ESPN FPI - Post Week 2 SEC Projections & Rankings

[navyhog24]

Rank (Based on Game Proj W-L) (Conf Win %)

2. Alabama (12-0) (35.0%)

6. Tennessee (10-2) (31.9%) (L : @ A&M, Bama)

7. LSU (9-3) (10.3%) (L : Wisky, Bama, @ A&M)

11. Texas A&M (10-2) (5.2%) (L : @ AU, @ Bama)

12. Ole Miss (8-4) (4.5%) (L : FSU, Bama, @ LSU, @ A&M)

13. Auburn (8-4) (3.8%) (L : Clemson, LSU, @ OM, @ Bama)

18. Florida (9-3) (6.9%) (L : @ UT, LSU, @ FSU)

31. Georgia (8-4) (1.5%) (L : @ OM, vs UT, vs UF, vs AU)

34. Arkansas (5-7) (0.2%) (L : vs A&M, vs Bama, vs OM, @ AU, vs UF, vs LSU, @ MSU)

39. Mississippi State (5-7) (0.2%) (L : vs USA, @ LSU, vs AU, @ BYU, vs A&M, @ Bama, @ OM)

42. Missouri (5-7) (0.2%) (L : @ WVU, vs UGA, @ LSU, @ UF, @ USCe, @ UT, vs ARK)

49. Vanderbilt (3-9) (0.1%)

54. South Carolina (6-6) (0.1%) (L : MSU, A&M, UGA, UT, UF, Clem)

87. Kentucky (2-10) (0.0%)

http://www.espn.com/college-football/statistics/teamratings

[Hawgar The Horrible]

*Projected results are based on 10,000 pulls from a crack pipe while shooting up heroin, blackouts to date, and remaining life expectancy.

[NuttinItUp]

So, out of our remaining 10 games, they only think we are going to win 4 or 5 of them?  (And that is including Alcorn State and Texas State.)

Sure...

[dhizzle]

Whatever.

[ZERO]

Would anyone here be calling for BB's head if we missed a Bowl in year four? Honest question. What is the bare minimum we could accept for a fifth year? I would think at the very least a Bowl is possible.

[Bacon_Bitz]

You don't just add up the games we are favored to win in to come to a win total.  Even FPI (which has always disliked Arkansas more than S&P+ and FEI for some reason, if I had to guess I would say it's because it relies on recruiting rankings more) predicts the Hogs final win total as 6.7-5.3.

[MissippHog]

So, out of our remaining 10 games, they only think we are going to win 4 or 5 of them?  (And that is including Alcorn State and Texas State.)

Sure...

No, it's saying we only have 3 wins remaining on the season.  Alcorn State, Texas State, and Missouri.

[NuttinItUp]

No, it's saying we only have 3 wins remaining on the season.  Alcorn State, Texas State, and Missouri.

No, it is saying our projected overall record is 6.7-5.3

Like Bacon_Blitz said, you don't just add up the games you are favored in. It is based on probabilities.

[GuvHog]

So, out of our remaining 10 games, they only think we are going to win 4 or 5 of them?  (And that is including Alcorn State and Texas State.)

Sure...

No, they are predicting that the Hogs will win only 3 more games (Texas State, Alcorn State, and Missouri) and go 1-7 in conference play.

That would be an absolute disaster and I don't see it happening.

[NuttinItUp]

No, they are predicting that the Hogs will win only 3 more games (Texas State, Alcorn State, and Missouri) and go 1-7 in conference play.

That would be an absolute disaster and I don't see it happening.

Incorrect. See response above.

[PORKULATOR]

I pray this isn't the case.  That'd be a HUGE step backwards for us.

[GuvHog]

Incorrect. See response above.

They list 7 conference games they predict that the Hogs will lose. That would mean a 5-7 record.

[alohawg]

Let's just hope they're not getting input from SEC officials.

[NuttinItUp]

I don't think some of you guys understand probabilities.

Look at it this way: You have a weighted coin that comes up heads 60% of the time and tails 40% of the time.

If you flip the coin 10 times, do you expect to get heads all ten times and tails 0 times because you have only a 40% chance of "winning" (getting tails) each time? No, you expect to get tails 4 times and heads 6 times.

You can't just take the percentage of each win occurring and say that is an expected loss or an expected win. It is just a probability of each thing occurring.

So, overall, ESPN FPI is predicting we get 6.7 wins and 5.3 losses. (It gives us a 2-0 record, so it is saying we are predicted to have 4.7 wins and 5.3 losses over our remaining 10 games.)

http://www.espn.com/college-football/team/fpi?id=8&year=2016

[DeltaBoy]

They need to put down the bottle.

