Weekly Probabilities 10/28

Goalscorer9's picture
October 28, 2013 at 1:53p
27 Comments

Here they are!  Again, although the process is objective, I acknowledge my probabilities of individual games are subjective, though I think this gives us a pretty good picture of where we are at.  *I tried to find the vegas or massey odds people talked about last week, but after a short search was not able to find them.  If someone provides a link I’d be happy to use those in the future* 

As always, my numbers will be attached for anyone who would like to critique.

Assumptions:
-Ohio State finishes undefeated.  This whole argument starts with "If Ohio State goes undefeated", so we'll assume they do to calculate probabilities
-If Alabama, Florida State, Oregon, or Miami finish undefeated, they will jump Ohio State.
-No 1 loss team will finish in front of an undefeated Ohio State
-Baylor, Fresno State, and Northern Illinois will not jump Ohio State, even if they finish undefeated

Numbers you care about:

-Probability Alabama wins out: 29.5%
-Probability Oregon wins out: 25.7%
-Probability Florida State wins out: 23.4%
-Probability Miami wins out: 7.8%

-Probability all of them lose, and Ohio State finishes numero uno: 37%

-And most importantly, the probability that at least 3 out of these 4 teams lose, and an undefeated Ohio State team finishes in the top 2: 63.9%

Our odds took a bit of a dip this week.  Missouri losing certainly helped a little, but their odds were so slim it didn’t matter much.  Oregon clearing the UCLA hurdle was much more substantial.  Still, we’re more likely than not to finish in the top 2, should we go undefeated.  We definitely should be big Stanford and Miami fans this week, though.


For those of you who would like to include the possibility of Ohio State losing:
-Probability of Ohio State winning out- 42.3%
-No undefeated assumption odds of Ohio State being in the national championship game- 27.1%

Comments

Toilrt Paper's picture

Your numbers show there will be 4 undefeated teams.

Goalscorer9's picture

No, they don't.  Is that a possibility?  Sure.  But it is unlikely.

Toilrt Paper's picture

Alabama, nothing better than a 35% chance of losing.
Oregon, nothing better than a 45% chance of losing.
FSU, nothing better than a 40% chance of losing
For us to feel good about your math AND a BCS Title berth, two of those three teams must lose. Your odds say they won't.

Goalscorer9's picture

You must not be very well practiced in how probability works.

Roll a die once, the's a 5/6 chance it's not a 1.  But if you roll that die 6 times, it's probable that a 1 would come at least once.  (In this case, the 1 is a loss).  So even though each individual game Alabama, Oregon, and FSU will be favored, they are still not likely to finish undefeated.  (Alabama is the most likely at only 29%)

kholmes's picture

This statement below is incorrect:
-Probability all of them lose, and Ohio State finishes numero uno: 37%
The 37% refers to the probability that they all lose but that doesnt necessarily mean that OSU wins all of their game. Based on your numbers, the probability that they all lose and OSU finishes number one (by winning their games) is 15.7%. The probability that they all lose and OSU loses is 21.3%.
Also, you are overestimating the chance that they all lose because either FSU or Miami has to win that game and you essentially estimate some chance that they both lose that game where they play each other.
I know..nitpicking. Not trying to be a jerk.

Goalscorer9's picture

Kholmes.  You obviously didn't read the whole thing.  Read the first line in my assumptions.  I also accounted for Ohio State losing in the last line.

I did include that in my calculations.  There are about 20 lines of calculations I did not include in image, which account for the games that are not mutually exclusive (ie FSU Miami).

Also not trying to be a jerk haha :)

kholmes's picture

Yep - you're right - I missed the line with assumptions that OSU goes undefeated and I just looked at that statement by itself. 

Goalscorer9's picture

ha no worries!  We've all been guilty of skimming through long posts before haha  ;)

kholmes's picture

But I still think you did not account for the FSU/Miami issue correctly..although it makes only a small difference (at least I believe I will put as my qualifier). The 37.0% number can be tied out to easily by multiplying (1-Prob of winning all games) for each of the 4 teams..but then that doesnt account for the Miami/FSU issue. I see the prob of either Miami or FSU winning out as (.65*(prob FSU wins remaining games after first one)+.35*(prob of miami winning all remaining games after first one) since there is that 0.65/0.35 breakdown of each winning that game and then after the first game you only care about whoever won that first game. Using this would get the probability of all 4 teams losing down to 36.1% from the 37.0% if you didnt account for it. So overall I just wasted 10 minutes typing to say that I believe it should be 36.1% instead of 37.0% which is completely meaningless in the grand scheme.

Goalscorer9's picture

Ah you are right there.  I did do it correctly in my at least 3 lose section, but forgot to go back and do that part in the all 4 lose section.  Luckily after this week the remaining teams will all be mutually exclusive and it will be much easier to calculate haha

shaggybuckeye's picture

I really enjoy these posts Goal. Thank you for putting in the effort.

