538.com has a new chart, predicting the chance of each of the ranked teams (and more) to win out, and their chances of being selected for the playoff if they do.

The most noteworthy part of their prediction is, basically, more upsets will happen: None of the top teams have even a 50% chance of winning out from here through Championship Saturday. They have Clemson at a 49% chance of winning out, Notre Dame at 38%, Ohio St. at 27%, Alabama at 25%, Baylor and Okla. St. at 16% each.

No surprise that they predict that if those teams do win out, they're locks for the playoff -- except that they give Notre Dame only a 74% chance of being selected even if they win out.

http://fivethirtyeight.com/features/the-...s-are-too/

26% chance that ND is out at 11-1, because there is a less than 26% chance that the Big 12 goes unbeaten I feel.

I think this is the one thing people so easily miss in the sport. Even if you are an 80% favorite to win 3 games in a row, it is almost a 50/50 shot you lose at least one (just under 49% actually). People look through schedules and say well we're better than x, y, and z so we should be 3-0 after that. What they are missing is that it's not about just being better. On any given Saturday either team can win, it's just a matter of how much has to go right for the underdog and/or wrong for the favorite. Even if you'd beat a team 8 out of 10 (or whatever), there are 2 times you would have lost and the odds of winning all the match-ups like that are slim.

Put another way: Even if you are a great team, odds are you will lose to someone not quite as good as you at some point.

All of this is of course magnifcied when you start playing good teams whom you might only beat 60% of the time. If Ohio State has a 60% chance of beating Michigan State, Michigan, and (with those two wins) Iowa in the Big Ten Championship, then the odds of them winning all 3 is 21.6%.

Just for fun.

The numbers themselves are unverifiable and thus meaningless.


But no harm, so long as people understand that the numbers are meaningless.

I really don't think Clemson is going to lose....Syracuse, Wake Forest, then South Carolina, and then most likely North Carolina. NC could be tricky, but I think this is Clemson's year

Notre Dame has that date with Stanford...I'm starting to like the Cardinal Tree in this matchup

Alabama has a trap game, a rivalry game, and if they win those, a conference championship vs. Florida...who's going to show up? the team that destroyed LSU and Georgia, or the team that barely beat Tennessee and lost to Ole Miss??

Ohio State has Michigan State, Michigan, and a possible Big 10 Championship all back-to-back....are they really going to get through that unscathed?

There's still A LOT of football left to be played...

(11-12-2015 03:23 PM)ohio1317 Wrote:  I think this is the one thing people so easily miss in the sport. Even if you are an 80% favorite to win 3 games in a row, it is almost a 50/50 shot you lose at least one (just under 49% actually). People look through schedules and say well we're better than x, y, and z so we should be 3-0 after that. What they are missing is that it's not about just being better. On any given Saturday either team can win, it's just a matter of how much has to go right for the underdog and/or wrong for the favorite. Even if you'd beat a team 8 out of 10 (or whatever), there are 2 times you would have lost and the odds of winning all the match-ups like that are slim.

Put another way: Even if you are a great team, odds are you will lose to someone not quite as good as you at some point.

All of this is of course magnifcied when you start playing good teams whom you might only beat 60% of the time. If Ohio State has a 60% chance of beating Michigan State, Michigan, and (with those two wins) Iowa in the Big Ten Championship, then the odds of them winning all 3 is 21.6%.

Put another way:

- if the (supposed) probability of beating three teams is X

- then the average (supposed) probability of beating each team is X^(1/3)


Example: if the (supposed) probability of beating three teams is 80%, then the average (supposed) probability of beating each team has to be about 93%.

(11-12-2015 03:16 PM)RUScarlets Wrote:  26% chance that ND is out at 11-1, because there is a less than 26% chance that the Big 12 goes unbeaten I feel.

Yeah, the bulk of that 26% is the possibility there would be undefeated champs in all of the ACC, Big 12, and Big Ten and that all 3 of those plus Bama/Florida would be the playoff teams. The rest is the possibility that ND would be passed over in favor of a one-loss conference champ: Utah, Mich. St. or Oklahoma if they win out, or Clemson, Ohio St., or Iowa if they win their conference at 12-1.

