≡ Menu

Thoughts on the Oklahoma/Notre Dame Debate

If there’s one good rule of sportswriting, it’s that premature writing is sure to backfire. With Oklahoma (and Gameday) traveling to Stillwater, and Notre Dame on the road in Stanford, there’s a better than even chance that either Oklahoma or Notre Dame finishes with one loss. According to ESPN’s FPI, the Sooners have a 67% chance of beating [1]Of note: This ignores the fact that Sooners star quarterback Baker Mayfield left the game last week with a head injury, and is not certain to play this weekend. rival Oklahoma State, while Notre Dame is an underdog this weekend, expected to beat the Cardinal just 38% of the time.  By the power of multiplication, this implies only a 26% chance of both Notre Dame and Oklahoma finishing the regular season with an 11-1 record.

The SRS says that Oklahoma is the better team, but the only means the Sooners would be favored on a neutral field. But when it comes to making the playoff, we are more interested in rewarding performance than putting the best four teams in.  Two years ago, I wondered whether Ohio State going 13-0 in a watered-down Big 10 was more impressive than Auburn going 12-1 in the SEC.  Now, I ask:

Which is harder: Going 11-1 against Oklahoma’s schedule or 11-1 against Notre Dame’s schedule?

Let’s assume that Notre Dame had an SRS rating of 64.2.  Why that number, given that the Fighting Irish have an actual SRS of 54.6? You’ll see why in a moment.  If Notre Dame was a 64.2, how likely would they have been to say, win at home against Navy?  Well, if we give three points for home field, that would make Notre Dame 13.0 points better, since Navy has an SRS of 54.2.   And we can use the following formula in Excel to determine how likely a 13-point favorite is to win a given game:

(1-NORMDIST(0.5,-(home_fav),13.86,TRUE)) + 0.5*(NORMDIST(0.5,-(home_fav),13.86,TRUE) – NORMDIST(-0.5,-(home_fav),13.86,TRUE))

This formula says a team that is 13 points better should win 82.6% of the time.    Now, let’s run through Notre Dame’s entire schedule. Here’s how to read the Clemson line. On October 3rd, Notre Dame traveled to Clemson to face the Tigers. Assuming Notre Dame has an SRS rating of 64.2, that would imply a projected spread of Notre Dame -1.6 in Clemson, since the Tigers have an SRS rating of 59.6. And based on the formula above, it would mean Notre Dame had a 54.6% chance of winning that game.

DateTeamOppH/RSOSSpreadWin %
09/05/2015Notre DameTexasHome39.9-27.397.5%
09/12/2015Notre DameVirginiaRoad39.7-21.593.9%
09/19/2015Notre DameGeorgia TechHome44.7-22.594.8%
09/26/2015Notre DameMassachusettsHome23.2-44.099.9%
10/03/2015Notre DameClemsonRoad59.6-1.654.6%
10/10/2015Notre DameNavyHome54.2-13.082.6%
10/17/2015Notre DameSouthern CalHome51.7-15.586.8%
10/31/2015Notre DameTempleRoad46-15.286.3%
11/07/2015Notre DamePittsburghRoad49.3-11.980.5%
11/14/2015Notre DameWake ForestHome35.7-31.598.8%
11/21/2015Notre DameBoston CollegeNeut38.5-25.796.8%
11/28/2015Notre DameStanfordRoad54.7-6.568%

Now, what are the odds of Notre Dame losing to Clemson but winning every other game? That is simply the product of all of the numbers in the far right column, except when you get to the Clemson line, you need to use 45.4% and not 56.4%, since we are now putting that game down as a loss. Run those calculations, and there is a 13% chance a team with a 64.2 SRS rating goes 11-1 in exactly that manner.

But if the question is how likely Notre Dame is to go 11-1 against that schedule, we need to include other possibilities. For example, a 64.2 SRS team has a 7% chance of going 11-1 by winning every game but the Stanford game, and a 4% chance of winning all eleven games but losing to Pitt. And, we should acknowledge a 5.5% chance of going undefeated, since we really mean “11-1 or better.”

Add all of those possibilities, and you end up getting an exactly 50% chance that a team with an SRS rating of 64.2 would go 11-1 or better against Notre Dame’s schedule. That, as you can probably figure out, is why I chose that rating. A team with Notre Dame’s 54.6 rating has just a 6.9% chance of going 11-1 or better, so like all teams with excellent record, there was some luck involved.

