(10-01-2020 03:10 PM)beefcake0520 Wrote:  

(10-01-2020 02:14 PM)iStoopify Wrote:  

(10-01-2020 01:45 PM)beefcake0520 Wrote:  

(09-29-2020 05:48 PM)WinstonTheWolf Wrote:  SP+

https://www.espn.com/college-football/st...-things-up

42 APP

51 UL


74 stAte

80 Troy
82 Coastal

92 GSU
96 GS

104 TXST


117 USA
123 ULM

Wouldn't put a lot of stock in that, they have Marshall ranked below App....might be due to App playing 3 games and Marshall just 2 at this point, who knows.

Do people not understand probabilities? If Marshall and App played this game 3 times, do you believe Marshall would win all three? SP+ is a predictive measure.

App dropped a TD, fumbled another, missed a FG etc. Just because you beat a team once does not mean you are objectively the better team.

WTF, ok, you realize the other team can usually claim the same crap right? I didn't say we'd be app 3 times, but you don't know we wouldn't either. We dropped a TD, went for it instead of kicking the FG etc etc as well, smothered you on defense, it still evens out as I pointed out in another post. You got away with another turnover would be as well when your QB got blasted and fumbled the ball straight up in the air, but your man got the ball back being in the right place at the right time. You wanna talk about who the better team was? really? I usually come on here and have civil discussion but your BS just blew that up. You were beat in the trenches, you were beat in the running game, you were beat on special teams. What part of all of that makes Marshall not the better team? and you think that was a fluke, go watch the first game too, you will see it was not. But whatever. We were the better team for the day, for the season is still up for grabs.

haha, I thought what I said was pretty civil. The point is that those mistakes could cost you the next game and not us. Its a pretty simple concept. All major indicators (massey, SP+, sagarin) have app ranked ahead of marshall. doesn't mean they're crap, because they are actually used by casinos to set the initial line for games.

All i was pointing out is that you were wrong in that you wouldn't put stock in those ranking systems, because by and large they are excellent predictors.

Softy

(10-01-2020 02:28 PM)TroyFootball05 Wrote:  

(10-01-2020 02:14 PM)iStoopify Wrote:  

(10-01-2020 01:45 PM)beefcake0520 Wrote:  

(09-29-2020 05:48 PM)WinstonTheWolf Wrote:  SP+

https://www.espn.com/college-football/st...-things-up

42 APP

51 UL


74 stAte

80 Troy
82 Coastal

92 GSU
96 GS

104 TXST


117 USA
123 ULM

Wouldn't put a lot of stock in that, they have Marshall ranked below App....might be due to App playing 3 games and Marshall just 2 at this point, who knows.

Do people not understand probabilities? If Marshall and App played this game 3 times, do you believe Marshall would win all three? SP+ is a predictive measure.

App dropped a TD, fumbled another, missed a FG etc. Just because you beat a team once does not mean you are objectively the better team.

I would like to add Pythagorean's Argument: If A beats B, B beats C, and C beats A, how do you rank them?

So using the Pythagorean Argument, you would take the difference in each game squared and then pull one of the teams our of your nether region based on whatever bias is in the breeze that day. I think that's how it works.

(10-01-2020 02:14 PM)iStoopify Wrote:  

(10-01-2020 01:45 PM)beefcake0520 Wrote:  

(09-29-2020 05:48 PM)WinstonTheWolf Wrote:  SP+

https://www.espn.com/college-football/st...-things-up

42 APP

51 UL


74 stAte

80 Troy
82 Coastal

92 GSU
96 GS

104 TXST


117 USA
123 ULM

Wouldn't put a lot of stock in that, they have Marshall ranked below App....might be due to App playing 3 games and Marshall just 2 at this point, who knows.

Do people not understand probabilities? If Marshall and App played this game 3 times, do you believe Marshall would win all three? SP+ is a predictive measure.

App dropped a TD, fumbled another, missed a FG etc. Just because you beat a team once does not mean you are objectively the better team.

