Friday, January 19, 2018

Middle of the road

Last year, my bracket numbers were all over the place.  My formula was immature.  Metrics were poorly chosen. I had to hand-jam numbers so they were often not up to date (like, I would have never been tracking LSU).  My bracket last season typically had most of the correct at large teams,  but many were under or over seeded based on public perception.

This year, with better use of specific components of the RPI and efficiency metrics, my bracket has been more consistent week to week.  The seedings also fall in line with the rest of the matrix.  Check it out: no blue or red outside the top line (last week Texas Tech crept ahead of Duke, so I slid them into a 1 spot... oops).

Also, the formula has done a better job of balancing recent success with the entire body of work.  Ohio State may look overseeded right now based on their preseason perception, but their overall body of work is just as strong as their recent hot streak.  It does lean heavily on NCSOS, so this will damped the effect of a hot 3 game homestand against conference foes.  The RPI still gives a little too much credit to mid-pack power 5 teams that get trounced against all good teams, but steal enough home victories to remain middle of the pack.  Those miserable losses are enough to keep them relevant in the RPI, but often they are floating in line with top mid-majors by other metrics.

It will be interesting to see how this keeps up as more conference data is accumulated.

No comments:

Post a Comment