Page 1 of 2
Playoff Log5 Probabilities
Posted: Tue Nov 26, 2013 2:45 pm
by Adam H
I thought some people might be interested in this.
The "winpct" column is the estimated probability that the team would beat an "average" team on a neutral field. It's derived from my own power ratings which also include FBS teams. I'm sure you'll find some of the team ratings a bit wonky, but overall they're pretty similar to SRS, Sagarin, etc.
The other columns show the probability that the team will make it through that round.
Oh yeah, and home field advantage isn't taken into account, so basically what I'm saying is that this is all completely worthless. But anyways...
Code: Select all
winpct rd1 rd2 rd3 rd4 rd5
Eastern Illinois 74.3% 100.0% 81.5% 57.3% 40.0% 24.4%
North Dakota St 74.6% 100.0% 79.7% 55.7% 38.2% 23.7%
Southeastern LA 65.0% 100.0% 71.1% 45.9% 22.3% 11.5%
Towson 59.9% 100.0% 68.1% 27.5% 15.0% 6.8%
Montana 59.5% 100.0% 59.9% 23.9% 12.6% 5.8%
Eastern Washington 58.3% 100.0% 54.2% 30.2% 12.7% 5.5%
McNeese St 55.8% 100.0% 57.4% 28.1% 11.1% 4.6%
Maine 53.0% 100.0% 57.7% 25.3% 9.4% 3.7%
South Dakota St 59.3% 64.3% 32.8% 18.6% 7.9% 3.6%
New Hampshire 50.5% 65.1% 30.9% 12.8% 4.5% 1.7%
Coastal Carolina 53.1% 58.6% 25.5% 8.8% 4.1% 1.6%
Jacksonville St 49.6% 52.7% 23.1% 9.9% 3.4% 1.2%
Sam Houston St 47.7% 63.4% 20.9% 9.9% 3.2% 1.1%
Fordham 45.8% 63.5% 22.9% 6.6% 2.7% 0.9%
Samford 46.9% 47.3% 19.5% 7.8% 2.5% 0.8%
Tennessee St. 44.5% 68.3% 14.8% 6.0% 2.4% 0.7%
South Carolina St 44.7% 54.6% 11.8% 4.6% 1.8% 0.6%
Bethune-Cookman 44.4% 41.4% 14.6% 4.0% 1.5% 0.5%
Northern Arizona 44.7% 35.7% 13.1% 5.5% 1.7% 0.5%
Furman 40.2% 45.4% 8.5% 3.0% 1.0% 0.3%
Lafayette 35.3% 34.9% 11.4% 3.2% 0.7% 0.2%
Southern Utah 34.5% 36.6% 8.1% 2.8% 0.6% 0.1%
Sacred Heart 32.7% 36.5% 9.0% 1.7% 0.5% 0.1%
Butler 27.1% 31.7% 3.6% 0.9% 0.2% 0.04%
I'll come update this thread after the results of every round, at least if I can remember.
Re: Playoff Log5 Probabilities
Posted: Tue Nov 26, 2013 3:10 pm
by AZGrizFan
Sorry....nice attempt and all, but ANY poll, ANYWHERE, using ANY data that puts an OVC team at the top of the heap has serious issues that should be addressed before being made public.
Re: Playoff Log5 Probabilities
Posted: Tue Nov 26, 2013 3:28 pm
by 89Hen
Adam H wrote:Butler 0.04%
Re: Playoff Log5 Probabilities
Posted: Tue Nov 26, 2013 3:29 pm
by 89Hen
Adam H wrote:Butler 0.04%
Re: Playoff Log5 Probabilities
Posted: Tue Nov 26, 2013 5:00 pm
by jcfrancis
So, because an OVC team is on top of the list, it is automatically worthless? So, I guess the fact that EIU is the #2 overall seed in the playoff is wrong. But I guess that being 4 points away from being undefeated like NDSU doesn't matter. Or the fact that they are the top offense in the FCS is not of note. Oh, and while they are not the top defense in the FCS, they are a top 25 (top 10 in the playoff field). Should we mention they beat a top 5 defense this year as well? Yes, I am well aware of the fact that the OVC has been horrible when it comes to the playoffs. But looking at the past without a serious look at the present is not wise. I guarantee you that no team in the playoff field will overlook EIU.
