#1
|
|||
|
|||
CFB Project for the Forum
I was wondering which is harder, to get through an SEC schedule with one loss, or to get through Louisville's schedule undefeated. Florida is the obvious choice since they control their own destiny, where Auburn and Tennessee do not. I think this would be an interesting project.
1. Ask MyTurn2Raise to give us an estimation on what the lines would be for each of Lousiville's and Florida's games. Base the lines on what we know about the teams now (don't use UL -4 vs. Miami, use what the line would have been knowing what we know now.) MyTurn said that UL vs. UF would be pick'em so the lines should be about the same for either team. Use Arkansas as the SECWest champ. 2. Someone convert these lines to percentages. I'm sure there is a website that can do this fairly easily. I know this would not be exact, but close enough. 3. Someone with math skills determine the difficulty of getting through 'Ville's schedule undefeated and UF's schedule with 0 or 1 loss. 4. Repeat with USC and/or Michigan if interested. Is this possible, or is there a factor I'm not considering? Is this interesting to anyone else or am I just a CFB nerd? |
#2
|
|||
|
|||
Re: CFB Project for the Forum
sounds interesting and i dont even follow college football much.
|
#3
|
|||
|
|||
Re: CFB Project for the Forum
did someone just say playoff system...or am i hearing things?
|
#4
|
|||
|
|||
Re: CFB Project for the Forum
is there some way to estimate a team's probability of winning from a computer rating? That would make it pretty easy to do, I think.
|
#5
|
|||
|
|||
Re: CFB Project for the Forum
Give me 20 minutes, it should be pretty easy to code up and run a simulation using Sagarin's predictor ratings.
|
#6
|
|||
|
|||
Re: CFB Project for the Forum
First, here's my approach. I used Sagarin to find the pointspread for each game, then used this to convert the spreads to moneylines. For those outside of the chart, I simply estimated. It shouldn't be too much error, since the difference between 95 and 97 is fairly negligible. Anyways, I then generated a random number for each game which determined the outcome. I ran 100000 seasons, and then found which percent of the time each team was undefeated or one loss. So, here are the results, you'll probably expect them.
Percent Florida undefeat: 3.7100 Percent Florida one loss: 16.3240 Percent Louisville undefeat: 8.9220 Percent Louisville one loss: 28.7220 Some notes: Sagarin hasn't been updated yet, so Louisville will probably win more once their win over WVU is taken into account. Also, I neglected juice, assuming that the moneyline was a true representation of the win/loss percentage of the game. |
#7
|
|||
|
|||
Re: CFB Project for the Forum
awesome....lots of error in this but a good estimate nonetheless
can you do it for USC/Michigan/tOSU/ND/etc as well |
#8
|
|||
|
|||
Re: CFB Project for the Forum
Dudd- you are my hero. Very cool. Thanks for doing that.
Could you post the point spreads? |
#9
|
|||
|
|||
Re: CFB Project for the Forum
It's a decent amount of work, looking at schedule, figuring out spreads, so I'm going to stop with Florida and Louisville. If anyone wants to do it and has a copy of Matlab, just PM me and I'll send you the file, it's really straightforward, just a lot of calculations.
|
#10
|
|||
|
|||
Re: CFB Project for the Forum
Here's the spreads I used. First is Florida, then Louisville. The current Sagarin predictor is given for each team, then an approximate spread is given, followed by the win percentage. I used a value of 3 points for home field, so that's why it's not just a straight subtraction of the two ratings.
%Florida=87.92 %game 1: SMiss=66.59 %spread=24.5, 98 %game 2: UCF=57.95 %spread=33, 99 %game 3: Tenn=85.11 %spread=0, 50 %game 4: Kent=68.55 %spread=22.5, 97 %game 5: Bama=77.45 %spread=13.5, 85.3 %game 6: LSU=92.15 %spread=+1, 47.6 %game 7: Auburn=84.87 %spread=0, 50 %game 8: Georgia=75.02 %spread=16, 88.2 %game 9: Vandy=71.18 %spread=14, 86.9 %game 10: SC=76.68 %spread=14, 86.9 %game 11: West Carolina=39.91 %spread=51, 100 %game 12: FSU=78.64 %spread=6.5, 73 %game 13: SEC Champ Arkansas=77.91 %spread=10, 80 %Louisville=88.97 %game 1: Kent=68.5 %spread=23.5, 97.5 %game 2: Temple=46.32 %spread=40, 100 %game 3: Miami=72.65 %spread=19.5, 90 %game 4: KState=71.08 %spread=15, 87.5 %game 5: Middle Tenn=64.2 %spread=22, 96.5 %game 6: Cincinnati=74.41 %spread=17.5, 89 %game 7: Syracuse=73.24 %spread=13, 84.6 %game 8: WVU=90.96 %spread=1, 52.4 %game 9: Rutgers=83.26 %spread=3, 60.8 %game 10: South Fla=69.4 %spread=23, 97 %game 11: Pitt=87.12 %spread=+1, 47.6 %game 12: UConn=67.57 %spread=24.5, 98 |
|
|