Vegas over under on season 7.5

#51
#51
marmoteer said:
And here are the probabilities of getting to at least n wins:
#wins %prob
1 99.999959
2 99.998747
3 99.976737
4 99.732950
5 98.084062
6 91.263209
7 73.984160
8 47.152834
9 21.601745
10 6.680720
11 1.340457
12 0.170262
Click to expand...

Last year FPI had us so over-rated that it gave us a greater than 8% chance at 12-0 regular season.

I really like some of the timing of our schedule this year. The teams we face often have some tough games before we face them, which matters in the SEC but doesn't get factored into most models.
 
  • Like
Reactions: 2 people
#52
#52
marmoteer said:
Using these probabilities, you should expect 7.9 wins. Here's how you can calculate: for each game, choose a random number between 0 and 1; if the number is less than the "win probability, count it as a win (else, loss). Count up the wins for the season. Do this for a million seasons. The average number of wins is 7.9.

This assumes that 1) the win probabilities from the FPI are meaningful and 2) one game doesn't depend on any other game. Definitely 2) is not true; for example, losing a key player in Game 2 affects Game 3. Whether or not FPI takes this into account, I have no idea.
Click to expand...

You can use any data set you want, but compound probability is still computed in the same way. What you are proposing is to basically not use the FPI numbers.
 
  • Like
Reactions: 1 person
#53
#53
Wylo said:
Save some cash for the winter. You're gonna need it.
Click to expand...

:lolabove::lolabove::lolabove::lolabove:

This thread proves 7.5 is the correct number. About half the people say over and half say under.

If the talent Butch has recruited performs then 8 wins should be doable.
 
  • Like
Reactions: 2 people
#54
#54
Boca Vol said:
:lolabove::lolabove::lolabove::lolabove:

This thread proves 7.5 is the correct number. About half the people say over and half say under.

If the talent Butch has recruited performs then 8 wins should be doable.
Click to expand...

It's the highest number that I feel comfortable betting the over on.
If we were only replacing a few stars then I'd be willing to go higher but not that much talent and so many assistants.

So many unknowns but at the same time, it gives a little extra excitement.
 
  • Like
Reactions: 1 person
#55
#55
BeardedVol said:
You can use any data set you want, but compound probability is still computed in the same way. What you are proposing is to basically not use the FPI numbers.
Click to expand...

I used the FPI like this: for Florida our probability of winning is .378. I pick a random number between 0 and 1, and if it's less than .378, I count it as a win. This is the correct way of calculating the expected number of wins if you can't (don't want to) calculate the probabilities analytically. The compound probability of about .25 that you cite only covers one case: there are 495 ways to get 8 wins out of a 12 game schedule. 220 ways to choose 9 wins, 66 ways to choose 10 wins, 12 ways to get 11 wins, and 1 way to get 12 wins.
 
  • Like
Reactions: 2 people
#56
#56
Rifleman said:
Last year FPI had us so over-rated that it gave us a greater than 8% chance at 12-0 regular season.

I really like some of the timing of our schedule this year. The teams we face often have some tough games before we face them, which matters in the SEC but doesn't get factored into most models.
Click to expand...

I like the timing too. I can't really remember having our big games so spread out. Playing LSU and Guice in November worries me form a DL health standpoint, but I also think our offense will be rolling by then.
 
  • Like
Reactions: 2 people
#57
#57
marmoteer said:
I used the FPI like this: for Florida our probability of winning is .378. I pick a random number between 0 and 1, and if it's less than .378, I count it as a win. This is the correct way of calculating the expected number of wins if you can't (don't want to) calculate the probabilities analytically. The compound probability of about .25 that you cite only covers one case: there are 495 ways to get 8 wins out of a 12 game schedule. 220 ways to choose 9 wins, 66 ways to choose 10 wins, 12 ways to get 11 wins, and 1 way to get 12 wins.
Click to expand...

