Is the Fourpeat bad for I-AA?

[Bison Fan in NW MN]

It's not NDSU's fault that some of their fans are so easily butthurt and insecure after 4 straight.

No butthurt.

I probably would be butthurt if the team I cheered for lost in the quarters of semis at home the last 3 years...

[kalm]

It's not NDSU's fault that some of their fans are so easily butthurt and insecure after 4 straight.

No butthurt.

I probably would be butthurt if the team I cheered for lost in the quarters of semis at home the last 3 years...

Thanks for proving my point. Do you need some quality salve? I know where to get some.

[89Hen]

I'm talking about a year long contest John. One out of 30 can get lucky every week, but over 12 weeks, that luck runs out. I want to see how your computer does against a bunch of people over the whole year. You seem to think the computer will finish first. Put up or shut up.

No, it doesn't work like that. When you compare one model system to human capability you give "human capability" a huge advantage due to random chance when you include a bunch of people then pick the one who did best after the fact. It doesn't mater if you do it for one week or do it for an entire season. To make a fair comparison you need to pick one person ahead of time who you think is good at it and compare what that one person does to what the model does.

Then you need to analyze the results to see if the difference that you end up could be accounted for by chance.

To have a fair test pick one person ahead of time based on your past "contest" results who you think is good at picking games.

Wrong Mr Science. One week, somebody is going to get lucky. Over the course of the whole year, the same person can't keep getting lucky. You are so full of shit. Put up or shut up.

[YoUDeeMan]

A few points:

JSO:

UNH had a 5 game win streak versus FBS teams: The Wildcats strung together consecutive victories over Rutgers (2004), Northwestern ('06), Marshall ('07), Army ('08) and Ball State ('09) before their unprecedented five-game winning streak vs FBS teams was snapped at Pittsburgh, 38-16, on Sept. 2, 2010.

http://www.unhwildcats.com/sports/fball ... 0904oym3o8" onclick="window.open(this.href);return false;

NDSU had three close games...but a 4th game, SHSU, was only 7-3 at half.

UNH would have beaten Montana...and SDSU. Since I said it, it must be true.

[Bison Fan in NW MN]

No butthurt.

I probably would be butthurt if the team I cheered for lost in the quarters of semis at home the last 3 years...

Thanks for proving my point. Do you need some quality salve? I know where to get some.

Did you think that up by yourself?



So, the Eagles could go from preseason Big Sky contender to mediocre over the weekend....

[Gil Dobie]

UNH would have beaten Montana...and SDSU. Since I said it, it must be true.

That's a tough call. Illinois St beat SDSU handily at home, when SDSU's QB was out with injury. Illinois St beat UNH on the road. The Jacks were a much better team with Sumner and the score would have been closer,and the Jacks may have split with UNH on a home and home basis.

[D1B]

I would favor SDSU...

It's fun to play that game, but that's all it is. That said, SDSU's #4 in Massey wasn't what was really egregious to begin with. It was a 5-7 WIU at 14, a 6-6 SIU at 15 and a 4-8 MSU at 20.

You don't understand Massey. Keep trying. When you figure it out, IF you figure it out, educate Z.

[JohnStOnge]

Wrong Mr Science. One week, somebody is going to get lucky. Over the course of the whole year, the same person can't keep getting lucky. You are so full of ****. Put up or shut up.

It is indeed possible for someone to get lucky over the course of the whole year in terms of the total number of games predicted correctly. And the more people you include, the more likely you make it that you're going to include somebody that happens to.

Like for instance let's say you have 30 people who have an underlying probability of picking 65% of games correctly vs. one model that has un underlying probability of picking 70% of the games correctly. To idealize for purposes of illustration say that each of the 30 people have exactly the same proficiency: An underlying probability of 0.65.

Say you're going to allow each person to pick 275 games and compare how they did to the model. If you just select one person ahead of time there is only about a 4% chance that he or she is going to get lucky enough to get to 70 percent correct. But if you just throw 30 people in there there is better than a 70% chance that at least one of them is going to get at least 70 percent correct.

The model has about a 50% shot at getting at least 70 percent correct. If you pick one person ahead of time, chances are it's going to win. But if you throw 30 people in there and say you're going to pick the person who does best in the end for comparison, more likely than not it won't.

That's the way it works. If you want a fair test you have to decide ahead of time who you think is good and compare that one person to the model.

