I’ve been a little puzzled this year by what I perceive as an undeserved negative perception of this year’s basketball team, and I’m wondering if I’m the only one?
This notion was triggered over time as I watched the team’s ranking in the human polls plummet during Lo’s absence, all while the RPI stayed relatively unchanged and much stronger.
Couple of days ago I noticed it again, but this time I realized that the team’s RPI was high 20′s and yet they weren’t getting a single vote in either the AP or Coaches’ polls.
I decided to build a comparison between rankings based on polls of sportswriter and coaches (which I really don’t like) compared against rankings based on programmatic indices (which I do like), and to identify teams that show the most variance from one to the other.
Methodology: To create the population of teams for analysis, I included any team that is mentioned with even a single vote in either the AP or Coaches’ poll. I added any additional teams that were present in the top 30 of the RPI ratings and ESPN’s BPI ratings. This gave me a population of 46 teams.
There were some gaps in the data. For example, San Diego State is 34th in the Coaches’ poll but absent from the AP poll. For these cases, to keep the analysis as fair as I could conceive while not leaving any population teams out, here’s what I did. Let’s stick with the AP poll. This poll contained 39 teams. That means there were 6 teams in my population that weren’t included. For the sake of this analysis, I assumed that they all tied for the next spot in the AP poll and assigned them all a ranking of 40th. I applied the same approach with the Coaches’ poll. Interestingly at this point I noticed that my population included only 3 teams that were unranked in either “people” poll: Colorado, Wichita State and (you guessed it!) NC State.
To simplify a bit (this is necessary when you have a simple mind, like mine), I then created a poll average based on the two “people” polls, and an index average based on the two index polls. I then subtracted the computer index average from the people poll average to arrive at my differential index. So negative numbers means that the team is “over-rated” by the people polls (according to the computer polls) and vice-versa.
Here’s the top of the list, showing the most “over-rated” teams:
Some interesting results, but K-State grabs my attention from that list. Their poll average projects a much higher seed than the index average. Depending on where they end up, especially if they have a first round matchup with an under-rated lower seed, I’ll definitely consider them an upset possibility.
That right, boys and girls, friends and relatives, #1 on this list is the NC State Wolfpack. No other team is further apart in the minds of the collective sports writers and coaches vs. these computer indices. Surprise!
The other team here that grabs my attention is Colorado State. They’ve flown under the radar (at least under MY radar) this year, and I want to know a little more about them. I’ll take a really hard look at who they’re matched against this year.
What does it mean?
My take on this is that this year’s Wolfpack team has been punished by the media/coaches for not living up to the insanely optimistic pre-season expectations that were promulgated by….those same media/coaches. What’s your theory?
As I type this, I also wonder if I included other computer indices (say Pomeroy and/or Sagarin) would the results be substantially different? If I have time, I’ll give it a shot.