I think everyone understands the concept of probabilities. The issue is that, if this is supposed to represent a statistical probability, then there are drivers on which the score is based. Frankly, count me among those for whom those drivers are not clear at all. Maybe the lack of consistency in number of games played is creating some whacky results, or maybe this model incorporates data from previous year. I don't know but the rates don't make much sense.

I do think that if App and Marshall played 10 times this year, there would be a split. Which team the split favored probably depends on which team you pull for. We'll never know since those games won't be played. Unfortunately, Marshall won't the game that was played because they played better that day.

(10-01-2020 03:35 PM)Yosef84 Wrote:  

(10-01-2020 02:14 PM)iStoopify Wrote:  

(10-01-2020 01:45 PM)beefcake0520 Wrote:  

(09-29-2020 05:48 PM)WinstonTheWolf Wrote:  SP+

https://www.espn.com/college-football/st...-things-up

42 APP

51 UL


74 stAte

80 Troy
82 Coastal

92 GSU
96 GS

104 TXST


117 USA
123 ULM

Wouldn't put a lot of stock in that, they have Marshall ranked below App....might be due to App playing 3 games and Marshall just 2 at this point, who knows.

Do people not understand probabilities? If Marshall and App played this game 3 times, do you believe Marshall would win all three? SP+ is a predictive measure.

App dropped a TD, fumbled another, missed a FG etc. Just because you beat a team once does not mean you are objectively the better team.

I think everyone understands the concept of probabilities. The issue is that, if this is supposed to represent a statistical probability, then there are drivers on which the score is based. Frankly, count me among those for whom those drivers are not clear at all. Maybe the lack of consistency in number of games played is creating some whacky results, or maybe this model incorporates data from previous year. I don't know but the rates don't make much sense.

I do think that if App and Marshall played 10 times this year, there would be a split. Which team the split favored probably depends on which team you pull for. We'll never know since those games won't be played. Unfortunately, Marshall won't the game that was played because they played better that day.

They absolutely were the better team. But rankings systems, specifically massey and sagarin, have been around forever. To say you wouldn't put stock in them is silly, because people use them to make business decisions (casinos specifically).

It's too bad you can't quantify the gut of the general fan base. There's been plenty of times where I just knew there weren't many bullets in that gun before the team trots off to the Alamo

(10-01-2020 03:39 PM)iStoopify Wrote:  

(10-01-2020 03:35 PM)Yosef84 Wrote:  

(10-01-2020 02:14 PM)iStoopify Wrote:  

(10-01-2020 01:45 PM)beefcake0520 Wrote:  

(09-29-2020 05:48 PM)WinstonTheWolf Wrote:  SP+

https://www.espn.com/college-football/st...-things-up

42 APP

51 UL


74 stAte

80 Troy
82 Coastal

92 GSU
96 GS

104 TXST


117 USA
123 ULM

Wouldn't put a lot of stock in that, they have Marshall ranked below App....might be due to App playing 3 games and Marshall just 2 at this point, who knows.

Do people not understand probabilities? If Marshall and App played this game 3 times, do you believe Marshall would win all three? SP+ is a predictive measure.

App dropped a TD, fumbled another, missed a FG etc. Just because you beat a team once does not mean you are objectively the better team.

I think everyone understands the concept of probabilities. The issue is that, if this is supposed to represent a statistical probability, then there are drivers on which the score is based. Frankly, count me among those for whom those drivers are not clear at all. Maybe the lack of consistency in number of games played is creating some whacky results, or maybe this model incorporates data from previous year. I don't know but the rates don't make much sense.

I do think that if App and Marshall played 10 times this year, there would be a split. Which team the split favored probably depends on which team you pull for. We'll never know since those games won't be played. Unfortunately, Marshall won't the game that was played because they played better that day.

They absolutely were the better team. But rankings systems, specifically massey and sagarin, have been around forever. To say you wouldn't put stock in them is silly, because people use them to make business decisions (casinos specifically).