What about Oklahoma? Let’s use that same 64.2 rating.  How likely would they have been to win on the road at Tennessee?  The Vols have an SRS of 51.8, which means OU would be considered a 9.4 point favorite given the location. Using the formula above, there is a 75.1% chance of Oklahoma winning given those variables.  Here’s how the entire schedule looks:

DateTeamOppH/RSOSSpreadWin %
09/05/2015OklahomaAkronHome31.8-35.499.5%
09/12/2015OklahomaTennesseeRoad51.8-9.475.1%
09/19/2015OklahomaTulsaHome32.3-34.999.4%
10/03/2015OklahomaWest VirginiaHome50.6-16.688.4%
10/10/2015OklahomaTexasNeut39.9-24.396%
10/17/2015OklahomaKansas StRoad41.6-19.692.1%
10/24/2015OklahomaTexas TechHome45.6-21.694%
10/31/2015OklahomaKansasRoad22.4-38.899.7%
11/07/2015OklahomaIowa StHome41-26.297.1%
11/14/2015OklahomaBaylorRoad58.8-2.456.9%
11/21/2015OklahomaTCUHome53.8-13.483.3%
11/28/2015OklahomaOklahoma StRoad54.7-6.568%

Here, the toughest opponent would be Baylor. Oklahoma would have a 13% chance of going 11-1 against this schedule but losing to Baylor. And an 8% chance of starting 11-0 but then losing to the Cowboys. And, given the high SRS rating, a 17.0% chance of going undefeated.

Look at all possibilities, and a team with a 64.2 SRS rating would have a 53.3% chance of going 11-1 or better against Oklahoma’s schedule. To have a 50% chance, a team would “only need” to have an SRS of 63.6. So in that regard, by a hair, it looks like getting to 11-1 or better against Notre Dame’s schedule is ever so slightly more impressive than getting to 11-1 with Oklahoma’s schedule.

If you want to summarize why, it’s really more about the depth of the Irish’s schedule. According to the SRS, playing at Baylor and at Oklahoma State is functionally equivalent to playing at Clemson and at Stanford. But OU only had two other games that really move the needle: at Tennessee (which, to be fair, is harder than Notre Dame’s third toughest game) and home against TCU. Notre Dame played at Pitt, against Navy, at Temple, and against USC. I’m not suggesting we laud Notre Dame for scheduling the Owls and the Midshipmen, but as it turns out, those teams are pretty darn good! Combined, they are 18-1 in games against teams other than Notre Dame.

There is still some time for these SRS ratings to change, and given the close margin, it’s conceivable that by next week, the numbers would flip. If Houston beats Navy and Connecticut looks good against Temple, that could be enough to switch things. And obviously if Clemson struggles against South Carolina, that would be bad for Notre Dame here, too.

But I’m tempted, for now, to side with the Irish. That’s because Oklahoma’s probably had it slightly easier than this analysis assumes. The Sooners played Baylor without quarterback Seth Russell, and while Jarrett Stidham is pretty good, the Baylor team Oklahoma faced was probably not as good as the SRS thinks. Similarly, the Sooners beat TCU without a pair of stars in quarterback Trevone Boykin and wide receiver Josh Doctson. And while the SRS views Oklahoma State and Stanford as even, I’m not sure I agree with that. Stanford’s SRS rating is depressed by its first two games of the year; remove those two games, and the Cardinal jump from a 54.7 to a 58.3.

So I do think Notre Dame has had a slightly tougher road to navigate. On the other hand, I do think Oklahoma is slightly better. That makes this one a toss-up. What do you think?

Oh, and Iowa? Well, the Hawkeyes’ resume will look more impressive if Iowa beats Nebraska and then either Ohio State or Michigan State in the B10 Championship Game.  And Notre Dame obviously hasn’t beaten Stanford yet.  But right now, this model basically says going 11-1 against Notre Dame’s schedule is just as difficult as going 11-0 (to date) against Iowa’s schedule. A team with an SRS of 64.2 would have a 49.7% chance of going undefeated against Iowa’s schedule.

I’ll spare you the math, but here’s the simple version. Both teams played Pitt (although Iowa played them at home), so let’s leave them out.  Otherwise, Iowa has really only beaten two teams of note: Wisconsin and Northwestern.  This model says there is nothing inherently more impressive about going 2-0 against Wisconsin and Northwestern (acknowledging that both games were on the road) than going 4-1 against Clemson, Stanford, USC, Navy, and Temple.  Both are those feats are equally impressive.

Oh, and one other thought. ESPN has a like-minded model called Strength of Record, which reflects the chance that an average Top 25 team would have that team’s record or better, given the schedule. And ESPN has OU ahead of Notre Dame. But I don’t think it’s because our systems are different, but rather, our team ratings are different. The SRS has Navy at #11 and TCU at #12; ESPN’s FPI has TCU at #5 and Navy at #38. Those are pretty big swings, and my model would certainly vault OU ahead of ND if we used those inputs. As always, a system’s outputs are only as good as its inputs. Here, I don’t know if I buy Navy quite at #11 or at #38, but Navy is a big part of Notre Dame’s resume. And the difference between #5 and #12 is big, too. As you can see, this one is really close to call.

References

References
1 Of note: This ignores the fact that Sooners star quarterback Baker Mayfield left the game last week with a head injury, and is not certain to play this weekend.
{ 0 comments }