Need in help in developing an algorithm for the 2014 college football season

[4down20]

My point is look at history.

Alabama was suppose to rip apart Utah and Oklahoma. They didn't.
Florida was suppose to be no match for Louisville.
Clemson was a big favorite over West Virginia.

Why must rankings be base on vegas odds? It should be about RESUME, not on subjective assumptions or probabilities.

Probability is a measure on what is most likely outcome. Just because upsets and things which weren't the most probable outcome do not happen, it doesn't negate the value of probability.

Probability in computers doesn't come from vegas odds, maybe vice versa at times. It comes from looking at the data and using a formula to derive useful information.

If something has a 90% chance of happening, then it can still happen. It's just not likely. If Alabama were to play Oklahoma again, Alabama would be the likely favorite despite the loss. And that would be based on stats.

 

[WNY_FOOTBALL_DUDE]

Probability is a measure on what is most likely outcome.

I know that, but that's for sports betting purposes or forecasting, not for determining resume or deserve.

Probable outcome is more in the OPINION territory. I am looking at something more objective.

 

[4down20]

I know that, but that's for sports betting purposes or forecasting, not for determining resume or deserve.

Probable outcome is more in the OPINION territory. I am looking at something more objective.

It's not opinion, it's a measure.

 

[WNY_FOOTBALL_DUDE]

It's not opinion, it's a measure.

Measurement based on what factors? College football lacks common opponents between the conferences.

 

[4down20]

Measurement based on what factors? College football lacks common opponents between the conferences.

It doesn't lack common opponents. There are links between the conferences. More links would make the probabilities more accurate, but it doesn't have to be direct links between every team to have value.

The stats you choose and the weight you give to them in the formula is what makes it your formula.

 

[iowajerms]

Some suggestions

Record
Opponents record
Opponents opponents record
H/A
Score differential or PF and PA

 

[WNY_FOOTBALL_DUDE]

Some suggestions

Record
Opponents record
Opponents opponents record
H/A
Score differential or PF and PA

Thanks. I decided to go with the following rules:

50% winning + 25% opponents winning% + 25% opponents of opponents winning%
FCS opponents are going to be worth a flat .167.
Home wins = .70, Road wins = 1.3. Vice verse for losses. Neutral site is 1.0

 

[boxedlunch]

Thanks. I decided to go with the following rules:

50% winning + 25% opponents winning% + 25% opponents of opponents winning%
FCS opponents are going to be worth a flat .167.
Home wins = .70, Road wins = 1.3. Vice verse for losses. Neutral site is 1.0

A technical question for you. By giving FCS opponents a flat percentage, does that mean you intend to average opponents winning percentage, give them a flat win loss record (ie 2-10), or perhaps a weighted win loss record for the season?

 

[WNY_FOOTBALL_DUDE]

A technical question for you. By giving FCS opponents a flat percentage, does that mean you intend to average opponents winning percentage, give them a flat win loss record (ie 2-10), or perhaps a weighted win loss record for the season?

Opponents win% and opponents of opponents winning% will be at .167 each.

 

[boxedlunch]

Opponents win% and opponents of opponents winning% will be at .167 each.

What I'm asking it this. You might have a team with opponents that have these records:

Team A 3-9
Team B 8-5
Team C 7-6
Team FCS 11-4

Now to calculate opponents winning percentage, you add the wins, 3 + 8 + 7 + 11 and divide by the wins + losses. However, you're saying Team FCS is going to be viewed with a winning percentage of .167. You can't calculate total winning percentage without assuming a certain number of wins and losses. Now, I get you can assume 1-5 or 2-10 and maybe that's what you have in mind, or maybe since you're giving them a winning percentage you have something else in mind. Just wondering how you would do that functionally?

