UConnSwag11
Storrs, CT The Mecca
- Joined
- Aug 26, 2011
- Messages
- 14,129
- Reaction Score
- 55,180
Jacobs has become a UConn troll extraordinaire.He seems strangely fixated on our demise.
And hopefully the NCAA Tournament Selection Committee also.We've absolutely confused the computers. And our fans. And our coaches.
Nothing confusing lately when you’re losing with consistency. Let’s get after it on TuesdayWe've absolutely confused the computers. And our fans. And our coaches.
The wild thing is (as pointed out by Gary Parrish on the CBS podcast) Torvik let's you sort the rating by date, and if you start at the start of our 6/8 losing streak until now only... we still rate as the 24th best team in the country. Pretty wild.Nothing confusing lately when you’re losing with consistency. Let’s get after it on Tuesday
Yes, we’ve lost a couple games that the statistics show we should have won 99% of the time based on shot quality and opponent’s shot quality. At least 3 losses heavily influenced by extremely lopsided officiating which is part of the reason the metrics show us as still being better than our record. Not blaming the refs, but the data is there and speaks for itself.The wild thing is (as pointed out by Gary Parrish on the CBS podcast) Torvik let's you sort the rating by date, and if you start at the start of our 6/8 losing streak until now only... we still rate as the 24th best team in the country. Pretty wild.
NET is difficult to understand at times. I'm still not sure how we're holding at 7. Figured we'd be around 10-11 by now.
Yesterday Clemson beat Q3 Florida St. by one on the road and dropped three spots while Florida St. gained three. Duke beat a Q3 GaTech team on the road (albeit by a huge margin) and gained eight spots while GaTech dropped 29. Pitt beat #20 Miami at home and stayed at 62 while Miami dropped nine spots for that Q1 loss.
While I neglected to specifically say so, the games I cited were all ACC and these are teams that play each other and have all already played each other, so there would be direct ramifications between them. Even noting your boldface comment, it still doesn't appear to make sense.Too many factors besides a single game. It's formula is pretty in-depth overall. Better than RPI. Plus it's not like you are talking about things in a bubble with a static group of comparison teams. Did the teams around those teams win/lose. Did they play poorly at home, did the teams they had "good" wins/losses against become worse. Etc. Unlike voters who won't move a team down without a loss, the computer will jump teams all the time if the data changes.
"NET takes into account game results, strength of schedule, game location, scoring margin, net offensive and defensive efficiency, and the quality of wins and losses. It's both a results-driven and predictive metric."
On KenPom, Pitt was favored by 1, so a 3 point margin is essentially a non-mover (2 point difference in expectation). Miami ended up moving up because Florida missed their projection by 9 points, Utah missed by 10 points, etc.While I neglected to specifically say so, the games I cited were all ACC and these are teams that play each other and have all already played each other, so there would be direct ramifications between them. Even noting your boldface comment, it still doesn't appear to make sense.
I follow Pitt as my wife went there. They beat the #20 AP team (albeit ranked 46 in NET) by 3 at home, Miami went up 1 spot in NET, Pitt stayed at 62 with no other movement in teams 61-70. Meantime Penn St. beat Michigan (was 75 NET) at home by 22, went up 8 spots in NET while Michigan dropped seven spots. It simply seems to me that parts of the algorithm are either overweighted or underweighted.
Thanks for the explanation.On KenPom, Pitt was favored by 1, so a 3 point margin is essentially a non-mover (2 point difference in expectation). Miami ended up moving up because Florida missed their projection by 9 points, Utah missed by 10 points, etc.
Penn St beat the projection by 17 points. Led to large movement.
As always, the margin is the most important aspect. Basically for the NET think of it as "margin vs. expected with a bonus for a win", whereas KenPom and Torvik have no bonus for the win, though I think Torvik does factor in how close it was throughout. Haslametrics has a bonus for making the game noncompetitive earlier. Forget if Miya has any quirks.