Re: Playoff Log5 Probabilities
Posted: Tue Nov 26, 2013 5:39 pm
by JohnStOnge
Why do you call it "Log5?"
Re: Playoff Log5 Probabilities
Posted: Tue Nov 26, 2013 6:12 pm
by AZGrizFan
jcfrancis wrote:So, because an OVC team is on top of the list, it is automatically worthless? So, I guess the fact that EIU is the #2 overall seed in the playoff is wrong. But I guess that being 4 points away from being undefeated like NDSU doesn't matter. Or the fact that they are the top offense in the FCS is not of note. Oh, and while they are not the top defense in the FCS, they are a top 25 (top 10 in the playoff field). Should we mention they beat a top 5 defense this year as well? Yes, I am well aware of the fact that the OVC has been horrible when it comes to the playoffs. But looking at the past without a serious look at the present is not wise. I guarantee you that no team in the playoff field will overlook EIU.
OVC teams are something like 0-50 in the last 50 years (OK, I made that stat up, but you get my drift). When one team, ANY team from the OVC wins more than one playoff game, call me. Until then, I seriously question the integrity of your computerized model.
Re: Playoff Log5 Probabilities
Posted: Tue Nov 26, 2013 6:35 pm
by jcfrancis
As stated before, I am fully aware of the failures of past OVC teams, and understand why anyone would be skeptical. However, to negate the validity of any poll, survey, or odd-maker because it does not reflect the past in one's eye is not wise. Why have a playoff then? NDSU is undefeated and won two straight titles. Just give it to them.
Re: Playoff Log5 Probabilities
Posted: Tue Nov 26, 2013 9:20 pm
by AZGrizFan
jcfrancis wrote:As stated before, I am fully aware of the failures of past OVC teams, and understand why anyone would be skeptical. However, to negate the validity of any poll, survey, or odd-maker because it does not reflect the past in one's eye is not wise. Why have a playoff then? NDSU is undefeated and won two straight titles. Just give it to them.
Many, many programs in the playoffs have won games, multiple games, and championships. OVC? Not so much. I have no problem with EIU being "near" the top...but AT the top? Nope...
Re: Playoff Log5 Probabilities
Posted: Tue Nov 26, 2013 9:55 pm
by grizzaholic
jcfrancis wrote:As stated before, I am fully aware of the failures of past OVC teams, and understand why anyone would be skeptical. However, to negate the validity of any poll, survey, or odd-maker because it does not reflect the past in one's eye is not wise. Why have a playoff then? NDSU is undefeated and won two straight titles. Just give it to them.
Why not. Their fans have already ordered their three-peat T-shirts and are planning their parties.
EARLY RUMOR ALERT: NDSU will be playing the Super Bowl Champion the following week for best football team in the Universe.
Re: Playoff Log5 Probabilities
Posted: Wed Nov 27, 2013 12:14 am
by craneac27
Re: Playoff Log5 Probabilities
Posted: Wed Nov 27, 2013 5:51 am
by Bison Fan in NW MN
grizzaholic wrote:jcfrancis wrote:As stated before, I am fully aware of the failures of past OVC teams, and understand why anyone would be skeptical. However, to negate the validity of any poll, survey, or odd-maker because it does not reflect the past in one's eye is not wise. Why have a playoff then? NDSU is undefeated and won two straight titles. Just give it to them.
Why not. Their fans have already ordered their three-peat T-shirts and are planning their parties.
EARLY RUMOR ALERT: NDSU will be playing the Super Bowl Champion the following week for best football team in the Universe.
No they haven't because they realize now that there is no way NDSU is going to get that 3-peat because SUU is now in the field....
Re: Playoff Log5 Probabilities
Posted: Wed Nov 27, 2013 6:22 am
by Gil Dobie
jcfrancis wrote:So, because an OVC team is on top of the list, it is automatically worthless? So, I guess the fact that EIU is the #2 overall seed in the playoff is wrong. But I guess that being 4 points away from being undefeated like NDSU doesn't matter. Or the fact that they are the top offense in the FCS is not of note. Oh, and while they are not the top defense in the FCS, they are a top 25 (top 10 in the playoff field). Should we mention they beat a top 5 defense this year as well? Yes, I am well aware of the fact that the OVC has been horrible when it comes to the playoffs. But looking at the past without a serious look at the present is not wise. I guarantee you that no team in the playoff field will overlook EIU.