 

[WNY_FOOTBALL_DUDE]

What I'm asking it this. You might have a team with opponents that have these records:

Team A 3-9
Team B 8-5
Team C 7-6
Team FCS 11-4

Now to calculate opponents winning percentage, you add the wins, 3 + 8 + 7 + 11 and divide by the wins + losses. However, you're saying Team FCS is going to be viewed with a winning percentage of .167. You can't calculate total winning percentage without assuming a certain number of wins and losses. Now, I get you can assume 1-5 or 2-10 and maybe that's what you have in mind, or maybe since you're giving them a winning percentage you have something else in mind. Just wondering how you would do that functionally?

An FCS school is an FCS School. Their schedules are weaker than the FBS ones. It would be unfair to treat FCS as an FBS school. I figure I either treat FCS opponents as extremely weak FBS opponent or I provide no credit for playing FCS schools or I ignore their existence. I pick option#1.

If you have any suggestions on how I should treat FCS opponents, let me know.

 

[boxedlunch]

An FCS school is an FCS School. Their schedules are weaker than the FBS ones. It would be unfair to treat FCS as an FBS school. I figure I either treat FCS opponents as extremely weak FBS opponent or I provide no credit for playing FCS schools or I ignore their existence. I pick option#1.

If you have any suggestions on how I should treat FCS opponents, let me know.

I have no objection to they way you decided to treat FCS schools, I was just asking how you planned to physically calculate it. You can't calculate true win/loss percentage by just giving them a winning percentage. You have to give them win/loss record if it works, and what you give them would affect their score. I was wondering how you intended to work this out.

 

[WNY_FOOTBALL_DUDE]

I have no objection to they way you decided to treat FCS schools, I was just asking how you planned to physically calculate it. You can't calculate true win/loss percentage by just giving them a winning percentage. You have to give them win/loss record if it works, and what you give them would affect their score. I was wondering how you intended to work this out.

Here's my spreadsheet:

Column one = List of FBS teams
Column two = Winning% per FBS team
Column three = Opponents winning% (average together all your opponents winning% together)
Column four = Opponents of opponents winning% (average together all your opponents winning% together)
Column five = Total or final rating.

For columns 3 and 4, the FCS opponent receives a .167 calculation.

 

[Codaxx]

Here's my spreadsheet:

Column one = List of FBS teams
Column two = Winning% per FBS team
Column three = Opponents winning% (average together all your opponents winning% together)
Column four = Opponents of opponents winning% (average together all your opponents winning% together)
Column five = Total or final rating.

For columns 3 and 4, the FCS opponent receives a .167 calculation.

No home field adjustment?

 

[WNY_FOOTBALL_DUDE]

No home field adjustment?

Yes, home wins = .7, away wins = 1.3.

I was just trying to illustrate how I will input FCS opponents into the formula.

 

[4down20]

Yes, home wins = .7, away wins = 1.3.

I was just trying to illustrate how I will input FCS opponents into the formula.

So an away win is worth nearly double in your opinion?

 

[Codaxx]

Yes, home wins = .7, away wins = 1.3.

I was just trying to illustrate how I will input FCS opponents into the formula.

I think I would lower that. We know homefield is worth around a little over 3 pts. That equates to about a 60% winning percentage. That goes to 1.2 and .8. I know Baseball RPI uses your number and basketball uses 1.4 and .6..

 

[WNY_FOOTBALL_DUDE]

So an away win is worth nearly double in your opinion?

Yes sir. It's slightly harder to win on the road. I gave out a modest .3 aka field goal difference. I wanted to find a happy medium. Not too liberal and not too conservative.

 

[boxedlunch]

Here's my spreadsheet:

Column one = List of FBS teams
Column two = Winning% per FBS team
Column three = Opponents winning% (average together all your opponents winning% together)
Column four = Opponents of opponents winning% (average together all your opponents winning% together)
Column five = Total or final rating.

For columns 3 and 4, the FCS opponent receives a .167 calculation.

Okay, so you intend to average the opponents winning percentage together rather than calculating actual winning percentage. Thanks, that's all I wanted to know.