An Explanation of TRMDi

WHAT, WHY, AND HOW

The rankings I develop are meant to both create a systematic comparison between teams, while also creating a field for the NCAA tournament that aligns with how I think the Selection Committee will act (not necessarily what they say).  The calculation is made up of three components.  The decision to include these specific measurements comes from my projection of how the Selection Committee will ultimately select teams, not necessarily because I believe this is the best way to predicatively measure future games between two specific teams.  (Editorial note, I'd prefer more weighting to home/road games, and the elimination of an arbitrary cut-off for top wins).

RPI:
The first is the RPI calculation, as defined here.  In short, a team’s winning percentage accounts for 25% of the calculation.  A team’s opponents’ winning percentage account for 50%, and a team’s opponents’ opponents’ winning percentage the final 25%.  The team’s winning percentage is also weighted with home wins and road losses at 60%, while home losses and road wins are weighted at 140%.  

BOOST:
The second is point differential (what I call Boost), with margins grouped into 12 tiers.  There are 6 winning tiers and 6 losing tiers.  This is a simplistic approach, but the intention here is to provide some balance to the RPI and big wins, and just speak to a team’s general strength in relation to their opponents.  Teams that have lopsided records in close games one way or will see an adjustment/regression in the boost.  In short, this is the “eye test” of TRMDi.

TOP 50 WINS:
The third is top 50 win share (using RPI).  Previous years I used a composite of top 50 wins, top 100 wins, losses to 100+ and losses to 200+, but the leading indicator most recently has been to prize the top 50 wins.



2 comments:

  1. if they committee next year is planning on using the new metrics to place teams into the tourney this is going to impact pretty much everyone's projections. Next year should be very interesting.

    ReplyDelete
  2. Agreed, and the most difficult part will likely be separating true new parameters vs lip service. That said, improvements to the measuring tools can only help in the long run of the tournament.

    ReplyDelete