Why this week’s poll speaks volumes on the 2017 champion


Indiana is ranked No. 13 in week four. Could be good to know later…. (AJ Mast, AP)

We are now in week four of the college basketball season, and here are the top 13 teams from the latest AP poll:

1.  Kentucky
2.  Villanova
3.  North Carolina
4.  Kansas
5.  Duke
6.  Virginia
7.  Xavier
8.  Gonzaga
9.  Baylor
10. Creighton
11. UCLA
12. Saint Mary’s
13. Indiana

You might be asking why I brought the curtain down on the nation’s top teams at No. 13 instead of an equally arbitrary but more customary number ending in a zero or a five.

Let me stress the word “arbitrary,” but, for now, here’s a fact worth pondering:

Every year since 2004, the eventual national champion has come from one of the top 13 teams in week four’s AP poll.

It turns out these weekly spasmodic seismographs that rather presumptuously recast the nation’s longest current winning streaks as something called rankings may have some predictive value after all. And that value achieves a new level of evaluative heft each year when we put Thanksgiving behind us.

The week-four rankings have actually outperformed same-season polls from the preseason and from week three when it comes to picking the eventual title-winner.

                                       AP rank
      Eventual champion    Preseason   Week 3   Week 4
2016  Villanova                11         8        8
2015  Duke                      4         4        4
2014  Connecticut              18        18       13
2013  Louisville                2         2        5
2012  Kentucky                  2         2        1
2011  Connecticut              37        36        7 
2010  Duke                      9         7        6
2009  North Carolina            1         1        1
2008  Kansas                    4         4        4
2007  Florida                   1         1        4                    
2006  Florida                  41        14       11
2005  North Carolina            4         9        8
2004  Connecticut               1         3        2
Average                       10.4       8.4      5.7 

It will be noted that the statistical difference in ranking of nascent title-winners between these polls comes down to just two extreme outliers: Florida in 2006, and Connecticut in 2011.

Well, that’s kind of the point. Two seasons out of 13 is a non-negligible frequency, after all, and in the here and now we have again seen a team vault from zero-vote oblivion all the way to the AP top 10 in just four weeks. Baylor may turn out to be a flash in the pan, who knows, but on the edge of December the Bears are at a minimum a fit subject for evaluation. We didn’t know that before.

The Old Faithful of college hoops analysis, the Thanksgiving epiphany
Extreme outliers do happen, and past history confirms what common sense suggests, namely, that we’re way better at identifying black swans at the end of November than we are at the end of October.

Besides, the week-four advantage here isn’t all about the Joakim Noah-era Gators and Kemba Walker-led UConn. Nine of the 13 eventual champions listed here did improve their rankings over the course of November or kept them right where they had been in the preseason. Only four teams fell in the rankings over those four weeks and then rallied to win it all the following April.

Now for the full disclosure. This Thanksgiving epiphany thing denotes a streak of some duration and (I dare say) a strong tendency, but, of course, it is no iron law. Syracuse in 2003 was ranked No. 52 in week four. So if you’re way down in the fine print this week, dream big. Maybe Colorado, Rutgers, or San Diego State can do something that hasn’t happened in 14 years.

(For the record the 2003 preseason poll really didn’t perform much better with respect to Jim Boeheim’s group than the week-four rankings did, pegging the Orange at No. 29. Note that prior to that season my ESPN.com archive of weekly AP polls vanishes into the oubliette of the far distant Juan Dixon and Lonny Baxter past, back when it took 12 minutes to download a song and the NCAA acted as though the RPI has analytic value.)

Why are we, as college basketball evaluators, so much smarter and more clairvoyant on November 30 than we were just four weeks prior to that? I’ll leave that Aristotelian query to my colleague Jeff Goodman, who, in this instance, speaks for us all….

Precisely. I ranked Connecticut No. 39 in the nation just 22 days ago, an unduly rosy evaluation that laptops from near and far now roundly mock and ridicule. Sure, the Huskies constitute a huge, loud, thunderous, egregious evaluative outlier (they very often do, don’t they, one way or another?), but when you’re talking about populations of 300-some teams there are always going to be UConn-variety erratics.

Getting back to that word “arbitrary,” in the literal sense there is, of course, nothing magical about being ranked No. 13 at the end of November (Indiana) versus being placed one spot lower (Louisville). Actually the No. 13 team that set this boundary for us, UConn in 2014, came within just 12 votes of being ranked No. 12 instead in the fourth week of that season. This stuff isn’t Newtonian; think of the Thanksgiving epiphany instead as a handy rule of thumb.

Purely in terms of locating the eventual champion, week four is when the rankings have traditionally made the jump from work-in-progress to clear signal. In fact if the past 13 seasons are any indication, the poll this week (average ranking of eventual champion: 5.7) is already the functional equivalent of the perfect-information one that will come out on March 13, 2017, the day after Selection Sunday (5.5).

Naturally, we’re far from omniscient just four weeks into the season. When March rolls around and I’ve seen way more basketball, I’ll whittle my national champion shortlist down to a mere eight teams. Still, Baylor and UConn have now happened in their diametrically opposed fashions, meaning we’ve just made a very rapid and steep ascent up the 2016-17 learning curve. For now I’ll ride with the AP top 13.

The 2017 title will be won by one of the following teams: Kentucky, Villanova, North Carolina, Kansas, Duke, Virginia, Xavier, Gonzaga, Baylor, Creighton, UCLA, Saint Mary’s or Indiana. Unless a latter-day Carmelo arises on one of the other 338 teams, I have history on my side.