SportsHoopla Top 25 Week 6 discussion

[Deep Creek]

Well I didn't say it was ranked worse in my rankings(I don't rank either), I said they were ranked less in the ESPN FPI.

ESPN Football Power Index - 2015 - ESPN

MTSU: 59
Minn: 69

So, the ESPN FPI is the holy grail?

 

[uncfan103]

Yes you did. You called them "Trash" and then later called them "Bye" weeks.

For arguments sake. Not when I put my poll together. But, when I'm comparing top 25 teams, I'd rather look at the games against teams in which the outcome wasn't decided before the coin toss. When I'm comparing schedules, or when looking at SoS, I don't believe that winning at home against North Texas or Army should be more beneficial to your SoS than winning on the road against Lamar. I think the games against opponents that begin the game with a legitimate chance (teams that can complete a pass, compete against other winning football programs) are the ones that be given the most weight when it comes to SOS. It's ludicrous that a game against Rice is more beneficial than a win against Elon. Or it's silly to say that a win against Kansas is better than a win against North Dakota State. A team with no chance whatsoever shouldn't boost your strength of schedule. I'm much more concerned with the top teams that people have played. Or the number of games against teams that could possibly compete for a conference title if they were in a group of five conference.

 

[4down20]

For one your projection ratings have Stanford higher than Georgia

two

USC West>>>>>>>>>>>>USC east
Arizona>>> Vanderbilt
Oregon State >> UL-Monroe

UCF is really bad but they can't possibly be worse than Southern.

So yes, you live in a black and white world.

 

[4down20]

So, the ESPN FPI is the holy grail?

Fuck no.

It's a pretty decent tool if you understand what it is however. As are a few other computer stats programs out there.

Computers aren't perfect, but they are able to look at the data from all the games and treat it all equally and without bias.

ESPN FPI is actually not even in the top of what I would consider good, as unlike others it adds human input(not sure for how long), but the website is convenient.

I'll take FPI over what random people think, and often look towards it and others like it to see if there are areas I should look at for my own opinions.

 

[BoiseStateFan27]

It's a rating of team strength, not a season ranking.

Also, it's not dumb enough to be binary and actually measures what happens in games beyond "well you lost, lets make it all equal with everyone else that has a loss".

Well usually the other team scoring more points than you in a game is a pretty important thing that happened during a game now isn't it?

Arkansas was flat out dominated by Texas Tech and failed in the redzone against Toledo, good teams can't fail that badly in the redzone that is a key important factor that must be overlooked by this apparently.

 

[Deep Creek]

Alabama stops the run well and Georgia doesn't have good QB play.

Which is exactly why I pick Bama to beat LSU.

 

[rmilia1]

So, when you finish looking at the comparison in the schedule do you really think to yourself...

"ULM is just soooooooo much better than Ball State it makes up for Stanford being a little better than Georgia, Duke being just a little better than Wisconsin, and Minnesota and MTSU being just about even. They're soooooo much better than Ball State that it offsets not only the difference in the other two games, but also the loss that Alabama had this year"?

To be fair Bamas schedule IS better because of Ole Miss. However Bama lost that game so I don't know you can equate that to Bama being better since they didn't win.

 

[uncfan103]

And that right there is why you'll never get it. You actually believe that bullshit, and yet you have absolutely not the first clue in how computer formulas work.

Computers actually do what you claim you are doing but aren't.

1. I have heard you say multiple times that you're looking at future schedules and projecting when you put your polls together. You have said things like teams are that high because they don't have anymore losable games on their schedule, or that they're as low as you have them because they're not going to win their next three games.

2. I'm not trying to be a computer. And don't need to "get" anything. If a computer would tell me that Alabama or Stanford is more deserving of a higher seed than Northwestern, than I don't really mind being different than the computer.

3. I'm not a fan of letting statistics override actual outcomes. Games have to mean something.

You're computer could be spot on, and what you have right now could be perfect at the end of the season but that's not what i'm going for. If the computers were so accurate, we'd stop having rankings after like the 7th week because the computers would have enough data to finish out the rest of the season for us. OR, if computers were so accurate we wouldn't need a new poll every week.

 

[4down20]

For arguments sake. Not when I put my poll together. But, when I'm comparing top 25 teams, I'd rather look at the games against teams in which the outcome wasn't decided before the coin toss. When I'm comparing schedules, or when looking at SoS, I don't believe that winning at home against North Texas or Army should be more beneficial to your SoS than winning on the road against Lamar. I think the games against opponents that begin the game with a legitimate chance (teams that can complete a pass, compete against other winning football programs) are the ones that be given the most weight when it comes to SOS. It's ludicrous that a game against Rice is more beneficial than a win against Elon. Or it's silly to say that a win against Kansas is better than a win against North Dakota State. A team with no chance whatsoever shouldn't boost your strength of schedule. I'm much more concerned with the top teams that people have played. Or the number of games against teams that could possibly compete for a conference title if they were in a group of five conference.

