FiveThirtyEight CFP Odds

Submitted by michiganman001 on

So FiveThirtyEight has brought their model back to project the CFP rankings while using FPI to predict the outcomes of games. Obviously they have to predict human behavior, which isn't as easy as 11-5>10-6 and therefore team x makes the playoffs. 

Michigan is given a 13% chance of winning out and a 8% chance of making the playoffs and then a 1% chance of winning it all. Some of us have talked about how the committee would see a 11-2 UM team, according to these odds, an 11-2 UM team would have a 61.5% chance of getting in the playoffs.

It is worth noting that they gave TCU a 91% chance of making it last year and OSU a 40% chance and we all know what happened.

The link: http://fivethirtyeight.com/features/ascending-tide-updated-college-foot…

 

 

MGoBoard
Football

Farnn

November 8th, 2015 at 7:03 PM ^

If Michigan wins out and doesn't make the playoffs I won't even be mad.  The season has been a nice surprise so far and every win above 8 will be above what reasonable preseason expectation were. 

Joined: 07/07/2008

MGoPoints: 12788

LSAClassOf2000

November 8th, 2015 at 7:29 PM ^

I would agree - if you go with Massey's estimates, our own chances of winning out are about 23%, but if we do it and don't quite make the playoff (we migh be able to see the floor, which isn't that bad really) and perhaps as a bonus slide into one of the other New Year's Six bowls, that's above what a lot of people thought would happen. Indeed, if it only gets better from there as we might project, I think we could live with that.

Joined: 01/07/2011

MGoPoints: 130608

Bocheezu

November 8th, 2015 at 7:24 PM ^

I wouldn't publish something where

-- TCU is more likely to make the playoffs than win a conference championship

-- MSU is more likely to make the playoffs than U-M, but U-M has a better chance to win the conference

Maybe these are the "if everybody in the country except TCU/MSU loses 2 games" scenarios, but come on, that doesn't account for a 1-2% discrepency in those probabilities.  Plus, show tenths of a percent for crying out loud.

Joined: 02/22/2009

MGoPoints: 2154

umfan83

November 8th, 2015 at 7:25 PM ^

As far as MSU/UM I'm sure UM is more likely to win the conference because Michigan plays Ohio State at home and MSU plays them in Columbus. Meaning it's more likely UM beats OSU with MSU losing to OSU. it probably has UM as the slightly better team as well.

Conversely, MSU probably has better playoff odds because if they won out they would be a 1 loss team and having won in Columbus and A2 and against Iowa. If Michigan wins out they'd still be a 2 loss team with wins at home over OSU and Iowa. 1 loss and 1 more quality win in conference play = more likely to make the playoffs



Sent from MGoBlog HD for iPhone & iPad

Joined: 03/10/2012

MGoPoints: 8466

MichiganMAN47

November 8th, 2015 at 7:36 PM ^

We should win both of these games, but they are far from a given, especially on the road. If we go into that OSU game with a chance at winning the division I really like our chances to finish the job. Should we win out, anything can happen. As strong as the Big Ten is this year, the committee will have a hard time keeping us out.

Joined: 02/24/2015

MGoPoints: 3216

Muttley

November 8th, 2015 at 7:47 PM ^

for UM to make the playoff using the input sources below:

 

For UM to make the playoff, the four events that need to happen are:  Note that the symbol "|" stands for "given" in probability & statistics.

1) UM wins out reg season
    • 35% = 85%(@IU) * 69%(@PSU) * 61%(OSU)
    • Etky Pting MGoBlog Diary
2) UM makes B1G Champ Game | win out
    • 77% OSU beats MSU
    • Etky Pting MGoBlog Diary
    • Any other MSU loss
        • Also advances UM to CCG
        • But simultaneously lowers #4 below
        • So I'm going to use #4 as a catchall
             for those two offsetting effects
3) UM wins B1G Champ Game | in B1G CCG
    • 52% UM beats Iowa
    • MasseyRatings.com UM v (undefeated) Iowa
    • Any Iowa loss
        • Increase UM's CCG chances
        • But simultaneously lowers #4 below
        • So I'm going to use #4 as a catchall
             for those two offsetting effects
4) UM makes Playoff | B1G CCG win    
    • 61%
    • Aww, heck, this is a SWAG at best
    • Originally, I picked 50% out of my butt
       after glancing at the info below
    • So I'll just go w/ the FiveThirtyEight #

This results in a probability of:

    • 8.5% = 35% * 77% * 52% * 61%

Note that I'm not saying that FiveThirtyEight is "accurate", but rather that their estimate seems ballpark reasonable to me.