[BDKJMU]

Haven't read this whole thread, but has anyone mentioned how close we are to talking about a 5 Peat? As I recall, in 2010 didn't NDSU lose @ EWU in the quarterfinals in OT on a blown call? And then of course EWU went on to win the NC..

[kalm]

Haven't read this whole thread, but has anyone mentioned how close we are to talking about a 5 Peat? As I recall, in 2010 didn't NDSU lose @ EWU in the quarterfinals in OT on a blown call? And then of course EWU went on to win the NC..

No.

[Brock Landers]

Haven't read this whole thread, but has anyone mentioned how close we are to talking about a 5 Peat? As I recall, in 2010 didn't NDSU lose @ EWU in the quarterfinals in OT on a blown call? And then of course EWU went on to win the NC..

Pretty much

[JohnStOnge]

Haven't read this whole thread, but has anyone mentioned how close we are to talking about a 5 Peat? As I recall, in 2010 didn't NDSU lose @ EWU in the quarterfinals in OT on a blown call? And then of course EWU went on to win the NC..

Yeah they had a shot but I think Delaware was the best team in FCS that year. I think what happened in the championship game was a huge upset and had it's own interesting officials' call thing. Delaware's fault though. I think they were way better than Eastern Washington and should not have been in a position to have a late spot by the officials on a fourth down play make a difference in who won the game.

[89Hen]

It's fun to play that game, but that's all it is. That said, SDSU's #4 in Massey wasn't what was really egregious to begin with. It was a 5-7 WIU at 14, a 6-6 SIU at 15 and a 4-8 MSU at 20.

You don't understand Massey. Keep trying. When you figure it out, IF you figure it out, educate Z.

I understand Massey and the computer models just fine. With 11 or 12 game seasons, there isn't enough data for a computer model to be accurate. Period.

[JohnStOnge]

I understand Massey and the computer models just fine. With 11 or 12 game seasons, there isn't enough data for a computer model to be accurate. Period.

As far as I can tell you have absolutely no basis for believing that except that you just assume it. Of course they're not 100% accurate. Nothing is. But it's very clear that they have predictive value.

I have a question for you: Suppose someone tells you there is a FBS game coming up but you can't know who the teams are. All you can know is that one team is ranked higher in the polls but the other team is favored according to all of the well established vegas handicappers. In other words: Favored by the line. You are told you have to bet your house on who is going to win straight up. Which team are you going to bet on? The one that is ranked higher in the polls or the one the Vegas handicappers have as the favorite?

And don't lie.

[GannonFan]

I understand Massey and the computer models just fine. With 11 or 12 game seasons, there isn't enough data for a computer model to be accurate. Period.

As far as I can tell you have absolutely no basis for believing that except that you just assume it. Of course they're not 100% accurate. Nothing is. But it's very clear that they have predictive value.

"Predictive value"? Now you're getting pretty slim on value. If all it is is predictive value why not just flip a coin? That has "predictive value" as well but no one would put stock in that.

[YoUDeeMan]

I understand Massey and the computer models just fine. With 11 or 12 game seasons, there isn't enough data for a computer model to be accurate. Period.

As far as I can tell you have absolutely no basis for believing that except that you just assume it. Of course they're not 100% accurate. Nothing is. But it's very clear that they have predictive value.

I have a question for you: Suppose someone tells you there is a FBS game coming up but you can't know who the teams are. All you can know is that one team is ranked higher in the polls but the other team is favored according to all of the well established vegas handicappers. In other words: Favored by the line. You are told you have to bet your house on who is going to win straight up. Which team are you going to bet on? The one that is ranked higher in the polls or the one the Vegas handicappers have as the favorite?

And don't lie.

The Vegas line does not determine who will win the game or who is the better team. Vegas simply sets a line that they think will balance the money on both sides.

[89Hen]

I understand Massey and the computer models just fine. With 11 or 12 game seasons, there isn't enough data for a computer model to be accurate. Period.

As far as I can tell you have absolutely no basis for believing that except that you just assume it. Of course they're not 100% accurate. Nothing is. But it's very clear that they have predictive value.

I have a question for you: Suppose someone tells you there is a FBS game coming up but you can't know who the teams are. All you can know is that one team is ranked higher in the polls but the other team is favored according to all of the well established vegas handicappers. In other words: Favored by the line. You are told you have to bet your house on who is going to win straight up. Which team are you going to bet on? The one that is ranked higher in the polls or the one the Vegas handicappers have as the favorite?

And don't lie.