You need more paragraphs, that was hard to read.

There is no way in hell you are able to look at all the teams and accurately gauge their strength. That is the reason I use things like advanced stats and FPI. They are able to.

 

[uncfan103]

To be fair Bamas schedule IS better because of Ole Miss. However Bama lost that game so I don't know you can equate that to Bama being better since they didn't win.

I was comparing Bama to Northwestern

 

[4down20]

To be fair Bamas schedule IS better because of Ole Miss. However Bama lost that game so I don't know you can equate that to Bama being better since they didn't win.

You only count it for SoS rankings, not season rankings.

When I did my computer rankings, the power score of a team was between 0-100. The SoS rankings were just a ranking of the average of all. But when it came to actually scoring the season rankings, you either got the positive points for a win, aka if you beat a team with a power score of 80, you got 80 points. And if you lost to a team then I would subtract the 80 from 100 and you lose that many points, so -20.

It would only matter when it comes to say 2 teams with the same amount of losses. As 1 team would have say -40 for a worse loss and the other team -20 for their loss to the better team.

However. As it is still early in the season it does matter some in terms of how much a team has been tested and how much we really know about them. When a team is playing a bunch of cupcakes that many teams could be undefeated on, they are untested and it's data is not going to be an accurate rating for that team.

 

[uncfan103]

You need more paragraphs, that was hard to read.

There is no way in hell you are able to look at all the teams and accurately gauge their strength. That is the reason I use things like advanced stats and FPI. They are able to.

There is no way to accurately compare all 128 teams + FCS teams. None. If you believe that stats can do it, that's fine. But, your stats tell you that Northwestern isn't better than stanford and the outcome showed otherwise. I'm sure the computers are smarter than I am, but they get things wrong sometimes.

I don't have a problem using statistics, but I also think the games are what's most important and I'm not going to question a head to head outcome because of some statistics until there are perfomances on the field that make me question those games. Right now, Northwestern has done nothing to show they'd lose to Stanford if they played again.

 

[rmilia1]

I was comparing Bama to Northwestern

Right. You were only discussing the wins but 4D was talking about SoS. I was just pointing out Bamas SoS is better but SoS doesn't factor in if you won or lost. Actually most teams with superior SoS have bad records because by losing it makes your opposition higher ranked

 

[Deep Creek]

To be fair Bamas schedule IS better because of Ole Miss. However Bama lost that game so I don't know you can equate that to Bama being better since they didn't win.

We don't know if Ole Miss is any good or not yet. They were rated like this last year and TCU beat the ever loving shit out of them at the end of the year. They could end up real good or end up like last year. Quien sabe?

 

[uncfan103]

You only count it for SoS rankings, not season rankings.

When I did my computer rankings, the power score of a team was between 0-100. The SoS rankings were just a ranking of the average of all. But when it came to actually scoring the season rankings, you either got the positive points for a win, aka if you beat a team with a power score of 80, you got 80 points. And if you lost to a team then I would subtract the 80 from 100 and you lose that many points, so -20.

It would only matter when it comes to say 2 teams with the same amount of losses. As 1 team would have say -40 for a worse loss and the other team -20 for their loss to the better team.

However. As it is still early in the season it does matter some in terms of how much a team has been tested and how much we really know about them. When a team is playing a bunch of cupcakes that many teams could be undefeated on, they are untested and it's data is not going to be an accurate rating for that team.

This sounds like a really, really good idea. I'm not trying to be sarcastic or anything. I mean it, I like the concept and I like that winning is important and beating crap teams is not. It also looks like it does a better job of spreading out values.

 

[4down20]

1. I have heard you say multiple times that you're looking at future schedules and projecting when you put your polls together. You have said things like teams are that high because they don't have anymore losable games on their schedule, or that they're as low as you have them because they're not going to win their next three games.

2. I'm not trying to be a computer. And don't need to "get" anything. If a computer would tell me that Alabama or Stanford is more deserving of a higher seed than Northwestern, than I don't really mind being different than the computer.

3. I'm not a fan of letting statistics override actual outcomes. Games have to mean something.

You're computer could be spot on, and what you have right now could be perfect at the end of the season but that's not what i'm going for. If the computers were so accurate, we'd stop having rankings after like the 7th week because the computers would have enough data to finish out the rest of the season for us. OR, if computers were so accurate we wouldn't need a new poll every week.

