Thursday, October 6, 2011

New for 2011: Undefeated ... but for how long?

Yes, it's time to preview some more new things for our blog this year. Technically this feature isn't really "new", but rather a promotion to regular appearance of a previously-rare feature. Last year we took two brief looks at the slate of undefeated teams, what their schedule looked like, and how likely they were to make it to the end of the regular season undefeated.

Well we've added that feature into our weekly rotation. Every Thursday from now until the end of the regular season we'll post an update on who is likely to keep their record unblemished, a game-by-game listing complete with projected final scores and odds of winning, special attention to the most difficult game remaining on their schedule, and any other notes.

The new twist on this data is the inclusion of treemaps to help visualize the relative odds of each team making it through undefeated. We've used treemaps before, but as a visualization of NCAA basketball tournament win shares. The bottom line is that the larger a team's square is, the better its odds are of going undefeated. At this point we don't take into account conference championships; this means we could be overestimating the odds that a team from the ACC, SEC, Pac-12, Big-10, or other two-division conference will go undefeated since they will have one more game against an as-yet-unknown quality opponent standing in their way of a perfect regular season.

We'll break the candidates into a few categories:

  • Legitimate title contenders;
  • Second-tier hopefuls;
  • Long shots;
  • Absolute phonies; and
  • What are YOU doing here??

From there we'll discuss their respective odds and track how their odds change as they win (or lose) and their upcoming opponents do. I doubt we'll see another charge like Auburn's last year, where they went from 100-to-1 odds of just making it to the SEC title game unscathed, but it'll be fun making pronouncements on par with Ken Pomeroy's "Kansas Will Not Win It All" in the meantime.