[Cobbler]

I don't think some of you guys understand probabilities.

Look at it this way: You have a weighted coin that comes up heads 60% of the time and tails 40% of the time.

If you flip the coin 10 times, do you expect to get heads all ten times and tails 0 times because you have only a 40% chance of "winning" (getting tails) each time? No, you expect to get tails 4 times and heads 6 times.

You can't just take the percentage of each win occurring and say that is an expected loss or an expected win. It is just a probability of each thing occurring.

So, overall, ESPN FPI is predicting we get 6.7 wins and 5.3 losses. (It gives us a 2-0 record, so it is saying we are predicted to have 4.7 wins and 5.3 losses over our remaining 10 games.)

http://www.espn.com/college-football/team/fpi?id=8&year=2016

I think both ways are correct. With the probabilities of all games cumulative, our expected win total is 6.7 - 5.3. When you look at each game individually, however, we are projected to lose 7 of the remaining 10 games, hence the 5-7 record that was stated above.

[010HogFan]

Tennessee and LSU at 6 and 7 tells me all I need to know about this measure.

[010HogFan]

and Oklahoma St. at 15 after losing to Central Michigan. Ha!

[Fatty McGee]

and Oklahoma St. at 15 after losing to Central Michigan. Ha!

Give LaTech that much luck and a free play and we might be right there with OSU.

[NuttinItUp]

I think both ways are correct. With the probabilities of all games cumulative, our expected win total is 6.7 - 5.3. When you look at each game individually, however, we are projected to lose 7 of the remaining 10 games, hence the 5-7 record that was stated above.

You can look at it that way if you want, I am just telling you how ESPN expects you to look at it.

For instance, if you look at it game by game like you are saying, it gives a 49.5% chance of beating Florida (almost a 50/50 tie) and only a 17.4% chance of beating Alabama. It doesn't make any sense to count both of those equally as losses. They are meant to be viewed differently which is why they give the probabilities.

[Großer Kriegschwein]

nm

[a0ashle]

You can look at it that way if you want, I am just telling you how ESPN expects you to look at it.

For instance, if you look at it game by game like you are saying, it gives a 49.5% chance of beating Florida (almost a 50/50 tie) and only a 17.4% chance of beating Alabama. It doesn't make any sense to count both of those equally as losses. They are meant to be viewed differently which is why they give the probabilities.

I think I had this same discussion last year. Don't try to explain how they come up with the "odds of going undefeated", it will only be painful.

[Bacon_Bitz]

These computer models are also conveniently self-correcting as the season goes along, and early on in the season rely heavily on last year's ratings combined with some recruiting estimate factored in. After we beat A&M in a couple of weeks all of a sudden we will probably be "favored" in more games. Right now our results from last year are keeping us sort of high while the poor performance against La Tech, relatively low recruiting rank in the SEC, the lack of a returning starter WB (for which FPI gives teams like LSU bonus points), and the low recruiting rank/lack of on-paper returning starters b/c of last season's injuries for TCU are all hurting us in this silly model. But make no mistake - we just gave TCU only their fourth loss in the past three years and the first at home in a long time. That's impressive.

[a0ashle]

These computer models are also conveniently self-correcting as the season goes along, and early on in the season rely heavily on last year's ratings combined with some recruiting estimate factored in. After we beat A&M in a couple of weeks all of a sudden we will probably be "favored" in more games. Right now our results from last year are keeping us sort of high while the poor performance against La Tech, relatively low recruiting rank in the SEC, the lack of a returning starter WB (for which FPI gives teams like LSU bonus points), and the low recruiting rank/lack of on-paper returning starters b/c of last season's injuries for TCU are all hurting us in this silly model. But make no mistake - we just gave TCU only their fourth loss in the past three years and the first at home in a long time. That's impressive.

I've also noticed a few drops in our numbers due to our high output games from last year rolling out of the data.

[NuttinItUp]

I think I had this same discussion last year. Don't try to explain how they come up with the "odds of going undefeated", it will only be painful.

Yeah, that I am not sure about.

I do know the odds change after every game. So, if you keep winning (especially if your measurable stats look good) then the odds will improve, or vice-versa.

They claim that it is almost as accurate as Vegas lines. Of that, I am not sure.

[a0ashle]

Yeah, that I am not sure about.

I do know the odds change after every game. So, if you keep winning (especially if your measurable stats look good) then the odds will improve, or vice-versa.

Multiply all the %s together.
If you have 2 games left, one 10% and one 50%, you would have a 5% chance of winning out.