(11-12-2015 03:23 PM)ohio1317 Wrote:  I think this is the one thing people so easily miss in the sport. Even if you are an 80% favorite to win 3 games in a row, it is almost a 50/50 shot you lose at least one (just under 49% actually). People look through schedules and say well we're better than x, y, and z so we should be 3-0 after that. What they are missing is that it's not about just being better. On any given Saturday either team can win, it's just a matter of how much has to go right for the underdog and/or wrong for the favorite. Even if you'd beat a team 8 out of 10 (or whatever), there are 2 times you would have lost and the odds of winning all the match-ups like that are slim.

Put another way: Even if you are a great team, odds are you will lose to someone not quite as good as you at some point.

Exactly right. Every year, early in the season, people conjure up these "OMG what if there are 8 undefeated teams this year" scenarios. Never happens.

Plus, you do have schools that will be playing much harder to win to be bowl eligible like South Carolina, Syracuse, and so forth. You do see upsets.

Iowa is in danger of fallen this week to Minnesota because the Gophers are trying to become bowl eligible.

(11-12-2015 03:05 PM)Wedge Wrote:  538.com has a new chart, predicting the chance of each of the ranked teams (and more) to win out, and their chances of being selected for the playoff if they do.

The most noteworthy part of their prediction is, basically, more upsets will happen: None of the top teams have even a 50% chance of winning out from here through Championship Saturday. They have Clemson at a 49% chance of winning out, Notre Dame at 38%, Ohio St. at 27%, Alabama at 25%, Baylor and Okla. St. at 16% each.

No surprise that they predict that if those teams do win out, they're locks for the playoff -- except that they give Notre Dame only a 74% chance of being selected even if they win out.

http://fivethirtyeight.com/features/the-...s-are-too/

The make playoff if win out number (last columns) is the most interesting, at least for the Notre Dame discussions. 74% chance of making playoff if they win out. 10 teams have a higher % chance, including OU at 87%. Baylor and Oklahoma State 99 and 98%. Goes against the CW on this board.

(11-12-2015 04:28 PM)Frog in the Kitchen Sink Wrote:  

(11-12-2015 03:05 PM)Wedge Wrote:  538.com has a new chart, predicting the chance of each of the ranked teams (and more) to win out, and their chances of being selected for the playoff if they do.

The most noteworthy part of their prediction is, basically, more upsets will happen: None of the top teams have even a 50% chance of winning out from here through Championship Saturday. They have Clemson at a 49% chance of winning out, Notre Dame at 38%, Ohio St. at 27%, Alabama at 25%, Baylor and Okla. St. at 16% each.

No surprise that they predict that if those teams do win out, they're locks for the playoff -- except that they give Notre Dame only a 74% chance of being selected even if they win out.

http://fivethirtyeight.com/features/the-...s-are-too/

The make playoff if win out number (last columns) is the most interesting, at least for the Notre Dame discussions. 74% chance of making playoff if they win out. 10 teams have a higher % chance, including OU at 87%. Baylor and Oklahoma State 99 and 98%. Goes against the CW on this board.

But it's just for fun, because the numbers are made up.

So it's not really disproving any wisdom, conventional or not.

(11-12-2015 03:23 PM)ohio1317 Wrote:  I think this is the one thing people so easily miss in the sport. Even if you are an 80% favorite to win 3 games in a row, it is almost a 50/50 shot you lose at least one (just under 49% actually). People look through schedules and say well we're better than x, y, and z so we should be 3-0 after that. What they are missing is that it's not about just being better. On any given Saturday either team can win, it's just a matter of how much has to go right for the underdog and/or wrong for the favorite. Even if you'd beat a team 8 out of 10 (or whatever), there are 2 times you would have lost and the odds of winning all the match-ups like that are slim.

Put another way: Even if you are a great team, odds are you will lose to someone not quite as good as you at some point.

All of this is of course magnifcied when you start playing good teams whom you might only beat 60% of the time. If Ohio State has a 60% chance of beating Michigan State, Michigan, and (with those two wins) Iowa in the Big Ten Championship, then the odds of them winning all 3 is 21.6%.

