Thursday, January 18, 2018

A quick word on selection committee reform

Here, glance at this first:

Not many new details, but just an acknowledgement that change is being discussed.  Two points of legitimate concern are in the details, though.

1) Should ESPN (or any other prominent media member/partner) have their metric included in this?  Conflict of interest is extraordinary high, even if the formula is known and independent.

2) Some of the metrics are predictive of future results, while others attempt to quantify only results that have already happened.  You just can't blend the two together.  In order to have a useful composite, the data you're including needs to be of the same format (using the same data) and needs to have a common goal.

The NCAA has said in the past they review resumes and try to prevent projecting out what happens in a bracket.  Some of these metrics are designed with the specific goal of projection.  If the NCAA wants to use these, fine, but it must acknowledge the change in the selection process publicly.

Two things to think about as discussion continues.


Unknown said...

One worrying thing to me about using forward predicting stats is that they undervalue wins and losses, which should always be important in sports. Example:

Team X beats Team A by 21, loses to Team B by 1.
Team Y beats Team A by 14, beats Team B by 6.

Each of these teams has a +10 Efficiency Rating, so according tho Kenpom they're the same. And we probably should expect the same from each moving forward, but shouldn't the loss hurt Team X's resume? If we just went by raw efficiency, then the bucket that lets you win by 20 instead of 18 is just as valuable to your resume as the bucket that wins you the game by 1 instead of losing by 1. I can't imagine why anyone would want to watch sports that way.

HenryMuto said...

Assuming they change stuff down the line I wonder how much that will wreck how good I have been doing lately at predictions.