1. I do, the power rankings I use do not. They are not "guesses" that have a 50/50 chance of being right. They are estimates based on data that have a high chance of being right.

It's the difference in knowing if you should play that 7/2 offsuit hand or pocket Aces in hold em. Sure, it may ultimately be a "guess" on if you win or not, but it's not a random guess.

2. Exactly, you are not a computer. Yet you keep trying to claim what you do is what the computer does and you don't. You are not watching that many games, you are not looking at things with an unbiased point of view either. The fact is, you watch probably less than 15% of the football - if that, and your ability to even hold the stats of all 128 FBS teams in memory is even much lower.

3. Games do mean something. They get played every week and then new data comes out and everyone updates their opinions. Been that way my entire life, going to be that way next year too. But of course, you need that strawman for a reason.

 

[rmilia1]

We don't know if Ole Miss is any good or not yet. They were rated like this last year and TCU beat the ever loving shit out of them at the end of the year. They could end up real good or end up like last year. Quien sabe?

True but right now they help Bamas SoS. This is why you can't truly depend on that Stat though. Too many different SOS ratings and it doesn't help your SoS when you win but it does when you lose. It's weird

 

[4down20]

This sounds like a really, really good idea. I'm not trying to be sarcastic or anything. I mean it, I like the concept and I like that winning is important and beating crap teams is not. It also looks like it does a better job of spreading out values.

I can't take credit for it, the idea for that actually came from a Boise St fan friend of mine back in the early early days of my rankings.

I agree, it was great and worked extremely well.

 

[uncfan103]

1. I do, the power rankings I use do not. They are not "guesses" that have a 50/50 chance of being right. They are estimates based on data that have a high chance of being right.

It's the difference in knowing if you should play that 7/2 offsuit hand or pocket Aces in hold em. Sure, it may ultimately be a "guess" on if you win or not, but it's not a random guess.

2. Exactly, you are not a computer. Yet you keep trying to claim what you do is what the computer does and you don't. You are not watching that many games, you are not looking at things with an unbiased point of view either. The fact is, you watch probably less than 15% of the football - if that, and your ability to even hold the stats of all 128 FBS teams in memory is even much lower.

3. Games do mean something. They get played every week and then new data comes out and everyone updates their opinions. Been that way my entire life, going to be that way next year too. But of course, you need that strawman for a reason.

It may be an educated guess, but if you're just ranking teams based on what has occurred up to this point in the season it eliminates the need to guess, regardless of how educated your guess may be.


Games don't matter one bit if the only thing that comes from them are stats. That's not why games are played.
If computers use what has happened in the games to say that Northwestern deserves to be ranked below Stanford, we know something isn't right. I can defend NW being ahead of Stanford based on play on the field, I can't defend Stanford being ahead of NW in laymans terms.


I don't claim to be some computer, and I don't think my rankings are meant to be computerized, it's something I just tried doing for fun. I feel like I can defend my rankings when questioned about certain teams. I'd rather keep it simple and put teams in order based on who I think has the best wins.

 

[4down20]

It may be an educated guess, but if you're just ranking teams based on what has occurred up to this point in the season it eliminates the need to guess, regardless of how educated your guess may be.

Games don't matter one bit if the only thing that comes from them are stats. That's not why games are played.
If computers use what has happened in the games to say that Northwestern deserves to be ranked below Stanford, we know something isn't right. I can defend NW being ahead of Stanford based on play on the field, I can't defend Stanford being ahead of NW in laymans terms.

I don't claim to be some computer, and I don't think my rankings are meant to be computerized, it's something I just tried doing for fun. I feel like I can defend my rankings when questioned about certain teams. I'd rather keep it simple and put teams in order based on who I think has the best wins.

An educated guess is better than no guess at all.

Was it an educated guess that Alabama would beat ULM? Was it bad? Nope. What would be bad? Pretending that you didn't know what the outcome of the games would be and as if it was no good.

The amount of errors in those educated guesses are much smaller than the errors you are purposely creating by ignoring it. It is more ignorant to believe all teams are equal to start the season than it was to believe Auburn to be a top25 team. And what goes ignored in all the focus on teams like Auburn, Oregon and such who may have been over-ranked are all the teams that weren't that far off and were generally good. Which far outweighed the errors, and to which the methods being used by some created even bigger errors in.

Where the hell do you think the stats come from? They come from the games. The games are the only things that matter. Literally.