I didn't say that I don't put stock in predictive rankings or scores. They are fine, although certainly have their drawbacks. I simply said that the drivers for this one aren't very clear. Usually you can make SOME sense of the rankings based on YTD performance, prior year, recruiting, etc. This one seems all over the board but my guess is that it is still incorporating quite a bit of 2019 data.

(10-01-2020 03:52 PM)Yosef84 Wrote:  

(10-01-2020 03:39 PM)iStoopify Wrote:  

(10-01-2020 03:35 PM)Yosef84 Wrote:  

(10-01-2020 02:14 PM)iStoopify Wrote:  

(10-01-2020 01:45 PM)beefcake0520 Wrote:  Wouldn't put a lot of stock in that, they have Marshall ranked below App....might be due to App playing 3 games and Marshall just 2 at this point, who knows.

Do people not understand probabilities? If Marshall and App played this game 3 times, do you believe Marshall would win all three? SP+ is a predictive measure.

App dropped a TD, fumbled another, missed a FG etc. Just because you beat a team once does not mean you are objectively the better team.

I think everyone understands the concept of probabilities. The issue is that, if this is supposed to represent a statistical probability, then there are drivers on which the score is based. Frankly, count me among those for whom those drivers are not clear at all. Maybe the lack of consistency in number of games played is creating some whacky results, or maybe this model incorporates data from previous year. I don't know but the rates don't make much sense.

I do think that if App and Marshall played 10 times this year, there would be a split. Which team the split favored probably depends on which team you pull for. We'll never know since those games won't be played. Unfortunately, Marshall won't the game that was played because they played better that day.

They absolutely were the better team. But rankings systems, specifically massey and sagarin, have been around forever. To say you wouldn't put stock in them is silly, because people use them to make business decisions (casinos specifically).

I didn't say that I don't put stock in predictive rankings or scores. They are fine, although certainly have their drawbacks. I simply said that the drivers for this one aren't very clear. Usually you can make SOME sense of the rankings based on YTD performance, prior year, recruiting, etc. This one seems all over the board but my guess is that it is still incorporating quite a bit of 2019 data.

SP+ does well against the spread, historically.

We knew this year would be a bad one for ULM going in. With Caleb Evans and the whole O-line graduating, our offensive struggles aren't too surprising. Then the DC that we all wanted fired decided to quit a week before the season started, and a few more players have opted out in the past week, including one of our 3 probable NFL guys Corey Straughter (for surgery reasons).

I can't speak for the financial side of things, but we still haven't sold out a game even with (unenforced for the most part) COVID restrictions. Our reported attendance numbers are lower but the crowds look the same as last year, which is bad.

(10-01-2020 03:26 PM)iStoopify Wrote:  

(10-01-2020 03:10 PM)beefcake0520 Wrote:  

(10-01-2020 02:14 PM)iStoopify Wrote:  

(10-01-2020 01:45 PM)beefcake0520 Wrote:  

(09-29-2020 05:48 PM)WinstonTheWolf Wrote:  SP+

https://www.espn.com/college-football/st...-things-up

42 APP

51 UL


74 stAte

80 Troy
82 Coastal

92 GSU
96 GS

104 TXST


117 USA
123 ULM

Wouldn't put a lot of stock in that, they have Marshall ranked below App....might be due to App playing 3 games and Marshall just 2 at this point, who knows.

Do people not understand probabilities? If Marshall and App played this game 3 times, do you believe Marshall would win all three? SP+ is a predictive measure.

App dropped a TD, fumbled another, missed a FG etc. Just because you beat a team once does not mean you are objectively the better team.