Goalscorer9's picture

No problem!  Glad you enjoy them :)

AndyVance's picture

I love your probability column each week, GoalScorer. I saw this in Gene Wojomanyletters' column over at the Four Letter Network, and thought it was an interesting look into the probabilities a non-Ohio State fan might assign to the same question:

In: The chances of Alabama, Oregon and Florida State finishing the regular season undefeated. According to ESPN researcher extraordinaire Chris Fallica, Alabama has a 66.2 percent chance to win its remaining regular-season games, followed by Florida State (63.2 percent) and Oregon (55.1 percent). And Fallica's number crunching shows that FSU (54.7 percent) actually has a better chance of entering the bowls undefeated than Bama (54.3 percent). Oregon is at 45.5 percent. The conference title game opponents are based on the highest teams in the standings.
Out: Ohio State's chances of finishing the regular season unbeaten. Of the top four teams in the BCS standings, the Buckeyes had the lowest probability of surviving the regular season without a loss (48.4 percent) and entering the bowl season without a loss (26.1 percent). Fallica also calculates that there's a 10.3 percent chance that all four teams are undefeated on Championship Saturday, Dec. 7.

I'm having a hard time understanding how they view Ohio State as the least likely of the heavyweights in the BCS of finishing undefeated... I suppose they're figuring the Ross-Ade voodoo will hold sway, M*ch*g*n will suddenly become a good team, or that half the team will get dysentery and die. 

Goalscorer9's picture

If I remember correctly Fallica has undervalued us all season.  (for a while, he had us like 8th in the power rankings, I believe).  They either: a) just assigned, imo, ridiculously high probabilities to the top 3's remainging games or b) are really horrible at math.

My guess would be a.  Every year ESPN likes to act like they have the entire season figured out by week 6, and then are so shocked when teams lose.  In college football, upsets happen a lot.  My guess is that they even have games like Alabama vs. Auburn at 90+% for Bama, which I think is silly.

Goalscorer9's picture

Yeah, this week is the first time Fallica has us in the top 5.  He had us at 9th preseason, so he's clearly still bitter about how wrong he was about the buckeyes.

AndyVance's picture

I was reading right along on his probabilities for the top three, but such a low percentage for the Buckeyes seemed farcical. Predicting a loss to a 31-point underdog is silly, so that leaves two teams who are almost as hopelessly overmatched as the Boilermakers, and a M*ch*g*n team who might legitimately drop two or three more games before season's end... That said, I always peg The Game as 50/50, because anything can - and has - happened there... but still, it seemed like a crazy low number.

Goalscorer9's picture

I disagree with his assessment, but can understand his Ohio State number.  My 42% could easily drop down to a 26% if you think the TTUN game is 50/50, and the B1GCG is closer to 50/50.  So I guess that didn't surprise me coming from someone who clearly assesses Ohio State as a worse team than I do.

I just don't understand how he's getting such high probs for the other teams.  He has all 3 at above 50%, the highest I have out of those three is 29%.  I don't think he's including the championship games in his numbers, which at least partially explains it, but still...  Let's just say I think he's being much harder with his projected probs on Ohio State than the other 3.  I guess our remaining schedule is just tougher

AndyVance's picture

I guess our remaining schedule is just tougher.

Yes, that is the only plausible explanation :)
That said, Alabama has two decent games left compared to Ohio State's one (regular season, that is), Florida State has a cakewalk after this week's contest, and Oregon only has one ranked team left on the slate (although the game versus Oregon State should be a good one as well).
Our schedule is pretty darn easy to navigate, but the other three have pretty good paths to an undefeated season, too. Still, I agree far more with your assessments than I did his.

Goalscorer9's picture

For some reason I see Florida rallying up for a season defining win and knocking out FSU.  I know Florida is beat up and not playing well, but there is talent there.  I just think everyone is overlooking that game when penciling FSU in for an undefeated season.  Not to mention they probably have 2 match ups left with Miami.  Are Miami world beaters?  No.  But they're undefeated through this much of the season.  I wouldn't want to have to play a team like that twice.

Stanford is a good team.  They're clearly Oregon's biggest hurdle.  But a rematch with UCLA won't be easy in the championship game.  They have a rivalry game with Oregon St. who has some talent and is a fringe top 25 team.  Even Utah showed it has some bite with a big win over Stanford.

Alabama Obviously has LSU, Auburn in a rivalry game, and the SEC championship.

 
So the way I see it, each of these teams have at least 3 games that aren't gimmes.  So all this "I don't see any of them losing" talk is a bit premature in my eyes.  But, it happens every year, I don't know why I'm surprised by it haha.