=============================================

Games that will or may help Michigan's ranking: 
(w/ present records and CFP rankings. Rankings to be updated Tuesday.  Teams that lost this weekend are inred.)

• 5 remaining B12 round robin games & BU-Tex
    • #6  Baylor 8-0 v  #15 Okla 8-1 11/14
    • #6  Baylor 8-0 @ #15 OkSt 9-0 11/21
    • #6  Baylor 8-0 @ #  8 TCU 8-1 11/27
    • #6  Baylor 8-0 v         Tex  4-5 12/05
    • #8  TCU     8-1 @ #15 Okla 8-1 11/21
    • #14 OkSt 9-0   v  #15 Okla 8-1 11/28

• SEC cannibalism 

    • #2 LSU 7-1 v Bert 5-4 11/14
    • #2 LSU 7-1 @ #18 OleMS 7-3 11/21
    • #2 LSU 7-1 v #19 TAMU 6-3 11/28
    • #4 Bama 8-1 @ #20 MsSt 7-2 11/14
    • #4 Bama 8-1 @ Aub 5-4 11/28
    • CCG SEC West v #10 Fla 8-1 12/05

• PAC12
    • #12 Utah 8-1 v #23 UCLA 7-2 11/21
    • CCG #11 Stan 8-1 v PAC12 North 12/05

• ACC
    • CCG #1 Clem 9-0 v ACC Coastal 12/05

• Cannibalistic OOC games including ND
    • #5 ND 8-1 @ #11 Stanf 8-1 11/28
    • #10 Fla 8-1 v #16 FSU 7-2 11/28

Joined: 07/07/2009

MGoPoints: -73245

Muttley

November 9th, 2015 at 9:09 AM ^

I took the regular season individual game win likelihoods from the latest diary post from Ecky Pting as of 11/05.  He quotes Bill Connelly's model estimates, the most recent of which can be found here:

 http://www.footballstudyhall.com/pages/2015-michigan-advanced-statistical-profile

I think the most useful takeaway is that, even with this Michigan-loving, backwards stats-only-based 61% win likelihood over OSU, winning out through OSU is still by far the least likely of the four events we need to have happen for Michigan to reach the College Football Playoffs.

You make an insightful point about the likelihood of Michigan beating Iowa increasing after a (scenario-defined) win over Ohio State, but I don't know how/where to observe that.

Joined: 07/07/2009

MGoPoints: -73245

SFBlue

November 8th, 2015 at 8:38 PM ^

At this time last year Ohio State was #14 in the AP, #13 in the coaches poll, and Mississippi State was number 1. Statistics like those used by 538 are not of much use, because the assumptions and the landscape change by the week.  

Joined: 11/02/2009

MGoPoints: 15699

bluesalt

November 8th, 2015 at 10:26 PM ^

It's probabilistic. If they predict something should happen 60% of the time, and they got it right all the time, there's something wrong with the model. Is there a decent margin of error with these predictions? Sure. But that doesn't mean there isn't anything to take away from it.

Joined: 11/02/2011

MGoPoints: 8541

Avon Barksdale

November 8th, 2015 at 10:20 PM ^

If we win our first Big Ten Championship since 2004, and go 11-2, I won't care if we miss the playoffs. It would be fantastic to go, but I'm rational and understand it's going to take a lot for a two loss B1G Champion to get into the CFP.

I would be ecstatic with an 11-2 season that included wins over four ranked teams (BYU, Northwestern, Ohio State, Iowa) and a Rose Bowl berth. Hopefully we put it all together this last month of the season.

Joined: 04/27/2013

MGoPoints: 12859