WTF, ok, you realize the other team can usually claim the same crap right? I didn't say we'd be app 3 times, but you don't know we wouldn't either. We dropped a TD, went for it instead of kicking the FG etc etc as well, smothered you on defense, it still evens out as I pointed out in another post. You got away with another turnover would be as well when your QB got blasted and fumbled the ball straight up in the air, but your man got the ball back being in the right place at the right time. You wanna talk about who the better team was? really? I usually come on here and have civil discussion but your BS just blew that up. You were beat in the trenches, you were beat in the running game, you were beat on special teams. What part of all of that makes Marshall not the better team? and you think that was a fluke, go watch the first game too, you will see it was not. But whatever. We were the better team for the day, for the season is still up for grabs.

haha, I thought what I said was pretty civil. The point is that those mistakes could cost you the next game and not us. Its a pretty simple concept. All major indicators (massey, SP+, sagarin) have app ranked ahead of marshall. doesn't mean they're crap, because they are actually used by casinos to set the initial line for games.

All i was pointing out is that you were wrong in that you wouldn't put stock in those ranking systems, because by and large they are excellent predictors.

Softy

No, making excuses and using those excuses in your selling point is very demeaning to the program you lose to, especially if the other program could use those very same excuses as to why you didn't get beat by 3 or 4 scores instead of 2. There is a difference in losing because of unforced errors (all of apps were forced which equals good defensive play) and losing because you just got beat. App has a good squad, I haven't seen any herd fan say otherwise, we have been complimentary of your program, one we wouldn't mind sharing a conference with. Excuses are like buttholes, everyone has them. We played with a redshirt freshman who never took a snap until this year, do you see any herd fans use that as an excuse because we didn't roll up 600 yards on you, no, you see most if not all give credit where it was due, your defense prevented him from doing what he did to EKU. If you go on SP ratings and use that as your measuring stick, that's like making a bad bet in the stock market, no more, no less. Vegas does not watch the games, they play on statistics and how people are betting to come up with their lines. The actual game that gets played is all that matters. Agree to disagree, don't really care.

(10-01-2020 03:52 PM)Yosef84 Wrote:  

(10-01-2020 03:39 PM)iStoopify Wrote:  

(10-01-2020 03:35 PM)Yosef84 Wrote:  

(10-01-2020 02:14 PM)iStoopify Wrote:  

(10-01-2020 01:45 PM)beefcake0520 Wrote:  Wouldn't put a lot of stock in that, they have Marshall ranked below App....might be due to App playing 3 games and Marshall just 2 at this point, who knows.

Do people not understand probabilities? If Marshall and App played this game 3 times, do you believe Marshall would win all three? SP+ is a predictive measure.

App dropped a TD, fumbled another, missed a FG etc. Just because you beat a team once does not mean you are objectively the better team.

I think everyone understands the concept of probabilities. The issue is that, if this is supposed to represent a statistical probability, then there are drivers on which the score is based. Frankly, count me among those for whom those drivers are not clear at all. Maybe the lack of consistency in number of games played is creating some whacky results, or maybe this model incorporates data from previous year. I don't know but the rates don't make much sense.

I do think that if App and Marshall played 10 times this year, there would be a split. Which team the split favored probably depends on which team you pull for. We'll never know since those games won't be played. Unfortunately, Marshall won't the game that was played because they played better that day.

They absolutely were the better team. But rankings systems, specifically massey and sagarin, have been around forever. To say you wouldn't put stock in them is silly, because people use them to make business decisions (casinos specifically).

I didn't say that I don't put stock in predictive rankings or scores. They are fine, although certainly have their drawbacks. I simply said that the drivers for this one aren't very clear. Usually you can make SOME sense of the rankings based on YTD performance, prior year, recruiting, etc. This one seems all over the board but my guess is that it is still incorporating quite a bit of 2019 data.

This ^^^^^^ makes a lot more sense. For SP to be accurate, you have to have a larger sample to go off of, 2 or 3 games is not large enough to use in an accurate rating, the creator of SP even states that. https://www.sbnation.com/college-footbal...s-rankings It would make sense where they would use 2019 data with the small sample size of this year to come up with an accurate rating. App was the better team last year.

