B1G Championship/Playoff Probabilities based on S&P+

Submitted by Pelini's Cat on November 12th, 2018 at 7:45 PM

So I was really bored in class and decided to put together a quick quantification of our basic B1G championship/playoff odds based on the S&P+ Team Profiles  . It's nothing groundbreaking but I figured I would share. You can see the results below. 

Screen Shot 2018-11-12 at 3.35.13 PM.png

A couple notes if you're interested in my methodology:

  • The UM/OSU "Results" refer to the results of their last two games in order
  • Win Probabilities are taken from the S&P + team profile pages mentioned above
  • Win Probabilities for Northwestern are estimated based on M's postgame win expectancy from the Team Profile page
    • This is somewhat generous to NW considering we played a pretty sub-par game in Evanston, but given the past decade of Michigan football, you'll forgive me for being conservative
    • I rounded down to 80% for OSU's win probability vs. Northwestern, based on the fact that OSU is a little worse than UM according to S&P+
  • The resultant B1G east champions in each scenario are color coded accordingly
  • I made a couple of assumptions with regards to the playoff:
    • UM gets in if they win out and doesn't if they don't.
      • Honestly I don't see a scenario where this isn't the case. A two loss team doesn't get in and a one loss B1G champ doesn't get left out. 
    • OSU is a coin flip to get in if and only if they win out
      • I think some things need to break their way. No team that has been blown out has ever gotten in. I just divided the probability they win out in half and called it a day.



November 12th, 2018 at 9:49 PM ^

Of course, ALL OF US would be more than happy with a 1-point win, having so many near misses and heartbreak -- among the blowouts -- vs. OSU in the past 15 years. That said, yeah...wouldn't it be just the best to blow them out, to NOT have to sweat through the win, and to leave their fans in stunned silence?

We can only hope. 

Again, I'd be more than happy to win 7-6, though. 


November 12th, 2018 at 9:08 PM ^

Does S&P win probability take home vs away into account?  Giving us a 60% chance to win seems high given that we'll probably be playing in one of the most hostile environments in all of sports. I think we'd be 7 point favorites on a neutral field, but in Columbus it's a toss-up.  


November 13th, 2018 at 7:16 AM ^

All stats & math aside, I can’t forget two years ago watching Meyer running up & down the sideline exhorting the fans. This year I hope he does the same with about 17 of his fans left in the stadium because they all left after the first half blowout.


November 13th, 2018 at 10:13 AM ^

UM gets in if they win out and doesn't if they don't.

  • Honestly I don't see a scenario where this isn't the case. A two loss team doesn't get in and a one loss B1G champ doesn't get left out.

There's a really easy scenario where the B1G champ gets left out.

A 1 loss 'Bama is getting in over Michigan.  It rounds out with SEC champ, Clemson, ND.  Bank on it.


November 13th, 2018 at 10:40 AM ^

People keep saying this, but I can't really believe it. In the scenario you describe, Alabama and Michigan will both be 12-1. Michigan will have 10 conference wins compared to Alabama's 8, plus a conference championship. Alabama's OOC wins are an FCS team, two mediocre Sun Belt teams, and a really bad Louisville team. Losses would be comparable, both to other playoff teams. 

Alabama may be the best team ever, but something as subjective as the "eye test" really can't override the resume comparison without causing an uproar.


November 13th, 2018 at 12:41 PM ^

People keep saying this, but I can't really believe it.


When has logic and reason applied to CFP decision making?

We're told, or it used be on the CFP's website in their charter (did you ever read it), that SoS matters.  Hint: it doesn't. 

Bama doesn't have to travel or play P5 OOC because THEY ARE STILL ALWAYS GONNA BE RANKED #1.

We're told, or it used be on the CFP's website in their charter, that similar opponents and H2H matters.  It doesn't.  It's a lie.

What the CFP does is not an accomplishment process.  It's a beauty contest... and people are delusional if they think 'Bama isn't the hottest chick even though Michigan has the better personality.

Ecky Pting

November 13th, 2018 at 12:49 PM ^

This is a misguided assessment on so many levels...

If Bama's loss is in the SEC championship to Georgia, who also has 1 loss in the end and has just beaten Bama, then Georgia goes to the CFP and Bama does not. The SEC championship is a de facto quarter-final game, and the CFP is not going to go for a rematch of a game that was just played 3 weeks prior on a neutral field. Moreover, the CFP looks for teams that have won their conference championships, and by inference, will eliminate teams that have lost their conference championship. The only reason Bama got in last year is because they had one loss, did not play in the SEC championship, did not play the SEC champion, Georgia, at any other time, and there were no other teams with one-loss with as good an argument for being in the playoff as Bama.

Ecky Pting

November 13th, 2018 at 10:22 AM ^

The win probability for Michigan vs. Northwestern can be computed based on the difference in the S&P+ ratings, which projects a point spread on a neutral field.

Michigan        = +24.9
Northwestern =    -0.8
Spread           =  25.7

Given that the probability distribution of point spreads follows a Gaussian distribution with zero mean and standard deviation of 15.53 points, the win probability for Michigan to beat Northwestern is 95.1% (about 1:19 odds).

The probability of Michigan advancing to the Big Ten Championship is reduced to 3 possible mutually exclusive scenarios:

  1. Maryland beats OSU (22.8%) and UM beats IU (97.0%)
    => 0.228*0.970 = 22.1%
  2. OSU beats Maryland (77.3%) and  UM beats OSU (60.2%)
    => 0.773*0.602 = 46.5%
  3. Maryland beats OSU (22.8%) and IU beats UM (3.0%) and UM beats OSU (60.2%)
    => 0.228*0.030*0.602 = 0.4%

So the total probability of being the B1GE champ is P(1) + P(2) +P(3) = 22.1% + 46.5% +0.4% = 69.1%, which is more or less in agreement with the OP.

From there, the probability of M winning the B1G championship is 65.7%, or the probability of making it to the game times the probability of winning it (0.691*0.951).