That isn't using FPI as a basis, but rather you boiling down FPI to a binary calculation of 'will win' or 'will lose'. Unfortunately that's a poor way to try to predict the outcome of a football game with their complex nature.

With any collection of independent events the probability is still computed as P(A) and P(B)=P(A*B). You can plug in the number for any 8 games on the schedule and get the compound probability of winning all 8 of those games.
 
Last edited:
  • Like
Reactions: 1 person
#58
#58
I see we are talking statistics again using metrics that ESPN clearly states is error prone and will be adjusted as the year progresses.

Based on the RPI, Tennessee is currently favored to win 9 games. The numbers will change week to week based on how Tennessee does and how the opponents do.

Georgia is also favored to win 9 games (projected losses to Notre Dame, Auburn and Tennessee)

Florida if favored to win 10 games including beating Michigan ... losing to GA and FSU.
 
  • Like
Reactions: 1 person
#59
#59
volfan102455 said:
I see we are talking statistics again using metrics that ESPN clearly states is error prone and will be adjusted as the year progresses.

Based on the RPI, Tennessee is currently favored to win 9 games. The numbers will change week to week based on how Tennessee does and how the opponents do.

Georgia is also favored to win 9 games (projected losses to Notre Dame, Auburn and Tennessee)

Florida if favored to win 10 games including beating Michigan ... losing to GA and FSU.
Click to expand...

Nooo, say it ain't so. I thought all this was straight up legitimate. I was getting ready to write it all down. I've got my pencil, give me something to write on.
 
  • Like
Reactions: 2 people
#60
#60
CAVPUT said:
I see our ceiling being 7 wins. Just need to somehow pull out a win against GT and come out injury free. A team like Bama can lose the level of talent we did and simply reload. We aren't at that level. I see a step back this year. 6-7 wins looks about right for our roster facing this schedule.
Click to expand...

I just don't see us losing to Ga, Fl, Al, Lsu and one more if this team stays healthy.
 
  • Like
Reactions: 2 people
#61
#61
Boca Vol said:
If I was in Vegas, I would bet the over. IMO, UT will be 8-4 or 9-3 in the regular season.
Click to expand...

I just don't see 9 wins with Butch and this team, especially breaking in a new QB and offensive coordinator... we went 8-4 with a more talented roster and the top offensive player in the conference at QB last year. I think 8 wins is Butch's ceiling as a HC in the SEC in my opinion. I think we go 7-5 or 6-6 with losses to Bama, LSU, UGA, Florida, and one or two losses in the SC, UK, Vandy, Mizzou or GT games.
 
Last edited:
  • Like
Reactions: 2 people
#62
#62
we should be happy with that this year. we need to get over fire the coach will fix it idea. we need to keep up grading our players. Bama and Ohio State have good seasons because they have the best players.
 
#63
#63
Behr said:
Nooo, say it ain't so. I thought all this was straight up legitimate. I was getting ready to write it all down. I've got my pencil, give me something to write on.
Click to expand...

LOL :) It is in black and white on the top of the site.

The Football Power Index (FPI) is a measure of team strength that is meant to be the best predictor of a team's performance going forward for the rest of the season. FPI represents how many points above or below average a team is. Projected results are based on 10,000 simulations of the rest of the season using FPI, results to date, and the remaining schedule. Ratings and projections update daily.

I would go with over based on my gut. To get to under implies losses to both GT and GA or a surprise losses to one or two from the following - SC, KY, Missouri or Vandy.
 
  • Like
Reactions: 1 person
#64
#64
Evet Vols said:
I just don't see us losing to Ga, Fl, Al, Lsu and one more if this team stays healthy.
Click to expand...

Me neither which is why taking the "under" doesn't make sense at this point. The GT game is a little concerning not because of who it is but because they are very different offensively and that offensive if they are successful will keep Tennessee's offense off the field. That will keep the score lower and that neutralizes the talent somewhat.
 