(10-02-2020 10:17 AM)beefcake0520 Wrote:  

(10-01-2020 03:52 PM)Yosef84 Wrote:  

(10-01-2020 03:39 PM)iStoopify Wrote:  

(10-01-2020 03:35 PM)Yosef84 Wrote:  

(10-01-2020 02:14 PM)iStoopify Wrote:  Do people not understand probabilities? If Marshall and App played this game 3 times, do you believe Marshall would win all three? SP+ is a predictive measure.

App dropped a TD, fumbled another, missed a FG etc. Just because you beat a team once does not mean you are objectively the better team.

I think everyone understands the concept of probabilities. The issue is that, if this is supposed to represent a statistical probability, then there are drivers on which the score is based. Frankly, count me among those for whom those drivers are not clear at all. Maybe the lack of consistency in number of games played is creating some whacky results, or maybe this model incorporates data from previous year. I don't know but the rates don't make much sense.

I do think that if App and Marshall played 10 times this year, there would be a split. Which team the split favored probably depends on which team you pull for. We'll never know since those games won't be played. Unfortunately, Marshall won't the game that was played because they played better that day.

They absolutely were the better team. But rankings systems, specifically massey and sagarin, have been around forever. To say you wouldn't put stock in them is silly, because people use them to make business decisions (casinos specifically).

I didn't say that I don't put stock in predictive rankings or scores. They are fine, although certainly have their drawbacks. I simply said that the drivers for this one aren't very clear. Usually you can make SOME sense of the rankings based on YTD performance, prior year, recruiting, etc. This one seems all over the board but my guess is that it is still incorporating quite a bit of 2019 data.

This ^^^^^^ makes a lot more sense. For SP to be accurate, you have to have a larger sample to go off of, 2 or 3 games is not large enough to use in an accurate rating, the creator of SP even states that. https://www.sbnation.com/college-footbal...s-rankings It would make sense where they would use 2019 data with the small sample size of this year to come up with an accurate rating. App was the better team last year.

SP+ uses a LOT of historical data early in the season by necessity. It's why it typically isn't hyperreactive to early season results although I believe it's been altered more recently to be more responsive to teams that start strong even if they had a bad previous season. Around midseason is when the picture clears up and you can see where teams kind of settle out at.

This is not a normal year however. We're in October and teams have played a different amount of games. Teams like FAU and Houston have yet to even play a game although they've been trying.

Right now SP+ is still heavily weighted to historical data. It doesn't mean it's wrong, just there hasn't been enough data to get a good idea of where teams are at currently with everything going on. It might take to the end of October for it to come into focus a bit more...although it may take longer. We'll have to see.

(10-02-2020 10:27 AM)eaglewraith Wrote:  

(10-02-2020 10:17 AM)beefcake0520 Wrote:  

(10-01-2020 03:52 PM)Yosef84 Wrote:  

(10-01-2020 03:39 PM)iStoopify Wrote:  

(10-01-2020 03:35 PM)Yosef84 Wrote:  I think everyone understands the concept of probabilities. The issue is that, if this is supposed to represent a statistical probability, then there are drivers on which the score is based. Frankly, count me among those for whom those drivers are not clear at all. Maybe the lack of consistency in number of games played is creating some whacky results, or maybe this model incorporates data from previous year. I don't know but the rates don't make much sense.

I do think that if App and Marshall played 10 times this year, there would be a split. Which team the split favored probably depends on which team you pull for. We'll never know since those games won't be played. Unfortunately, Marshall won't the game that was played because they played better that day.

They absolutely were the better team. But rankings systems, specifically massey and sagarin, have been around forever. To say you wouldn't put stock in them is silly, because people use them to make business decisions (casinos specifically).

I didn't say that I don't put stock in predictive rankings or scores. They are fine, although certainly have their drawbacks. I simply said that the drivers for this one aren't very clear. Usually you can make SOME sense of the rankings based on YTD performance, prior year, recruiting, etc. This one seems all over the board but my guess is that it is still incorporating quite a bit of 2019 data.