EIU beating SIU 40-37 in 2OT and NDSU beating SIU 31-10. Last years 58-10 loss to SDSU doesn't look good either. EIU might have trouble giving up the run. I see SIU had over 200 yards rushing, NDSU gave up 46 yards to the same SIU team. And for arguments sake, NIU had a tough time with Idaho too.
Re: Playoff Log5 Probabilities
Posted: Wed Nov 27, 2013 6:27 am
by JohnStOnge
This is a good place to talk about something. I don't know what the model is but it's a model. There are "margins of error" around each probability estimate so it's really not possible to separate teams so precisely as 1 vs. 2, 2 vs. 3, etc. It's kind of like that "margin of error" thing you see on political poll results. When the difference between the two candidates is within the "margin of error," which is probably a 95 percent confidence interval, the talking heads say it's a "statistical tie."
What it would basically tell you if you just woke up from a coma and didn't know anything at all about the season that just transpired is that Eastern Illinois is very good. It can give you an idea as to who the better teams in FCS are and who the worst ones are. Like if you compare Eastern Illinois to Butler you can be pretty confident that Eastern Illinois is better. The "margins of error" around each probability estimate are probably mutually exclusive (do not overlap) and there's even probably a good bit of separation between the upper bound of the confidence interval for Butler and the lower bound for Eastern Illinois. But if you're trying to compare Eastern Illinois and North Dakota State it's going to be one of those "statistical tie" things.
It's like that with all of the football models (power rating systems). But I don't worry about that because there's at least as much and probably more error in human polls. And human brains are not capable of processing all of the results comprehensively and simultaneously. There's a certain advantage to human polls in that people can watch games and see things that aren't captured by the scores. But there are questions about objectivity and also nobody sees anything close to all the games anyway. Chances are most of the people voting in FCS polls have never seen Eastern Illinois play and if they have they haven't seen much of it. So I don't think the advantages of the human thought process outweigh the advantage the models have in being able to "see" everything comprehensively, simultaneously, and objectively.
Of course it's got to be a good model and there are ways of testing that.
Re: Playoff Log5 Probabilities
Posted: Wed Nov 27, 2013 6:44 am
by grizzaholic
Bison Fan in NW MN wrote:grizzaholic wrote:
Why not. Their fans have already ordered their three-peat T-shirts and are planning their parties.
EARLY RUMOR ALERT: NDSU will be playing the Super Bowl Champion the following week for best football team in the Universe.
No they haven't because they realize now that there is no way NDSU is going to get that 3-peat because SUU is now in the field....
Damn straight Skippy!
Re: Playoff Log5 Probabilities
Posted: Wed Nov 27, 2013 6:57 am
by grizinidaho
I was really hoping the Griz would line up against SELA or EIU. I know they are both good teams, but I'd much rather play them than NDSU, at least early.
Re: Playoff Log5 Probabilities
Posted: Wed Nov 27, 2013 8:07 am
by Adam H
JohnStOnge wrote:Why do you call it "Log5?"
I have no idea!
http://en.wikipedia.org/wiki/Log5
AZGrizFan wrote:When one team, ANY team from the OVC wins more than one playoff game, call me. Until then, I seriously question the integrity of your computerized model.
If Tennessee St beats EIU in the second round would that count?
Also, if it makes you feel better, the model does say that NDSU is a bit better than EIU. NDSU just has a little tougher path to the championship so their odds of winning are predicted to be lower than EIU.
Re: Playoff Log5 Probabilities
Posted: Wed Nov 27, 2013 1:14 pm
by jcfrancis
Tougher path? EIU potentially has TSU with their team healthier than last game, Towson, and EWU. I am not saying NDSU has an easy path, as I don't think there is an easy path. There will be some great match ups.
Don't get me wrong. I am. to saying that EIU is the best team. I do think they are a great team, but there are tough match ups across the board. The teams that get to Frisco earn it.
