Thursday, March 15, 2018

My Final Thoughts on the Selection


Selection Sunday came and went again, leaving the majority of people unsatisfied.  While 2017 was extremely straight forward as far as who was getting, but the debate was left over how they were seeded, 2018 was the wild west.  My “bubble” list went 41 deep.  Some of these teams were the likes of Loyola and Murray State who had already secured bids.  Given that we were still dealing with the uncertainty of the Davidson/URI bid, and you could very well have made an argument that the loser of that game was going home, even if you felt URI was “safe” but not a lock, bubble volatility was high.

Basically, the goal was to select 13 teams from the pool 33 at-large candidates.  It was not easy.

My initial ranking system integrated 3 main components to for a Total Efficiency: Basic Efficiency (KenPom/Sagarin components), Strength of Efficiency (Some SoS/SoR components), and Trend, which puts emphasis on the last 5, 10 games.  This season, knowing more about how the NCAA uses its Team Sheets, a Quadrant Win Value (Q-Rate) was derived and applied to create a new Total Efficiency (New-T).  An addition qualifier was created called T-Factor, which applies the actual RPI to each team’s numbers.

For the most part, these were consistent with the NCAA selection process.  The Q-Rate was actually a huge discriminator for many of the committee’s selections, with a couple outliers.  First, the stamp of the RPI ruled many tie-breakers.  Second, it was clear that the only thing that was used to evaluate bubble teams outside outside of Q-Rate and RPI (T-Factor) was their standing with the NCAA.

Of the 33 Bubble Teams, Louisville Ranked 2 in T-Factor, 1 in New-T, but 20th in Q-Rate.  Overall, this averaged out to be the second best profile.  The best? Well that team was 1 in T-Factor, 2 in New-T, and 17th in Q-Rate.  Their blind profile was essentially identical, only one team received an 8 seed (2-9 in Q1 games) and Louisville (3-10 Q1 games) managed to not even make the First Four Out.  Sure, that had nothing to do with the ongoing FBI investigations.  Ironically, in the development of the tool, I had deliberately shaded gray any team that was involved, and included other teams that had postseason bans were in the eye, like Georgia Tech, UConn, Michigan State, etc.  These pictures shows the Cardinals and Blind Team Two well ahead of the field, with Penn State as being good enough to compete, but woefully lacking win quality (33rd).




When sorting by T-Factor, Middle Tennessee creeps back into the discussion, while NC State (and OK State) fall to the bottom of the bin.  Knowing that the committee only uses the RPI when it is convenient, I didn’t want to put too much stock in to these.  Instead, I really felt that the Q-Rate would tell the story.  

NC State and OK State perform well here, and I had been touting the Cowboys for a couple weeks as a sleeper (before they beat Kansas again).  NC State I was a little cooler than most.  I never moved them above a low 10, and really felt that their resume as a whole (wildly inconsistent) looked more like a play in candidate than a true high bubble.  Even after the BC loss, I felt they were safe, until chaos ensued in the MWC.  when the numbers settled Sunday, NC State was not in great shape, but the Win Quality was great.



Now sorting by an average rank of the Q-Rate, T-New, and T-Factor, I got this list, and we rolled with it to narrow the list, Missouri Oklahoma, and KSU went in quickly.  After that, I held off on Creighton, since they were in the same boat as Louisville. Next three were pretty easily FSU, VT, and Alabama.  We acknowledged Providence’s run made them safe, and then were torn between the other Catholic schools: Creighton, Saint Mary’s, and Saint Bonaventure.  We took Q-Rate as the primary discriminator and ranked them Bonnies, Gaels, and Jays.



Now we were down to slots, with Davidson lurking to snipe one, so we took a break.  When we came back, URI had yet to pull away so we ranked the next group of teams.  Some of the trends were easy to spot. NC State and OK State had similar resumes (good wins, poor SoS, questionable losses, poor RPI). Baylor and Notre Dame had similar resumes (strong metrics, poor wins). UCLA and Louisville had similar RPIs and win scores, but while UCLA got there by quality, Louisville by quantity.  Syracuse and Middle Tennessee had non-conference RPI numbers that kept them in the hunt.

When the numbers were all on the table, we ranked the bubble from there:
UCLA
Louisville
Middle Tennessee
NC State
Oklahoma State
Baylor
Notre Dame
Marquette
Western Kentucky
Penn State
Arizona State

The argument was that NC State was either in or out.  Same with OK State, as those Quality Wins would either carry them, or the RPI would be the bible.  We went RPI, with a caveat, as UCLA had a stronger finish, they were “safer” than Louisville.  Middle Tennessee was the “wait to see what Davidson does.” mid major, that would be cast out by a mid-major.

What did we learn?

  1. The NCAA does not use the numbers.  The keep them in the safe and pull them out when there is a talking point.
  2. There was heavy bias against the FBI teams.  For the most part, I understand that Miami and Larrannaga have been mostly cleared, but have to think that if their numbers dipped further, they were at risk.  USC was in low 30s overall in both New-T and T-Factor (basically showing that their RPI was backed up by metrics) and a Q-rate similar to Providence and Oklahoma State, finished second in their conference, played in their conference championship game, and were kicked to the curb.  That is completely indefensible without cause.
  3. Bad losses do not matter for major conference teams.  The matter doubly for Mid-majors.
  4. Q-Rate is definitely a factor, but needs to be tweaked
  5. RPI is king.  Xavier was anywhere from 6th to 14th in most metrics, but finished 3rd in RPI and 5th in Q-rate.  Only justification they end up on the 1 line.  Everything else pointed to lowest 2 seed.
  6. They NCAA will ignore every other rating and metric to choose a media narrative, and they REALLY like the Bobby Hurley story.
  7. “Body of Work” is the “Football Move” or “Becomes a Runner” of the Bracketology vernacular.
  8. The Bracket is pretty much set by Friday.  

The biggest mistake I made was Louisville over NC State. No excuses. Looking back through the last 2 weeks of brackets, everything I had SCREAMED NC State, with Louisville as the last team in, if at all.

Saint Mary’s was justifiably left off based on the Quality Win sniff test.  The difficult part of Q-Rate is that it treats a RPI 1 away win the same as an RPI 50 neutral win… they are not.  Next year, we will do what the NCAA did and lock in a many lines as possible on the seed list by Friday or Saturday, then reweigh the Wins for the 20-30 bubble candidates. However, MTSU was a much better fit here than Syracuse.

Comfortable missing USC for Arizona State.  What a travesty reach by the committee.  I mean, did you watch that ASU/Syracuse game?  It looked like two 14 seeds slugging it out.  I get that Syracuse plays the zone well, but Boston fucking College just hung 85 on them.  It ain’t that tough to solve.  The Sun Devils were a 9th place team in a mid-major caliber conference, and they were put on display for a national audience… thanks NCAA.

No comments:

Post a Comment