2015 Preseason Probability Poll Results

coit

Bullseye
Joined
Nov 29, 2007
Messages
93,399
Some of the polls are closed so I'll start tallying the results here.

Alcorn St: 0.970
Tulane: 0.892
ND: 0.631
Duke: 0.821
UNC: 0.748
Clem: 0.547
Pitt: 0.744
FSU: 0.584
UVA: 0.805
VT: 0.659
Mia: 0.639
Dwags: 0.626

For those teams we played last year, here are the change in probabilities from the 2014 poll to the 2015 poll.

Tulane: +0.054
Duke: +0.077
UNC: +0.067
Clemson: +0.100
Pitt: +0.014
UVA: -0.010
VT: +0.183
Miami: +0.186
Dwags: +0.302

So, a unanimous increase in confidence this year over last. Oddly, Pitt is showing the smallest increase in confidence by a fairly significant amount.

Here's a ranking of our opponents from least to most confident.

1. Clemson
2. FSU
3. UGA
4. ND
5. Miami
6. VT
7. Pitt
8. UNC
9. UVA
10. Duke
11. Tulane
12. Alcorn St

If you remove the 1.0 votes, UGA draws essentially even with Clemson and ahead of FSU.

In looking at the distributions for the polls, there are a inordinate number of 1.0 votes for games like ND, Duke and UGA, and maybe slightly higher 1.0 votes for games like UNC, Clemson, and VT. However, zeroing out the 1.0 votes entirely still doesn't give us a game where we are less than a coin flip to win.

These results aren't going to pass a normality test or anything but fun nonetheless.

Through the first 7 games, we are about 0.136 to be undefeated.
Through the first 9 games, we are about 0.064 to be undefeated.
Through the first 11 games, we are about 0.027 to be undefeated.
Through the regular season, we are about 0.017 to be undefeated.


September 29th Update - Through 4 Games:

Expected wins: 3.314
Actual wins: 2

Win Deviation: -1.314
 
Last edited:
Some of the polls are closed so I'll start tallying the results here.

Alcorn St: 0.970
Tulane: 0.892
ND: 0.631
Duke: 0.821
UNC: 0.748
Clem: 0.547
Pitt: 0.744
FSU: tbd
UVA: tbd
VT: tbd
Mia: tbd
Dwags: tbd


In looking at the distributions for the polls so far, there are a inordinate number of 1.0 votes for games like ND and Duke, and maybe slightly higher 1.0 votes for games like UNC, Clemson, and Pitt. However, zeroing out the 1.0 votes entirely still doesn't give us a game where we are less than a coin flip to win.

These results aren't going to pass a normality test or anything but fun nonetheless.

Through the first 7 games, we are about 0.136 to be undefeated.

Throw out %5 of the top and bottom outliers.
 
This is in effect going into a Catholic church and taking a poll about the distribution of Christians, Muslims, and Jews.
 
This is in effect going into a Catholic church and taking a poll about the distribution of Christians, Muslims, and Jews.

I'd say it's more along the lines of going into a catholic church and taking a poll of the probability that the members of that church would be able to beat the members of another church in a battle to the death.
 
Throw out %5 of the top and bottom outliers.

I did some crunching and threw out everything that was an outside of a two sigma type deviation. The results don't vary that much. I think the undefeated probability goes from 0.136 to 0.123.
 
This is in effect going into a Catholic church and taking a poll about the distribution of Christians, Muslims, and Jews.

I'd say it's more along the lines of going into a catholic church and taking a poll of the probability that the members of that church would be able to beat the members of another church in a battle to the death.


Well the polls last year weren't nearly as optimistic.
 
This is in effect going into a Catholic church and taking a poll about the distribution of Christians, Muslims, and Jews.

Did everyone select 100%?

The better analogy is that we're asking a group of Republicans the likelihood that the Republican nominee will win the 2016 presidential election. It's skewed but directionally correct.

The best way to compare is to look year-over-year if we're getting more or less confident.
 
You don't count your polling when you're in the off season. There'll be time enough for counting, when the polling's done.
 
I've updated the OP with the following:

For those teams we played last year, here are the change in probabilities from the 2014 poll to the 2015 poll.

Tulane: +0.054
Duke: +0.077
UNC: +0.067
Clemson: +0.100
Pitt: +0.014
UVA: tbd
VT: tbd
Miami: tbd
Dwags: tbd

So, a unanimous increase in confidence this year over last so far. Oddly, Pitt is showing the smallest increase in confidence by a fairly significant amount.
 
as far as I remember for those screaming about selection bias and what not:

These Stingtalk polls have been a pretty solid predictor of our wins and losses over the years or at the very least our win/loss total.

Probably because many GT fans enjoy being right more than they enjoy a GT victory.
 
as far as I remember for those screaming about selection bias and what not:

These Stingtalk polls have been a pretty solid predictor of our wins and losses over the years or at the very least our win/loss total.

Probably because many GT fans enjoy being right more than they enjoy a GT victory.

So we'll be undefeated?
 
If I gave you a coin that turned heads 60% of the time, and told you to flip it 12 times, are you of the opinion that it would always turn up heads?

So what you're telling me is that these polls are actually just a super fun way of saying "we might beat Clemson"?
 
Sadly, we will still have to actually play the games.

Sadly? Watching our banners raised over the battlefield is 80% of the reason we follow this religion. I'd rather play against them in their biggest and baddest than to be given a W. Get ur mind rite!
 
Sadly? Watching our banners raised over the battlefield is 80% of the reason we follow this religion. I'd rather play against them in their biggest and baddest than to be given a W. Get ur mind rite!


It was a joke.
 
Back
Top