Projecting Washington’s move up the playoff rankings after a wild week 10

We’ve spent a lot of time on the blog talking about the College Football Playoff and the manner in which the selection committee goes about doing their seeding.

We do know quite a bit about what the committee considers as criteria for ranking. The committee, in fact, published that criteria when they launched their venture four years ago. From their own “protocols” document, they provided much detail.

This charter document is divided up into several sections that outline the “what’s” and “how’s” of the selection process. In the “Principles” section, they identified the five factors that they cared most about:

... Conference championships won

... Strength of schedule

... Head-to-head competition

... Comparative outcomes of common opponents (without incenting margin of victory)

... Other relevant factors such as key injuries that may have affected a team’s performance during the season or likely will affect its postseason performance

In another part of the same document, there is a section called “Selection Process” that helps explain how the committee will apply the principles above to their method for selecting teams. Here is a verbatim lifted directly from it:

...establish a committee that will be instructed to place an emphasis on winning conference championships, strength of schedule and head-to-head competition when comparing teams with similar records and pedigree (treat final determination like a tie-breaker; apply specific guidelines).

The criteria to be provided to the selection committee must be aligned with the ideals of the commissioners, Presidents, athletic directors and coaches to honor regular season success while at the same time providing enough flexibility and discretion to select a non-champion or independent under circumstances where that particular non-champion or independent is unequivocally one of the four best teams in the country.

In addition to the committee’s own protocol document, we have the original publications from the NCAA when they were going through the process of defining the process and assigning the committee. This excerpt from the FAQ publication helps clarify (a little) the importance of a criterion that UW fans are banking heavily on:

A selection committee will pick the four teams -- using guidelines such as strength of schedule, head-to-head results and won-loss record -- after the regular season. The committee will give preference to conference champions.

All of this evidence is fine and good. But we still know very little about the actual process the committee engages in and how each individual member assembles their own ballots on a week to week basis. There have been some good attempts at reporting on this. You can find examples here and here.

If you read those reports, you will find that there are some big gaps that the reporters fail to address, particularly in how the committee actually assembles the “short list” of teams that they put on the initial ballot and to what extent members are given freedom (or not given freedom) to use or exclude data in how they score their ballots.

Because we don’t have this insight, we are left to debate the merits of every team that could make it in ad nauseum. Brad, Kirk and I, in fact, had a doozy of an exchange on Friday that would leave all of you both breathless and dismayed were I to share the full text of that exchange here.

I’ll spare you the details.

All of this is to say that projecting the committee’s output is difficult even though we have a very good idea of what criteria is being utilized. FiveThirtyEight.com has what I think is the best model out there, and we’ve shared it here several times. As of this morning, they are projecting UW with a 32% probability of getting into the playoff (sixth overall). If UW wins out, they model the Huskies odds of getting into the playoffs at 84%.

That projection will likely confuse many fans because when the rankings come out later this week, UW will not be included anywhere near the top four.

So, why the discrepancy? The short answer is that the committee won’t have scored one of their most important criteria in any ranking that they release before the end of the season: conference championships. Until they do that, the actual ranking is sure to look very unlike what will actually come to pass.

Last week, I introduced my own method of ranking the teams in the race using the criteria that the committee is using. I did it for the purposes of illustrating the method more so than actually projecting the weekly or final rankings. But I found that I was “in the ballpark” when I compared my method with the actual week 9 results. So I decided to do it again.

Week 10 was obviously a huge week for the Huskies. They had a “style points” kind of win against a long-time rival at the same time that these things were happening:

• tOSU got their asses kicked by an unranked team in a spectacular fashion
• Penn State lost their second game essentially ending their shot at the B1G championship
• Oklahoma State took a second loss
• Oklahoma got the big win over OKST but demonstrated that defense is optional on their team
• Clemson took control of the ACC
• Notre Dame gave up 37 at home to a basketball school
• Iowa State’s dream came to an end

It wasn’t exactly UW’s dream scenario weekend (losses by Georgia, Miami, TCU and Clemson would have made the weekend nearly perfect), but we’ll take it.

Now that I’ve digested all of the chaos AND the updated offensive, defensive, special teams and strength of schedule stats, here are my projections for what the updated CFP rankings will look like later this week:

If you happen to be reading this on a phone and can’t decipher the table, the punchline here is that UW is still on the outside looking in. I’m projecting them to be ranked this week as the eighth team in a virtual tie with TCU. The top four are Georgia, Alabama, Clemson and Notre Dame with Oklahoma the next team in.

In my last article, there was much discussion about style points and “quality wins” being a gap in my analysis. Since my own method does seem to generate results very similar to what the committee is generating, I’m not sure that this is the case. Strength of schedule is already in the model (I use updated rankings from SB Nation to score this section). But it is true that I haven’t done anything to try to capture the style points that many people think influence committee selections.

For fun, I went and looked up the scoring differential stats of all teams in the NCAA following week 10. Lo and behold, there are a lot of playoff contenders showing up on the leaderboard of that list. Alabama is number one, of course. But UW makes a strong showing at number four with an average margin of victory at 23.9 points. That’s pretty impressive.

It turns out that taking that data and using it to add “style points” into the model makes a big difference, particularly for UW.

This is what my week 10 projections would look like if “style points” (scoring margin) was included in it:

The top four stay the same, but UW jumps up to #5 in a virtual tie with Oklahoma. Keep in mind that strength of schedule is already scored here, so the Huskies are already being dinged for that. That said, it turns out that UW has been “handling its business” more impressively than some other teams and that they may not yet be getting full credit for it.

I suspect that scoring for style points, just like conference championships, will work itself out in the scoring as the committee processes full season data after the conference championship games have all been played. So I’ll go ahead and keep that metric in my model going forward.

When I forecast my model out to the end of the season, my results look very similar to FiveThirtyEight.com. And, yes, I’m still projecting UW into the playoffs.

To set the table, here are the key assumptions I make in scoring my forecast:

• TCU loses next week at Oklahoma but wins the rematch for the Big 12 championship
• Ohio State wins out and wins the B1G
• Notre Dame wins out
• Clemson wins out and takes the ACC
• Miami, by definition, gets two losses
• Alabama wins out, handing Georgia their only loss in the SEC championship

UW is a three seed in my final season rankings model mostly because of the fact that I weight the conference championship highly in my formula. That combined with the fact that the projected B1G and Big 12 champs will have two losses apiece make UW a clear #3.

We can certainly debate both my forecasted scenario (e.g. Oklahoma could easily win the Big 12 with one loss) and the weighting of certain criteria (especially conference championships relative to strength of schedule), but I’ve at least accounted for all the criteria that goes into the model. So the critical pieces of information that go into the selection process are on the table for discussion.

Regardless of what you think of my projection, the path for UW is relatively clear now. The most important factors for UW in getting to the playoff (in my model) are:

• UW winning out (not highly likely)
• have a two-loss team win the B1G (very likely)
• have a two-loss team win the Big 12 (somewhat likely)
• Miami not winning the ACC and hopefully taking a loss against Notre Dame (very likely)
• Alabama winning the SEC over Georgia (somewhat likely)

Oh, and Stanford beating Notre Dame in a few weeks would be helpful, too.

If you are so inclined, feel free to steal my structure, change the weightings and make up your own scenarios. I’d love to see what you come up with and to debate it as we savor whatever time we have left participating in this race for the playoff.