Re: Playoff Log5 Probabilities
Posted: Mon Dec 02, 2013 12:01 pm
by Adam H
Updated after first round:
Code: Select all
winpct rd2 rd3 rd4 rd5
North Dakota St 74.7% 81.0% 55.8% 38.0% 23.5%
Eastern Illinois 74.7% 76.7% 53.9% 36.8% 22.5%
Southeastern LA 65.5% 64.9% 41.0% 20.0% 10.3%
Towson 60.3% 64.8% 26.4% 13.9% 6.3%
South Dakota St 60.3% 52.5% 28.5% 12.6% 5.7%
Montana 58.7% 53.5% 21.0% 10.7% 4.8%
Eastern Washington 57.9% 47.5% 24.7% 10.3% 4.4%
McNeese St 56.7% 51.4% 24.4% 9.9% 4.2%
Jacksonville St 55.3% 48.6% 22.4% 8.9% 3.6%
New Hampshire 53.8% 52.7% 22.8% 8.5% 3.4%
Coastal Carolina 55.3% 46.5% 16.8% 8.0% 3.3%
Maine 51.1% 47.3% 19.2% 6.7% 2.5%
Sam Houston St 50.7% 35.1% 16.9% 5.9% 2.2%
Tennessee St. 47.3% 23.3% 9.9% 3.9% 1.3%
Fordham 45.1% 35.2% 9.8% 3.7% 1.2%
Furman 40.8% 19.0% 6.5% 2.2% 0.6%
After a mostly upset-free first round, the teams with the first round bye now all have a tougher path to the championship than expected, while the winners from this past Saturday basically double their chances of winning the championship.
jcfrancis wrote:Tougher path? EIU potentially has TSU with their team healthier than last game, Towson, and EWU. I am not saying NDSU has an easy path, as I don't think there is an easy path. There will be some great match ups.
Don't get me wrong. I am. to saying that EIU is the best team. I do think they are a great team, but there are tough match ups across the board. The teams that get to Frisco earn it.
I totally agree with the last part. The reason that NDSU's expected path of Montana and Southeastern LA looked a shade harder than EIU's Towson and [South Dakota St / Eastern Washington / McNeese St / Jacksonville St] is because my ratings disagree with the polls. I won't argue that my ratings are superior to the polls or anything like that, but most objective algorithms rank Eastern Washington lower than Southeastern LA (for example).
Re: Playoff Log5 Probabilities
Posted: Thu Dec 05, 2013 7:22 pm
by The Kicker
Adam H wrote:Updated after first round:
Code: Select all
winpct rd2 rd3 rd4 rd5
North Dakota St 74.7% 81.0% 55.8% 38.0% 23.5%
Eastern Illinois 74.7% 76.7% 53.9% 36.8% 22.5%
Southeastern LA 65.5% 64.9% 41.0% 20.0% 10.3%
Towson 60.3% 64.8% 26.4% 13.9% 6.3%
South Dakota St 60.3% 52.5% 28.5% 12.6% 5.7%
Montana 58.7% 53.5% 21.0% 10.7% 4.8%
Eastern Washington 57.9% 47.5% 24.7% 10.3% 4.4%
McNeese St 56.7% 51.4% 24.4% 9.9% 4.2%
Jacksonville St 55.3% 48.6% 22.4% 8.9% 3.6%
New Hampshire 53.8% 52.7% 22.8% 8.5% 3.4%
Coastal Carolina 55.3% 46.5% 16.8% 8.0% 3.3%
Maine 51.1% 47.3% 19.2% 6.7% 2.5%
Sam Houston St 50.7% 35.1% 16.9% 5.9% 2.2%
Tennessee St. 47.3% 23.3% 9.9% 3.9% 1.3%
Fordham 45.1% 35.2% 9.8% 3.7% 1.2%
Furman 40.8% 19.0% 6.5% 2.2% 0.6%
After a mostly upset-free first round, the teams with the first round bye now all have a tougher path to the championship than expected, while the winners from this past Saturday basically double their chances of winning the championship.
jcfrancis wrote:Tougher path? EIU potentially has TSU with their team healthier than last game, Towson, and EWU. I am not saying NDSU has an easy path, as I don't think there is an easy path. There will be some great match ups.
Don't get me wrong. I am. to saying that EIU is the best team. I do think they are a great team, but there are tough match ups across the board. The teams that get to Frisco earn it.
I totally agree with the last part. The reason that NDSU's expected path of Montana and Southeastern LA looked a shade harder than EIU's Towson and [South Dakota St / Eastern Washington / McNeese St / Jacksonville St] is because my ratings disagree with the polls. I won't argue that my ratings are superior to the polls or anything like that, but most objective algorithms rank Eastern Washington lower than Southeastern LA (for example).
Keep posting you were 7 of 8 in the first round. You have 4 close games this week, lets see if SDSU, New Hampshire, Montana and McNeese St. all pull out those closer ones.
