(it is the same of all computers), I'd say the use of it is way more subjective than objective to the point where it's not a good use of the statistics for ranking purposes.

If you wish to narrow it down to only the small part that is objective, then that's fine. As part of the composite it's fine and all because the teams are somewhat in the general area they should be mostly, but as an individual ranking system it sucks.

Kind of like the NCAA SoS rankings that use the same principle. They are ok for a ball park figure usually, but there are huge faults when it comes to some teams and it sucks for specific rankings.

But yes, they all have their shortcomings. The best we can do is understand the limitations and try to account for them in our head. I was always disappointed with mine not taking out garbage time and the corruption of the model that causes. It would have been better and more accurate doing that.

I think yours is faulted for only using the scores. There is so much more to a game than the score when it comes to team strength and measuring how different schools performed. Rushing for 100 yards against Alabama is a damn bit better than Rushing for 100 against Oregon. A team that rushes for 100 yards against Alabama should be accurately seen as having a stronger rushing attack than a team that rushed for 100 yards against Oregon. And to do that requires looking at a ton of data and adjusting for the opponents strength. And how the team scored matters as well etc. All these things make for more accurate rankings.

 

[4down20]

Good read about homefield advantage even though it centers around the NFL. It's not what you think that gives the homefield advantage.

Home field advantage is real, but it’s not because of the 12th man

 

[nddulac]

But yeah, I guess Hawaii doing about 19 pts better at home than on the road could be seen as a big difference. Yet, somehow Penn St and the majority of good teams are actually lower. Penn St came in with only a .2 homefield advantage.

Like I said - the variance is large.

And since he isn't using specific ratings, 3.5 is actually a good general rule and is widely used.

But to assign that as a constant value in all cases would be equivalent to me treating all students as C students, since that is the average grade I assign.

That not withstanding, I remember reading something years ago (perhaps from my rec.sport.football.college days) that the most important aspect of home-field advantage comes when playing teams from other time zones. The bigger the timezone difference (or deviation from normal kick off time, meal time, etc.) the larger the advantage. That is consistent with Hawaii enjoying one of the largest home field advantages.

I have no idea where to even begin to find that analysis (if it even exists out there in the cyber ether) but it is a concept that I have never forgotten.

 

[nddulac]

So if I were to rank teams based solely on the number of yards they get passing in their games, would you call it objective?

I believe I already answered that question when I said that the choice of criteria is always subjective.

I think yours is faulted for only using the scores. There is so much more to a game than the score when it comes to team strength and measuring how different schools performed.

I don't disagree, mostly because I recognize (and fully admit) that no fixed set of criteria will be perfect or ideal. What makes one team good is generally different than what makes another team good. And how you weigh those qualities in terms of ranking those will most likely be different. As such, no system is without flaws. That's why it is good to look at a wide variety of systems and methods and then look for the patterns that emerge.

However, in defense of my system, given that I am trying to rate 759 teams this year, meaning that there are some 300+ games each week and no centralized data source for anything other than scores, even given the limitations of using the scores and game location alone is all that is practical.

 

[4down20]

Like I said - the variance is large.

But to assign that as a constant value in all cases would be equivalent to me treating all students as C students, since that is the average grade I assign.

That not withstanding, I remember reading something years ago (perhaps from my rec.sport.football.college days) that the most important aspect of home-field advantage comes when playing teams from other time zones. The bigger the timezone difference (or deviation from normal kick off time, meal time, etc.) the larger the advantage. That is consistent with Hawaii enjoying one of the largest home field advantages.

I have no idea where to even begin to find that analysis (if it even exists out there in the cyber ether) but it is a concept that I have never forgotten.

But those are not the reasons why Hawaii has such a big "advantage". The variance was more to do with them being a bad team in general than actual home field advantage. That's why the better teams always came in with lower numbers. Now that could be useful in predicting scores, if you applied each one specifically, but those differences don't actually help the teams win more than any other team.

So while the data showed that large point difference for Hawaii, it didn't show a proportionate amount of extra home wins.

The NFL article I posted is a good read about it. They attribute it more to teams being unfamiliar with their surroundings than all other factors. It notes that division teams that play twice a year have less of a home field advantage. This also goes well with another article I read that only examined conference games found very little home field advantage on average.

There is really only 1 time I can think of where I thought the crowd was really affecting the game recently, and that was when Alabama played Auburn @ Auburn in 2009. Their student section was so loud it was shaking the scoreboard and you could see it on the TV as Alabama tried to score near it.

 

[4down20]

I believe I already answered that question when I said that the choice of criteria is always subjective.


I don't disagree, mostly because I recognize (and fully admit) that no fixed set of criteria will be perfect or ideal. What makes one team good is generally different than what makes another team good. And how you weigh those qualities in terms of ranking those will most likely be different. As such, no system is without flaws. That's why it is good to look at a wide variety of systems and methods and then look for the patterns that emerge.

However, in defense of my system, given that I am trying to rate 759 teams this year, meaning that there are some 300+ games each week and no centralized data source for anything other than scores, even given the limitations of using the scores and game location alone is all that is practical.

Just because they all have flaws doesn't mean some aren't better or more/less flawed than others.

I can sympathize with getting data, it's the reason I quit doing it.

 

[The Crimson King]

This year. Back in 2010 we had 6 teams take byes the week before, which was more than 50% of the teams(11 teams).

They were supposed to do something about it and did for like 1 year, but then this year they opened it back up again. Can't blame teams for doing it, it's smart. Can blame the SEC office for giving teams so many open dates before Alabama, while giving Tennessee a total of 1 team with an open date before them.

Speaking of Tennessee, they are supposed to be the game that's played on the "3rd Saturday of October" every year, but not this year. This year they are taking a bye week while we play A&M.

I do not count LSU or Auburn among the teams that would go in that club. Alabama has a bye week before LSU and it has been that way for like 5 years or longer with both teams taking a bye. And it's the same deal with Auburn and the cupcake, although Alabama played a cupcake first while Auburn was playing Georgia that week in the past.

So, MTSU before Ole Miss, ULM before Georgia, Bye before LSU, and Charleston Southern before Auburn? Nah, no one takes a week off before big games right?