This ^^^^^^ makes a lot more sense. For SP to be accurate, you have to have a larger sample to go off of, 2 or 3 games is not large enough to use in an accurate rating, the creator of SP even states that. https://www.sbnation.com/college-footbal...s-rankings It would make sense where they would use 2019 data with the small sample size of this year to come up with an accurate rating. App was the better team last year.

SP+ uses a LOT of historical data early in the season by necessity. It's why it typically isn't hyperreactive to early season results although I believe it's been altered more recently to be more responsive to teams that start strong even if they had a bad previous season. Around midseason is when the picture clears up and you can see where teams kind of settle out at.

This is not a normal year however. We're in October and teams have played a different amount of games. Teams like FAU and Houston have yet to even play a game although they've been trying.

Right now SP+ is still heavily weighted to historical data. It doesn't mean it's wrong, just there hasn't been enough data to get a good idea of where teams are at currently with everything going on. It might take to the end of October for it to come into focus a bit more...although it may take longer. We'll have to see.

Very good analysis and I agree. It's also a reason why is shouldn't be used for any type of ranking talk unless there is more of a sample size for the given season to utilize. UL should be above us and App, they are barely above us and below app though. Anyone that tries to use that system at this point of the year is really reaching for something to hold on to.

(10-02-2020 02:51 PM)beefcake0520 Wrote:  

(10-02-2020 10:27 AM)eaglewraith Wrote:  

(10-02-2020 10:17 AM)beefcake0520 Wrote:  

(10-01-2020 03:52 PM)Yosef84 Wrote:  

(10-01-2020 03:39 PM)iStoopify Wrote:  They absolutely were the better team. But rankings systems, specifically massey and sagarin, have been around forever. To say you wouldn't put stock in them is silly, because people use them to make business decisions (casinos specifically).

I didn't say that I don't put stock in predictive rankings or scores. They are fine, although certainly have their drawbacks. I simply said that the drivers for this one aren't very clear. Usually you can make SOME sense of the rankings based on YTD performance, prior year, recruiting, etc. This one seems all over the board but my guess is that it is still incorporating quite a bit of 2019 data.

This ^^^^^^ makes a lot more sense. For SP to be accurate, you have to have a larger sample to go off of, 2 or 3 games is not large enough to use in an accurate rating, the creator of SP even states that. https://www.sbnation.com/college-footbal...s-rankings It would make sense where they would use 2019 data with the small sample size of this year to come up with an accurate rating. App was the better team last year.

SP+ uses a LOT of historical data early in the season by necessity. It's why it typically isn't hyperreactive to early season results although I believe it's been altered more recently to be more responsive to teams that start strong even if they had a bad previous season. Around midseason is when the picture clears up and you can see where teams kind of settle out at.

This is not a normal year however. We're in October and teams have played a different amount of games. Teams like FAU and Houston have yet to even play a game although they've been trying.

Right now SP+ is still heavily weighted to historical data. It doesn't mean it's wrong, just there hasn't been enough data to get a good idea of where teams are at currently with everything going on. It might take to the end of October for it to come into focus a bit more...although it may take longer. We'll have to see.

Very good analysis and I agree. It's also a reason why is shouldn't be used for any type of ranking talk unless there is more of a sample size for the given season to utilize. UL should be above us and App, they are barely above us and below app though. Anyone that tries to use that system at this point of the year is really reaching for something to hold on to.

I'd rather use it than an arbitrary voted on ranking tbh.

(10-02-2020 02:51 PM)beefcake0520 Wrote:  

(10-02-2020 10:27 AM)eaglewraith Wrote:  

(10-02-2020 10:17 AM)beefcake0520 Wrote:  

(10-01-2020 03:52 PM)Yosef84 Wrote:  

(10-01-2020 03:39 PM)iStoopify Wrote:  They absolutely were the better team. But rankings systems, specifically massey and sagarin, have been around forever. To say you wouldn't put stock in them is silly, because people use them to make business decisions (casinos specifically).