Re: Playoff Log5 Probabilities
Posted: Mon Dec 09, 2013 9:53 am
by Adam H
AZGrizFan wrote:When one team, ANY team from the OVC wins more than one playoff game, call me.
What's your number?
Re: Playoff Log5 Probabilities
Posted: Mon Dec 09, 2013 9:55 am
by kalm
Adam H wrote:AZGrizFan wrote:When one team, ANY team from the OVC wins more than one playoff game, call me.
What's your number?
Re: Playoff Log5 Probabilities
Posted: Mon Dec 09, 2013 9:57 am
by Ibanez
Adam H wrote:Updated after first round:
Code: Select all
winpct rd2 rd3 rd4 rd5
North Dakota St 74.7% 81.0% 55.8% 38.0% 23.5%
Eastern Illinois 74.7% 76.7% 53.9% 36.8% 22.5%
Southeastern LA 65.5% 64.9% 41.0% 20.0% 10.3%
Towson 60.3% 64.8% 26.4% 13.9% 6.3%
South Dakota St 60.3% 52.5% 28.5% 12.6% 5.7%
Montana 58.7% 53.5% 21.0% 10.7% 4.8%
Eastern Washington 57.9% 47.5% 24.7% 10.3% 4.4%
McNeese St 56.7% 51.4% 24.4% 9.9% 4.2%
Jacksonville St 55.3% 48.6% 22.4% 8.9% 3.6%
New Hampshire 53.8% 52.7% 22.8% 8.5% 3.4%
Coastal Carolina 55.3% 46.5% 16.8% 8.0% 3.3%
Maine 51.1% 47.3% 19.2% 6.7% 2.5%
Sam Houston St 50.7% 35.1% 16.9% 5.9% 2.2%
Tennessee St. 47.3% 23.3% 9.9% 3.9% 1.3%
Fordham 45.1% 35.2% 9.8% 3.7% 1.2%
Furman 40.8% 19.0% 6.5% 2.2% 0.6%
After a mostly upset-free first round, the teams with the first round bye now all have a tougher path to the championship than expected, while the winners from this past Saturday basically double their chances of winning the championship.
jcfrancis wrote:Tougher path? EIU potentially has TSU with their team healthier than last game, Towson, and EWU. I am not saying NDSU has an easy path, as I don't think there is an easy path. There will be some great match ups.
Don't get me wrong. I am. to saying that EIU is the best team. I do think they are a great team, but there are tough match ups across the board. The teams that get to Frisco earn it.
I totally agree with the last part. The reason that NDSU's expected path of Montana and Southeastern LA looked a shade harder than EIU's Towson and [South Dakota St / Eastern Washington / McNeese St / Jacksonville St] is because my ratings disagree with the polls. I won't argue that my ratings are superior to the polls or anything like that, but most objective algorithms rank Eastern Washington lower than Southeastern LA (for example).
Re: Playoff Log5 Probabilities
Posted: Mon Dec 09, 2013 10:50 am
by Adam H
After second round:
Code: Select all
winpct rd3 rd4 rd5
North Dakota St 74.5% 69.5% 45.6% 26.7%
Eastern Illinois 75.2% 65.1% 44.0% 26.3%
Southeastern LA 63.9% 58.8% 25.8% 12.0%
Eastern Washington 60.7% 53.2% 20.7% 9.0%
Towson 61.9% 34.9% 18.4% 8.2%
Jacksonville St 57.5% 46.8% 16.8% 6.8%
New Hampshire 55.3% 41.2% 14.7% 5.6%
Coastal Carolina 56.2% 30.5% 13.9% 5.4%
Eastern Washington made the biggest jump, from 4.4% chance to win the championship to 9.0%. That was a tough SDSU team they just dominated, and their winpct went up accordingly. SELA didn't impress in their win over SHSU, so while they were once the clear cut #3 team, EWU and Towson have basically caught up with them.
Remember that this doesn't take into account home field advantage, so the home teams are more likely to win than these percentages indicate. Especially EWU on that nice red field. Still, I wouldn't be that shocked if JSU pulled off the upset.
Re: Playoff Log5 Probabilities
Posted: Mon Dec 09, 2013 12:58 pm
by AZGrizFan
kalm wrote:Adam H wrote:
What's your number?
Like Haley's Comet...once every 76 years.
