myDeaconmyhand
First man to get a team of horses up Bear Mountain
Ultimately Georgia Tech and Nevada’s NET rankings are based on an entire body of work, and the 1 close result between those teams is only so meaningful.
I think you are overly trivializing the “probability difference” between Q1 and Q3 opponents. There is no way to evaluate schedule difficulty without making arbitrary distinctions, and as much as you might disagree with the 4 quadrant system, it’s almost certain that whatever method you would prefer is just as arbitrary. Say you do away with quadrants - do you average opponents NET to determine overall schedule difficulty, then you could argue that averages are skewed at the margins, and so on. Do you create more quadrants? Less?
Bracket Watch!
Still in the last 4 out with UVA directly behind. Interestingly, Miami and VaTech are two of the last four in.
As opposed to 20 games for seeding in the conference tournament?
I assume an all-play tournament would have several play-in rounds before getting to the round of 64.
I agree wholeheartedly with all of this except 1. More data points is always better.If I’m designing the system, I do away with the Quad system altogether. The NET already accounts for schedule strength (as does KP, Torvik, etc) and the NET also has the secret adjustment for wins. So by also looking at “Quad 1” record, we’re double counting the impact of wins over good teams.
But we’re probably not going to do away with Quad 1 wins, because human beings like to look at simple stats. There are still ways we could tweak the system.
1. We could and should ignore the H2H impact of any particular game on the corresponding NET positions of both participants. Example - WF beat VPI. How strong is VPI? We should evaluate VPI vs every other team in the country while ignoring the WF-VPI result. If at the end of the year, VPI is 33rd in NET, but would have been top 30 if you exclude the WF game, that’s a Q1 win for WF. Similarly, if Louisville ends up near the Q3-Q4 border, Miami could potentially have a Q4 loss while others who played UL could get Q3 credit (for example). That’s impossible to do manually but it would be easy to program.
2. We could increase the number of breakpoints such that each set of teams are more similar than what we currently see.
3. We could have a “wins adjustment” where we give credit to wins based on the relative strength of each team (again, in theory we’re already doing this in the NET although the exact formula is a secret). If I’m designing the “wins adjustment” I’m giving credit directly corresponding to the relative strength of every team, 1-362.
All easy to implement ideas and all better than, oh congrats, you got to play the #25 team and after you beat them, they’re still top 30.
It's really even more than double counted:There is an argument that double counting wins over good teams is a feature of the current system. You want to be recognized as a "good team?" OK. Beat good teams and get rewarded.
Morning all. One day away from Wake Forest-Virginia, which means another edition of both the BOTG Pod and our previews on BSD.
All your traditional stuff in the preview: game info, history, the metrics, roster analysis, my take, keys to the game and prediction. Plus, some notes on how this season is just beginning for Wake Forest. You can read all that here.
EP 23 of the BOTG Pod has a very special guest, Conor O'Neill, who joins me for 25 minutes of refs, replays, Alabama football, the metrics and, of course, Wake Forest basketball. That segment is sandwiched by my thoughts on the FSU game and my Virginia preview. Listen below:
As always, thanks for listening/reading!
If folks are open to a BOTG thread, I'm certainly ok with making one. Being the new person/journalist, didn't want to change the flow of how things work here. But open to any ideas folks have. I'm incredibly appreciative of all the support from the OGBoards.I would not mind this having its own thread. Easier to follow/update
I 100% would like that!If folks are open to a BOTG thread, I'm certainly ok with making one. Being the new person/journalist, didn't want to change the flow of how things work here. But open to any ideas folks have. I'm incredibly appreciative of all the support from the OGBoards.