2018 Preseason Probability Poll Results

coit

Smarter than Diseqc
Joined
Nov 29, 2007
Messages
70,788
Here's the result of the 2018 Probability Polls.

Alcorn : 0.872 W
USF : 0.673 L
Pitt : 0.650 L
Clem : 0.316 L
BG : 0.827 W
Louis : 0.572 W
Duke : 0.665 L
VT : 0.550 W
UNC : 0.712 W
Miami : 0.519 W
UVA : 0.736 W
UGA : 0.423 L

Total: 7.515 wins

Adjusted total thru Game 1: 7.643 wins (+0.128)
Adjusted total thru Game 2: 6.970 wins (-0.673)
Adjusted total thru Game 3: 6.320 wins (-0.650)
Adjusted total thru Game 4: 6.004 wins (-0.316)
Adjusted total thru Game 5: 6.177 wins (+0.173)
Adjusted total thru Game 6: 6.605 wins (+0.428)
Adjusted total thru Game 7: 5.940 wins (-0.665)
Adjusted total thru Game 8: 6.390 wins (+0.450)
Adjusted total thru Game 9: 6.678 wins (+0.288)
Adjusted total thru Game 10: 7.159 wins (+0.481)
Adjusted total thru Game 11: 7.423 wins (+0.264)
Adjusted total thru Game 12: 7.000 wins (-0.423)

Changes in probability from 2017 / 2016 / 2015 repeat opponents:

Pitt: +0.008 / +0.065 / -0.094
Clem: -0.016 / -0.074 / -0.231
Duke: -0.055 / -0.001 / -0.156
VT: -0.046 / +0.007 / -0.109
UNC: +0.088 / +0.175 / -0.036
Miami: +0.081 / +0.057 / -0.120
UVA: -0.006 / +0.047 / -0.069
UGA: -0.193 / -0.049 / -0.203

Average : -0.017 / +0.028 / -0.127
Cumulative: -0.139 / +0.227 / -1.018

So, we are roughly 2% less optimistic this year over last year, but about 3% more optimistic than this time in 2016. Significantly, we are 13% less optimistic than in 2015 after the Orange Bowl high water mark.

Ranking of toughest opponents by probability:

1. Clem 0.316 L
2. UGA 0.423 L
3. Miami 0.519 W
4. VT 0.550 W
5. Louis 0.572 W
6. Pitt 0.650 L
7. Duke 0.665 L
8. USF 0.673 L
9. UNC 0.712 W
10. UVA 0.736 W
11. BG 0.827 W
12. Alcorn 0.872 W

Here's the complete voting spreadsheet. I'll be tracking each voters record through the season if you voted for all 12 games.

 
Last edited:

18in32

Petard Hoister
Joined
May 23, 2010
Messages
24,605
This board thinks there's a 13% chance that Alcorn St wins?
 

coit

Smarter than Diseqc
Joined
Nov 29, 2007
Messages
70,788
This board thinks there's a 13% chance that Alcorn St wins?

Well keep in mind that the maximum possible score is 90%. We've debated making some games higher and such but decided to stick with this. As such, for a gimme type game you can assume most of the 0.9s are really 1.0s and figure that the chance of Alcorn winning is closer to 3%.
 

LagrangeJacket

Dodd-Like
Joined
Oct 22, 2008
Messages
8,984
I apparently don't Stingtalk enough and missed some poles. Can you put me in at 0.9 for Alcorn, 0.7 for Dook, 0.5 for Louisville, and 0.5 for the Dwags?

<3
 

WreckRod

Dodd-Like
Joined
Apr 3, 2012
Messages
3,801
lol there's 250+ people in your sheet and ~ 36 have voted on all 12 games.

Also lots of 11 game voters. Come on people.

What do the results look like if you cut out all of the people who only voted once?
 

JJacket

1st to pass for 2500 and rush for 1000
Joined
May 20, 2003
Messages
59,869
What if you cut out everyone but JJacket?
 

coit

Smarter than Diseqc
Joined
Nov 29, 2007
Messages
70,788
lol there's 250+ people in your sheet and ~ 36 have voted on all 12 games.

Also lots of 11 game voters. Come on people.

What do the results look like if you cut out all of the people who only voted once?

