#KPI Blog: Evaluating the #KPI

Posted on by

Hello from beautiful Spokane.  Admittedly, it’s been a whirlwind couple of days as the #KPI computers have made their way from Indianapolis Sunday, to East Lansing late Sunday night via bus and to Spokane late Monday.  With Michigan State playing in the Big Ten Tournament championship game, there was no opportunity to update data prior to the Selection Show on Sunday (heck, I didn’t even make it to a television until the South Region was well announced!).

I felt badly that I didn’t get anything posted before Sunday’s Selection Show (though it was quite obvious that win or lose Sunday, nothing was getting posted).  I want to thank the folks at http://bracketmatrix.com/ for choosing the #KPI to be part of the formula. My ranking there is based as of math as of 7 a.m. on Sunday.  No human element, no update on Sunday.  And that’s ok and understood.  Their site is an incredible tool.

I have been keeping a 1-68 list separate from the #KPI.  That list got 67/68 teams correct.  40 teams were seeded correctly, 64 were within one seed line and the other 3 in that field were off by two seeds.  The average difference in seeding was .448.  I won’t publish that list until after the NCAA Tournament is over for several reasons.

Again, thank you for the incredible interest in the #KPI this year.  The interest has far exceeded my expectations and has been quite humbling.  It was quite fun getting texts and calls from people asking what their #KPI would be if they won this game or made it how far in their conference tournament.  There is still plenty more data coming as the month progresses.  I hope readers continue to be interested.

#KPI Headlines:

  • Greg Shaheen authored a fascinating look at bracketing through his Twitter timeline Monday night:

  • I had a fun look at some things to expect from the bracket (it may or may not have been published around 3:30 a.m. ET on Sunday morning): http://kpisports.net/2014/03/16/week-18-sunday-kpi-basketball-as-of-31614-7am-et/.
  • 19 of 31 conference tournaments were won by someone other than the No. 1 seed.
  • Scoring was up 2.87 PPG (up 4.39%) from 65.3 PPG to 68.2 PPG in conference tournament play this year when compared to 2012-13.  Week 19 (March 10-16) was the lowest scoring week of the season (67.4 PPG).  The next lowest total was 68.9 PPG (Jan 13-19).  Week 19 was also the lowest of the season for possessions (64.6), field goal attempts (53.9), 3-pt field goal attempts (17.7), assists (11.4), steals (5.6) and turnovers (11.3).  As has been said all year, the number of possessions (hence pace of play) has a lot to do with the decreases we have seen week to week.

Dissecting the #KPI formula how it relates to the actual NCAA Tournament seeding:

THE GOOD:

  • The top-8 teams in the #KPI all landed a 1 or 2 seed. Arizona, Florida, Wichita State, Wisconsin, Villanova and Michigan were all on the exact seed line.  Kansas was a No. 1 seed in the math while Virginia was a No. 2 seed. They were flipped.
  • All 16 teams seeded 1-4 appeared in the top-17 of the #KPI.  The lone exception was Mountain West Tournament champion New Mexico, who earned a 7 seed.
  • Of the last 18 AQ teams in the tournament (13 seed or below), 14 of 18 were seeded correctly.  Western Michigan and Delaware were swapped in the #KPI, as were Eastern Kentucky and North Carolina Central.
  • 50.8% of teams who made the tournament were placed on their exact seed line by #KPI math.  77.8% were exact or within one seed line while 88.9% were exact or within two seed lines.
  • The average differential between #KPI and tournament seeds was 0.825.
  • The last four teams below the cut line who made the NCAA Tournament: Tennessee, N.C. State, Iowa and Xavier (in that order).  All four of those teams are playing in Dayton Tuesday and Wednesday night.  The eight teams playing in Dayton are the exact four lowest #KPI AQ teams and the exact four lowest #KPI at-large teams based on who made the tournament.
  • Between the NCAA and NIT, the #KPI got 96 of 100 teams in the two tournaments (96%).  The #KPI had Georgia and LSU as the first two teams out of the NIT, while West Virginia and Indiana State were a few spots back.

THE NOT SO GOOD

  • I’ll admit that I wanted badly to land 68/68 teams this year.  It didn’t happen.  Because of how some things fell, I became at the mercy of certain teams winning their conference tournaments.  Green Bay, Louisiana Tech and Belmont all received automatic bids to the NIT, but were actually above my cut line.  Green Bay’s win over Virginia, Louisiana Tech’s win at Oklahoma and Belmont’s win at North Carolina were all great wins, but only enough to inflate their #KPI, not enough to get them in the tournament.
  • North Dakota State, Harvard and Stephen F. Austin were higher in the #KPI than the 12 seeds that each team received.
  • The #KPI math missed on New Mexico – the largest outlier at a difference of 4 seed lines.  Remove New Mexico from the equation, and the #KPI had 14 of the top 16 teams seeded correctly (with only Virginia and Kansas swapped on the first four lines).
  • Among at-large teams, the #KPI undervalued Colorado, Memphis and Kansas State.

This is the #KPI

Posted in Uncategorized and tagged .