AndyVance's picture

I've always said - as in basketball - that it's tough to beat a team twice in a season. Both FSU and Oregon have that possibility before them, which is part of the reason I like our chances to win out and get in the big dance.

kholmes's picture

If you notice, Fallica has two probabilities for each team (the chance they win "their remaining regular season games" and the chance of "entering the bowls undefeated"). From this, it appears he must define the regular season as not including conference championship games.
But given that he has both probabilities, you can get a glimpse of what he is assuming for the championship game to get a sense of what probabilities he is using for the individual championship game. Note that he says he is using games against the team they would currently be projected to play in conference championship game (so MSU for OSU). Based on numbers, it looks like he gives OSU a 54% chance (26.1/48.4)  to win the B1G championship game (assuming they make the championship game), FSU an 87% chance to win ACC game (same assumption), Oregon an 82% chance to win Pac 12 game (same assumption), and Alabama also 82% (with same assumption that they first make the championship game).
This doesnt really pass the smell test at first glance as putting OSU/MSU as a 54/46 game does not seem reasonable given the high numbers he is giving other teams. Presumably he probably has OSU/UM at similar numbers which also doesnt seem that reasonable (and I'm a Michigan fan so not biased towards OSU in looking at this probability objectively)...as a UM fan, I think your 75% probability seems reasonable for what its worth.
The 87% prob of FSU winning the ACC championship game against Miami seems high but I see that FSU is a 21 point favorite so it may be in line with those numbers then. This may be a case that people/lines have FSU, Bama, and Oregon as significantly overwhelming favorites against anyone which is what would be needed to get those probabilities. Is Alabama going to be a 20+ point favorite against Auburn? If so, then maybe these probs arent that out of whack with what gambling lines will be.

AndyVance's picture

Excellent analysis - I agree entirely that his prediction of Ohio State's chances in its final four games, and its chances against Sparty, are way too low while his predictions for the top three BCS teams seem unrealistically high.
Other threads have gone into this in more depth, but Alabama still has to face LSU and Auburn in the regular season, and either Mizzou or South Carolina in the SEC Championship. If Alabama manhandles LSU, that's probably in line with LSU being somewhat "down" this year, but if they blow out Auburn, doesn't that tell us that voters have had them highly overrated this year? But, I digress. LSU, Auburn and South Carolina (not so much Mizzou) all have the possibility of giving the Tide a game.
For FSU, the only real regular season contest is this week's tilt against The U, where we will probably see Miami unmasked as a team who isn't quite "back" yet. If FSU is going to "Clemson," it might be against a Wake Forrest team who has shown flashes this season, but really, who are we kidding. FSU might be the most complete ballclub in the BCS this year. A rematch against Miami, or a battle against Virginia Tech, could make for an interesting conference game. FSU's road to 13-0 is much easier, it seems, than Bama's.
That leaves Oregon. I would have said that they have the hardest road to an unblemished schedule, but now I'm not so sure. The PAC-12 may be the all-around best conference this season, despite heraldry earlier that the ACC was surging (it wasn't). There are several ranked teams competing for the right to play Oregon in the conference title match, and the Ducks still have to get past Stanford and Oregon State in the regular season. I'm hoping Stanford can do Stanford things and stymie the vaunted Oregon offense this week; if not, it's a rematch against UCLA, or perhaps Arizona State, to seal the deal.
At any rate, Ohio State playing Purdue, Indiana, Illinois and M*ch*g*n is a much easier slate - the pundits have been telling us how damn soft our schedule is all season, and now they only give us a 50/50 shot to win all those games? does not compute does not compute does not compute

Goalscorer9's picture

not even a mention of the gators on the FSU section? lol

Goalscorer9's picture

Even on a down year, playing Florida at the swamp in a rivalry game with the pressure of a BCS title appearance on the line raises some huge red flags for me.  Add in the fact that NOBODY is talking about this game or giving florida any credit and I just smell upset.

AndyVance's picture

From your mouth to the gods' ears... For once a Gator win wouldn't bother me at all :)

kholmes's picture

AndyVance...I love your last paragraph.
Basically it is saying that OSU is least likely to win out as it is playing competition closest to its own level than any of the other teams (since they have OSU near 50/50 while other teams are playing tought games in the 80-85% range). This can only be the case if 1) OSU is playing tougher competition than those other teams OR 2) OSU is just assumed to be a worse team than the Ala, Ore, FSU teams and so OSU is near the level of its competition even though the competition is supposedly weak. But the narrative all along was that OSU isnt playing anyone decent so they dont deserve to be in the title game over those times. From the probabilities and your last paragraph, the only conclusion is that the media/polls just made the assumption at the start of the season that OSU would not be that good this year because of their weak schedule and so therefore they are pretty close to the level of the supposedly weak competition. I think those probabilities just show how strong the perception of OSU nationally has been affected by this runaway "weak schedule" theory before any of the games were even played.
Despite being a Michigan fan, this sort of crap above is why I think I would be rooting for OSU to kill Alabama if they played in  a bowl game to put an end to this stupid line about B1G competition.

AndyVance's picture

Right on - the punditry was left little choice but to rank Ohio State #2 at the top of the season after it went undefeated under Urban Meyer last year... an empathy vote, if nothing else: "Awww, it's too bad you had nothing to play for in the post season. Here's a good look in the polls."
But they were essentially looking for any reason to drop us behind the two (or three) teams they really wanted in the national title fight: Alabama and some combination of Oregon, FSU or - until they tanked - Clemson. The B1G is weak narrative is a quasi-canard used to justify their banal treatment of OSU in the polls after the preseason gimmie.