I didn't say that I don't put stock in predictive rankings or scores. They are fine, although certainly have their drawbacks. I simply said that the drivers for this one aren't very clear. Usually you can make SOME sense of the rankings based on YTD performance, prior year, recruiting, etc. This one seems all over the board but my guess is that it is still incorporating quite a bit of 2019 data.

This ^^^^^^ makes a lot more sense. For SP to be accurate, you have to have a larger sample to go off of, 2 or 3 games is not large enough to use in an accurate rating, the creator of SP even states that. https://www.sbnation.com/college-footbal...s-rankings It would make sense where they would use 2019 data with the small sample size of this year to come up with an accurate rating. App was the better team last year.

SP+ uses a LOT of historical data early in the season by necessity. It's why it typically isn't hyperreactive to early season results although I believe it's been altered more recently to be more responsive to teams that start strong even if they had a bad previous season. Around midseason is when the picture clears up and you can see where teams kind of settle out at.

This is not a normal year however. We're in October and teams have played a different amount of games. Teams like FAU and Houston have yet to even play a game although they've been trying.

Right now SP+ is still heavily weighted to historical data. It doesn't mean it's wrong, just there hasn't been enough data to get a good idea of where teams are at currently with everything going on. It might take to the end of October for it to come into focus a bit more...although it may take longer. We'll have to see.

Very good analysis and I agree. It's also a reason why is shouldn't be used for any type of ranking talk unless there is more of a sample size for the given season to utilize. UL should be above us and App, they are barely above us and below app though. Anyone that tries to use that system at this point of the year is really reaching for something to hold on to.

If it is good enought to be predictive to have some success against the betting line, it is good enough for rankings assessment - at least it is objective. Early on it is going to be off on some teams - but there IS year to year statistical correlation in college football - is it the same for all teams - no but how a team did last year is a reasonable indicator of probable team strength the following year.

(10-02-2020 05:44 PM)WinstonTheWolf Wrote:  

(10-02-2020 02:51 PM)beefcake0520 Wrote:  

(10-02-2020 10:27 AM)eaglewraith Wrote:  

(10-02-2020 10:17 AM)beefcake0520 Wrote:  

(10-01-2020 03:52 PM)Yosef84 Wrote:  I didn't say that I don't put stock in predictive rankings or scores. They are fine, although certainly have their drawbacks. I simply said that the drivers for this one aren't very clear. Usually you can make SOME sense of the rankings based on YTD performance, prior year, recruiting, etc. This one seems all over the board but my guess is that it is still incorporating quite a bit of 2019 data.

This ^^^^^^ makes a lot more sense. For SP to be accurate, you have to have a larger sample to go off of, 2 or 3 games is not large enough to use in an accurate rating, the creator of SP even states that. https://www.sbnation.com/college-footbal...s-rankings It would make sense where they would use 2019 data with the small sample size of this year to come up with an accurate rating. App was the better team last year.

SP+ uses a LOT of historical data early in the season by necessity. It's why it typically isn't hyperreactive to early season results although I believe it's been altered more recently to be more responsive to teams that start strong even if they had a bad previous season. Around midseason is when the picture clears up and you can see where teams kind of settle out at.

This is not a normal year however. We're in October and teams have played a different amount of games. Teams like FAU and Houston have yet to even play a game although they've been trying.

Right now SP+ is still heavily weighted to historical data. It doesn't mean it's wrong, just there hasn't been enough data to get a good idea of where teams are at currently with everything going on. It might take to the end of October for it to come into focus a bit more...although it may take longer. We'll have to see.

Very good analysis and I agree. It's also a reason why is shouldn't be used for any type of ranking talk unless there is more of a sample size for the given season to utilize. UL should be above us and App, they are barely above us and below app though. Anyone that tries to use that system at this point of the year is really reaching for something to hold on to.