  • Like
Reactions: 1 person
#65
#65
Catbone said:
You mustve lost your house last year
Click to expand...

Good thing this isn't last year then huh? This thread is discussing the odds and THIS year's team. Everybody trying to spin the doom and gloom narrative for this year. I really think Tennessee is going to surprise a lot of people. I see 9 wins with a chance at 10 if a couple of bounces go our way.
 
  • Like
Reactions: 2 people
#67
#67
forgiven said:
we should be happy with that this year. we need to get over fire the coach will fix it idea. we need to keep up grading our players. Bama and Ohio State have good seasons because they have the best players.
Click to expand...

We had two top 5 classes and the best Butch could do with them was 9-4. He's had plenty of time to build the roster.
 
  • Like
Reactions: 2 people
#68
#68
BeardedVol said:
That isn't using FPI as a basis, but rather you boiling down FPI to a binary calculation of 'will win' or 'will lose'. Unfortunately that's a poor way to try to predict the outcome of a football game their complex nature.

With any collection of independent events the probability is still computed as P(A) and P(B)=P(A*B). You can plug in the number for any 8 games on the schedule and get the compound probability of winning all 8 of those games.
Click to expand...

This is the right way to calculate it; it's called Monte Carlo simulation.

I'm treating the % you cited for each game as a probability. You're trying to compute the probability of winning precisely those 8 games. I'm saying we have about 50% probability of winning at least 8 games, given the assumptions I listed previously.
 
  • Like
Reactions: 1 person
#71
#71
marmoteer said:
This is the right way to calculate it; it's called Monte Carlo simulation.

I'm treating the % you cited for each game as a probability. You're trying to compute the probability of winning precisely those 8 games. I'm saying we have about 50% probability of winning at least 8 games, given the assumptions I listed previously.
Click to expand...

FPI is essentially a monte carlo simulation already, and according to ESPN they run 10k simulations to come up with their individual FPI rankings.

UT does not have a 50% chance of winning 8 games, not according to FPI calculations, and calculating the chances to win the 8 games where UT is the most highly favored. According to FPI, the chance of winning those 8 games is just under 25%, and that is your best case scenario under FPI for winning 8 games, as once you start substituting in the games where UT has a lower chance of winning, the overall chance of winning all 8 games in the data set will be lower.

You are never going to get to an overall 50% chance of winning 8 games when the best case scenario for winning 8 games has only a 25% chance, and all of the other scenarios are less then 25% chance of occurring.

Again, this is all based off of pre-season FPI being considered relatively accurate, which it's probably not, but the math still stands.
 
  • Like
Reactions: 1 person
#72
#72
BeardedVol said:
FPI is essentially a monte carlo simulation already, and according to ESPN they run 10k simulations to come up with their individual FPI rankings.

UT does not have a 50% chance of winning 8 games, not according to FPI calculations, and calculating the chances to win the 8 games where UT is the most highly favored. According to FPI, the chance of winning those 8 games is just under 25%, and that is your best case scenario under FPI for winning 8 games, as once you start substituting in the games where UT has a lower chance of winning, the overall chance of winning all 8 games in the data set will be lower.

You are never going to get to an overall 50% chance of winning 8 games when the best case scenario for winning 8 games has only a 25% chance, and all of the other scenarios are less then 25% chance of occurring.

Again, this is all based off of pre-season FPI being considered relatively accurate, which it's probably not, but the math still stands.
Click to expand...

So you are saying that FPI is useless. It states that Tennessee has a higher than 50% chance of winning 9 games when evaluated opponent to opponent.

Yet you are saying that Tennessee has only a 25% of winning 8 games which implies FPI is wrong and useless in predicting wins and losses to individual games.

Sorry can't have it both ways. You can't use numbers that clearly predict that Tennessee has > 50% of winning a game and turn that into Tennessee losing more than predicted just because you need to prove your point.
 
  • Like
Reactions: 1 person
New posts
You must log in or register to reply here.

VN Store



Back
Top