Oregon Lost One Game. Now Nobody Can Agree How Good They Are.

Oregon lost once and the consensus dropped them from 4th to 19th. But one system still has them 5th and another has them 76th — and that gap tells you more than the ranking does.

The Sports Page · The Professor · with thanks to Kenneth Massey, who publishes all of it

The question: how much should one Saturday change your mind about a team?

Nobody agrees. Thirty-five rating systems looked at the same Oregon loss and moved them to somewhere between 5th and 76th.

Why that is surprising. A ranking is presented as a fact — one number, printed, settled. Underneath it is an argument that nobody has won.

What you get: a way to tell a solid ranking from a contested one, using a number nobody prints.

What happened

Oklahoma State beat Oregon 39–31 as a 23-point underdog. We wrote about that on Sunday; it was roughly a 1-in-13 result.

By Monday, Oregon had gone from 4th to 19th in the composite ranking. Oklahoma State had climbed 26 places.

That is the part everybody saw. Here is the part nobody prints.

The systems did not agree. They barely agreed at all.

Massey prints every system next to the composite. This week they scattered.

On Oregon: one system holds them 5th. Another has them 76th. The middle half of all thirty-five rankings spans twelve places.

On Oklahoma State it is worse. The rankings run from 38th to 116th. The middle half alone spans 44 places. That is seventy-eight ranks of doubt about a club that won one game.

Now take Notre Dame, who did nothing odd at all. Their middle half spans 2nd to 4th, and eight of the thirty-five systems have them first outright.

Look at the shape of it. The top six are tight little clusters. Then Michigan spreads, Oregon spreads further, and Oklahoma State smears across eighty ranks of the page.

What the spread actually measures

Every rating system starts with a belief about a team. Returning players, recruiting, last year. Call that its prior. Then results arrive and each system updates.

How far it moves depends entirely on how strongly it held that prior.

A system with a confident preseason view of Oregon treats one loss as a bad afternoon; it barely moves. A system that leans on results alone has no such anchor. In September, one loss is a large share of all it knows.

So Oregon sitting anywhere from 5th to 76th is not thirty-five systems failing. It is thirty-five answers to one question, laid side by side: how much should I believe what I just saw?

And notice where the spread is widest. It is widest on exactly the teams whose last result was most surprising. Nobody argues about Notre Dame, because nothing happened to argue about.

We have a number for this, and it is uncomfortable

On Monday we measured this across 236 teams. How far a team finishes from the line one week explains 0.3% of the next week; the correlation was +0.06.

Set that beside a fifteen-place drop.

We are not going to tell you the systems are wrong; we cannot show that. Ranking a team is not the same job as predicting next week’s margin. A loss carries something a margin does not. It happened, and it counts from now on.

But the tension is real and you should hold both numbers at once. Something moved Oregon fifteen places. Our best measurement of what a single result predicts says it should have moved them less than that.

What to take home: ask for the spread

A ranking is a point estimate. Nobody publishes the error bar, and a point estimate without one is half a sentence.

So when you see a number, ask what the disagreement around it looks like. Notre Dame at 2nd, with everybody between 2nd and 4th, is a different claim entirely from Oregon at 19th with everybody between 5th and 76th. Both are printed the same way; only one of them is settled.

You know where to find it now. Massey puts every system side by side, free. The spread across them is the closest thing college football has to an honest error bar.

Watch it this week. The teams that everybody agrees about are the teams nothing has happened to yet.