Exactly. The chances that all 5 of the top teams in the 538 link all winning out (Clemson, OSU, Alabama, Baylor and Notre Dame) is .002% or 1 in 500.

I dont know id say made up. They are certainly are making educated guesses about the odds of winning each game and the probability of who is selected in diffferent senarios but there is a lot more here than typical articles/talk.

(11-12-2015 05:59 PM)ohio1317 Wrote:  I dont know id say made up. They are certainly are making educated guesses about the odds of winning each game and the probability of who is selected in diffferent senarios but there is a lot more here than typical articles/talk.

I'm not suggesting they threw darts at a dartboard. No.

But the point is that no matter how intelligent the algorithm seems to be, you can never verify the number it spits out.

So in that case, what's the point of such a specific number at all?

(11-12-2015 04:24 PM)DavidSt Wrote:  Plus, you do have schools that will be playing much harder to win to be bowl eligible like South Carolina, Syracuse, and so forth. You do see upsets.

Iowa is in danger of fallen this week to Minnesota because the Gophers are trying to become bowl eligible.

Iowa is in danger of losing this week to Minnesota because it is Iowa vs Minnesota more than because of Minnesota trying to become Bowl Eligible. Minny does NOT want Iowa to go undefeated.

I agree getting actual numbers is very difficult. I suppose you might devise something from vegas odds, but regardless there isnt nearly enough data for football (baseball would be far easier). That said, i like the approach over a lot of what i read/see that i feel makes as many assumptions. The numbers are still a crapahoot to an extent, but it gives an interesting idea about probabilities if we make certain assumptions.

Note: all on my phone in hurry.

(11-12-2015 06:11 PM)ohio1317 Wrote:  I agree getting actual numbers is very difficult. I suppose you might devise something from vegas odds, but regardless there isnt nearly enough data for football (baseball would be far easier). That said, i like the approach over a lot of what i read/see that i feel makes as many assumptions. The numbers are still a crapahoot to an extent, but it gives an interesting idea about probabilities if we make certain assumptions.

Note: all on my phone in hurry.

I don't want to get into it again, so suffice for me to say: the actual number that the method spits out can never be verified.

So it doesn't matter what the number is ... so long as they don't use 0% and 100%.

(11-12-2015 05:59 PM)ohio1317 Wrote:  I dont know id say made up. They are certainly are making educated guesses about the odds of winning each game and the probability of who is selected in diffferent senarios but there is a lot more here than typical articles/talk.

If each teams power rankings are slightly different it doesn't chance much. The beauty of his model is using thousands of simulated seasons to give perspective of all the possible outcomes. The % chance for each team of making the playoffs, winning the conference championship, winning out etc is not only reflective of what each team does, but what everyone else does around them. So there will be years of order where everyone wins out and there will be years of disorder where all the top teams lose and bedlam ensues, and there will be everything in between. What this is useful is to get an idea of all the possibilities, and the interaction of those possibilities.

I think the biggest assumption his formula makes is how the committee responds to the various outcomes. As we've seen the committee is tough to predict sometimes. But it's still useful, as it is tough for the human mind to process the thousands of possible game outcomes ( not to mention score outcomes) and how those outcomes interplay with each other.

Question:

1. Big 12 cannibalizes each other to produce 1 loss champ.
2. Michigan wins B1G
3. ND wins out, but Stanford wins Pac 12.

Who gets in at number 4?

1. Clemson
2. SEC champ (one loss)
3. ND
4. LSU/Big 12/Pac 12?

(11-13-2015 07:43 AM)RUScarlets Wrote:  Question:

1. Big 12 cannibalizes each other to produce 1 loss champ.
2. Michigan wins B1G
3. ND wins out, but Stanford wins Pac 12.

Who gets in at number 4?

1. Clemson
2. SEC champ (one loss)
3. ND
4. LSU/Big 12/Pac 12?

If the 1 loss champ is OU, then them. If not, then going to depend on eye test/ game control of the various teams.

OU would be in in such a scenario no doubt, having the head to head wins over the top three teams in the conference. Unlikely, but possible. I think a one loss Utah would have the edge over them, but my intentions were the B1G/Pac producing two 2 loss champs.