If it is good enought to be predictive to have some success against the betting line, it is good enough for rankings assessment - at least it is objective. Early on it is going to be off on some teams - but there IS year to year statistical correlation in college football - is it the same for all teams - no but how a team did last year is a reasonable indicator of probable team strength the following year.

While that is true for a squad returning most of the same elements, that can't be true for a squad with key turnover. SP should start from scratch when that is the case because the squad could be stronger than predicted or weaker. It's a system that is in the mid 50s on accuracy in prediction. Well, doesn't seem that accurate as a ranking structure. Results on the field are a much more accurate ranking structure but only after 4-5 games have been played to make sure quality wins wasn't a fluke.

(10-03-2020 10:54 AM)beefcake0520 Wrote:  

(10-02-2020 05:44 PM)WinstonTheWolf Wrote:  

(10-02-2020 02:51 PM)beefcake0520 Wrote:  

(10-02-2020 10:27 AM)eaglewraith Wrote:  

(10-02-2020 10:17 AM)beefcake0520 Wrote:  This ^^^^^^ makes a lot more sense. For SP to be accurate, you have to have a larger sample to go off of, 2 or 3 games is not large enough to use in an accurate rating, the creator of SP even states that. https://www.sbnation.com/college-footbal...s-rankings It would make sense where they would use 2019 data with the small sample size of this year to come up with an accurate rating. App was the better team last year.

SP+ uses a LOT of historical data early in the season by necessity. It's why it typically isn't hyperreactive to early season results although I believe it's been altered more recently to be more responsive to teams that start strong even if they had a bad previous season. Around midseason is when the picture clears up and you can see where teams kind of settle out at.

This is not a normal year however. We're in October and teams have played a different amount of games. Teams like FAU and Houston have yet to even play a game although they've been trying.

Right now SP+ is still heavily weighted to historical data. It doesn't mean it's wrong, just there hasn't been enough data to get a good idea of where teams are at currently with everything going on. It might take to the end of October for it to come into focus a bit more...although it may take longer. We'll have to see.

Very good analysis and I agree. It's also a reason why is shouldn't be used for any type of ranking talk unless there is more of a sample size for the given season to utilize. UL should be above us and App, they are barely above us and below app though. Anyone that tries to use that system at this point of the year is really reaching for something to hold on to.

If it is good enought to be predictive to have some success against the betting line, it is good enough for rankings assessment - at least it is objective. Early on it is going to be off on some teams - but there IS year to year statistical correlation in college football - is it the same for all teams - no but how a team did last year is a reasonable indicator of probable team strength the following year.

While that is true for a squad returning most of the same elements, that can't be true for a squad with key turnover. SP should start from scratch when that is the case because the squad could be stronger than predicted or weaker. It's a system that is in the mid 50s on accuracy in prediction. Well, doesn't seem that accurate as a ranking structure. Results on the field are a much more accurate ranking structure but only after 4-5 games have been played to make sure quality wins wasn't a fluke.

50+% against the spread is really good actually.

1 APP
2 Coastal

3 UL

4 stAte

5 Troy

6 GSU
7 GS

8 TXST


9 USA




10ULM

(10-14-2020 10:35 PM)Coastal12 Wrote:  1 APP
2 Coastal

3 UL

4 stAte

5 Troy

6 GSU
7 GS

8 TXST


9 USA




10ULM

you should let the game week finish before posting this.

(10-15-2020 11:59 AM)balanced_view Wrote:  

(10-14-2020 10:35 PM)Coastal12 Wrote:  1 APP
2 Coastal

3 UL

4 stAte

5 Troy

6 GSU
7 GS

8 TXST


9 USA




10ULM

you should let the game week finish before posting this.

Was thinking the same.

Sent from my SM-G975U using Tapatalk

(10-14-2020 10:35 PM)Coastal12 Wrote:  1 APP
2 Coastal

3 UL

4 stAte

5 Troy

6 GSU
7 GS

8 TXST


9 USA




10ULM

Not now train bot