I haven't looked at cutting out single voters but I doubt it would have much of an impact on any single game.



What if you cut out everyone but JJacket?

I did look at cutting out everyone that voted 0.9 each week (you and @whenyousaybud) and those that vote 0.1 each week, and there's really not a significant change in the result.

I think the most useful information is the change in confidence year-over-year against each opponent, and the average changes in confidence. Definitely matches our roller coaster seasons.
 

WreckRod

Dodd-Like
Joined
Apr 3, 2012
Messages
3,801
I haven't looked at cutting out single voters but I doubt it would have much of an impact on any single game.






I did look at cutting out everyone that voted 0.9 each week (you and @whenyousaybud) and those that vote 0.1 each week, and there's really not a significant change in the result.

I think the most useful information is the change in confidence year-over-year against each opponent, and the average changes in confidence. Definitely matches our roller coaster seasons.
lol I think we are all about as bad as the sports writers - last year sucked so we think this year will suck.
 

18in32

Petard Hoister
Joined
May 23, 2010
Messages
24,605
lol I think we are all about as bad as the sports writers - last year sucked so we think this year will suck.
Yeah, there's only two kinds of analysis in CFB... if you are (or are trying to be) emotionally detached, you go by last year's results, modified up or down by 20% for experience, star players, brand name opponents, etc. If you accept your emotional investment, you start at perfection and discount a little as the facts may require.
 

ThisIsAtlanta

Actually Nicolas Cage
Staff member
Joined
Oct 18, 2009
Messages
59,358
Yeah, there's only two kinds of analysis in CFB... if you are (or are trying to be) emotionally detached, you go by last year's results, modified up or down by 20% for experience, star players, brand name opponents, etc. If you accept your emotional investment, you start at perfection and discount a little as the facts may require.
There must be a better way.
 

coit

Smarter than Diseqc
Joined
Nov 29, 2007
Messages
70,788
OP updated for Game 1 results:

Our Game 1 win has increased the expected win total by 0.128. The total expected win value is now 7.643.
 

coit

Smarter than Diseqc
Joined
Nov 29, 2007
Messages
70,788
OP updated for Game 2 results:

Our Game 2 loss has decreased the expected win total by 0.673. The total expected win value is now 6.970.
 
Last edited:

coit

Smarter than Diseqc
Joined
Nov 29, 2007
Messages
70,788
OP updated for Game 3 results:

Our Game 3 loss has decreased the expected win total by 0.650. The total expected win value is now 6.320.

Now that we are in Week 3, I'll post the current prediction leaders. In a couple of weeks I'll add in the modified measures we discussed in the pre-season discussion thread. For the measure below, a vote of higher than 0.5 represents a predicted win, and 0.5 or lower a loss.

We've currently got a set of perfect prognosticators sitting at 3-0.

@GT18YJjr.
@GTBJW
@ee8384
@RamblinWreck92
@Diseqc

Congrats, I guess.
 

JJacket

1st to pass for 2500 and rush for 1000
Joined
May 20, 2003
Messages
59,869
OP updated for Game 3 results:

Our Game 3 loss has decreased the expected win total by 0.650. The total expected win value is now 6.320.

Now that we are in Week 3, I'll post the current prediction leaders. In a couple of weeks I'll add in the modified measures we discussed in the pre-season discussion thread. For the measure below, a vote of higher than 0.5 represents a predicted win, and 0.5 or lower a loss.

We've currently got a set of perfect prognosticators sitting at 3-0.

@GT18YJjr.
@GTBJW
@ee8384
@RamblinWreck92
@Diseqc

Congrats, I guess.
Yea, öööö those guys
 
M

Member 6626

Guest
OP updated for Game 3 results:

Our Game 3 loss has decreased the expected win total by 0.650. The total expected win value is now 6.320.

Now that we are in Week 3, I'll post the current prediction leaders. In a couple of weeks I'll add in the modified measures we discussed in the pre-season discussion thread. For the measure below, a vote of higher than 0.5 represents a predicted win, and 0.5 or lower a loss.

We've currently got a set of perfect prognosticators sitting at 3-0.

@GT18YJjr.
@GTBJW
@ee8384
@RamblinWreck92
@Diseqc

Congrats, I guess.
Do I win anything?
 
Top