Completely stupid question JSO.

And I never even came close to saying 100% accurate. Do you agree that with more data, a computer model become more accurate? Most FCS teams play 1 or 2 games against other FCS teams out of conference (many don't play any). So the data points for ranking teams across conferences are minimal when you're talking about 120 teams and 1400 games in a 12 game season. Computers also can't take into account variables outside of the statistics such as new coaches that take a few games to get their team to adjust, an injury to a key player, weather, trap games, etc... It cracks me up that you think computer models can be anything more than fun.

[JohnStOnge]

Completely stupid question JSO

No, it's not. I have a reason for asking. Again (but with a slightly different scenario):

Let's say you have to bet your house. You are told that two teams are playing that are rated by the FBS college football championship committee. You are not told who the two teams are. But you are told that the college football championship committee has one team rated higher and the other team is the favorite by the line.

Which way are you going if you have to bet straight up on who is going to win?

[JohnStOnge]

The Vegas line does not determine who will win the game or who is the better team. Vegas simply sets a line that they think will balance the money on both sides.

Yes I know but what you are getting through that process is an effort to reflect a consensus of people who are putting money down on football. And, on the overwhelming majority of occasions, the line doesn't change by that much once they set it. They do a pretty darned good job of doing that.

[BDKJMU]

As far as I can tell you have absolutely no basis for believing that except that you just assume it. Of course they're not 100% accurate. Nothing is. But it's very clear that they have predictive value.

I have a question for you: Suppose someone tells you there is a FBS game coming up but you can't know who the teams are. All you can know is that one team is ranked higher in the polls but the other team is favored according to all of the well established vegas handicappers. In other words: Favored by the line. You are told you have to bet your house on who is going to win straight up. Which team are you going to bet on? The one that is ranked higher in the polls or the one the Vegas handicappers have as the favorite?

And don't lie.

Completely stupid question JSO.

And I never even came close to saying 100% accurate. Do you agree that with more data, a computer model become more accurate? Most FCS teams play 1 or 2 games against other FCS teams out of conference (many don't play any). So the data points for ranking teams across conferences are minimal when you're talking about 120 teams and 1400 games in a 12 game season. Computers also can't take into account variables outside of the statistics such as new coaches that take a few games to get their team to adjust, an injury to a key player, weather, trap games, etc... It cracks me up that you think computer models can be anything more than fun.

Since most of the games (all but 1-2 for most teams) are against other I-AA teams, in a 12 game season its actually going to be more like 800 games. You forgot to divide by 2. Example CAA 12 teams x 8 CAA games per team doesn't = 96 CAA games. Its 48, since 2 CAA teams playing in each game.

In a 12 game season if all 120 teams played only fellow I-AA it would be 120 x 12 divided by 2 = 720 games (not 1440).

In a 12 game season if all 120 teams played 11 fellow I-AA and one I-A or Div II team, then it would be:
120 x 11 divided by 2 = 660 I-AA vs I-AA games
120 x 1 = 120 out of division (I-AA vs I-A or Div II) games.
Total 780 games.

If all 120 teams played 10 fellow I-AA and 2 I-A or Div II games, would be:
120 x 10 divided by 2 = 600 I-AA vs I-AA games
120 x 2 = 240 out of division games.
Total 840 games.

Since the avg team is probably avg closer to 1.25-1.5 out of division games in a 12 game season, think the total number of games would be right around 800 in a 12 game season, and around 725 in an 11 game.

[89Hen]

Completely stupid question JSO

No, it's not. I have a reason for asking. Again (but with a slightly different scenario):

Let's say you have to bet your house. You are told that two teams are playing that are rated by the FBS college football championship committee. You are not told who the two teams are. But you are told that the college football championship committee has one team rated higher and the other team is the favorite by the line.

Which way are you going if you have to bet straight up on who is going to win?

The betting line? Again, WHAT does that have to do with this? PEOPLE set the betting line.

[89Hen]

Since most of the games (all but 1-2 for most teams) are against other I-AA teams, in a 12 game season its actually going to be more like 800 games. You forgot to divide by 2. Example CAA 12 teams x 8 CAA games per team doesn't = 96 CAA games. Its 48, since 2 CAA teams playing in each game.

In a 12 game season if all 120 teams played only fellow I-AA it would be 120 x 12 divided by 2 = 720 games (not 1440)...

That was a long post to say that you agree with me. Not enough data points for computer models to be accurate.