College Basketball: What does the AP Poll Tell Us?

College basketball offers fans a less predictable postseason and champion than most other major sports. There are likely several reasons for this (single elimination tournament, 6 rounds of tournament play, wide variance in roster experience). However, for the people that follow the sport regularly, a natural question arises: does anyone know anything about who's going to win? And, does the regular season tell us anything?

In order to answer the questions, the AP poll is probably the best source of information. Since it offers rankings at the preseason, mid-season (before conference play), and end of season (pre-tournament), what does it say about our collective knowledge of college basketball over the last ten seasons?

PRESEASON SUCCESS: 73%
MIDSEASON SUCCESS: 71%
PRETOURNEY SUCCESS: 81%

Over the last ten years, the AP top 20 basically yielded six of the Elite Eight teams before a game was played. Through midseason (basically, right before conference play starts), that number did not improve; in fact, it was slightly worse. Bottom line: at this level, conventional wisdom seems neither good nor bad.

PRESEASON SUCCESS: 55%
MIDSEASON SUCCESS: 53%
PRETOURNEY SUCCESS: 68%

This looks like the most varied level of success for the poll. The preseason success was higher than the midseason success, and the pretourney number was the highest of them all. But it's so inconsistent: in two years (2011 and 2006), no preseason top 10 team made the Final Four. In the same years, only one top 10 team from the last poll made the Final Four. Apparently, the difference between winning three tourney games and winning four is real.

For picking the national champion, the AP poll has some interesting results. In the preseason poll, three out of the last ten champions were unranked: 2011 (UConn), 2006 (Florida), and 2003 (Syracuse). In fact, all three remained outside the top ten (#21, #11, #11, respectively) going into the tourney.

The most interesting part of this is the national champions from the top 5 in the poll. Given the variability seen in the Elite Eight and Final Four participants, it's comparatively impressive to see that six champions came out of the top 5 in preseason, and eight out of the midseason top 5.

Why is the pre-tourney number so low? It's hard to say for sure. The easiest answer is probably the idea that the polls reflect record more than ability as the season drags on, and teams that lose a game are very likely to fall (only five of ten national champions won their conference tournaments). But that does not reflect a diminished ability to win it all come March.

CONCLUSIONS

The AP poll is likely the best reflection of popular wisdom regarding college basketball, but it's far from perfect as a predictive tool.

• Whether it's November, January, or March, it's likely we'll get at least one Final Four team that "doesn't belong" based on the polls, and probably more. That is and hopefully always will be the best part of March Madness.
• As far as the Elite Eight and Final Four go, we collectively know nothing in week 8 that we didn't know before the season started. In fact, we might know a little less. It's almost enough to say that pre-January games don't matter at all, except...
• Despite the single-elimination, six-wins required gauntlet of the tournament, the top 5 teams in the AP poll matter, even in preseason, and maybe even more in midseason. In fact, the top 5 AP teams in week 8 have won more national titles in the last decade than the #1 seeds. There's no way that's a sure thing, but regardless, I'll probably take a look at the odds on Duke, Michigan, Arizona, Louisville, and Indiana to win it all come March, no matter what happens between now and then.