Changes That You Think Would Improve NCAA Tournament Seeding Process

awhom111

Member
Maybe it just felt like it as a reader, but it seemed like we had more complaining than any other season about the selection and seeding process during the past season. With the Olympics just finishing and us getting eased into the Fall sports at Duke, I figured that this would be a good time to start a thread for ideas and upsides and downsides of various changes. I know that some changes would never happen because you can't put toothpaste back in the tube so my hatred of midseason releases of brackets if the postseason started right then will continue since it is pretty obvious from TV ratings that fans cannot resist watching them.

Here are some ideas that I have thought about over the past year that might add fairness to the process or not as well as some honesty about how likely they would ever be accepted (pretty much never):

As mentioned above, I think that midseason ranking reviews bias the final rankings because of our general obsession with keeping teams ranked in place until they lose. This is before even getting into the fact that schedules are arranged systematically and the games are not just shuffled arbitrarily, not to mention our favorite griping about ACC schools losing non-conference games, almost all of which happen at the beginning of the season.

One idea that I think would be interesting to try, maybe not in basketball at first, would be to not have the committee gather and create a consensus ranking. Instead, each member would submit their full field and ranking after the last game and then the rankings would be combined into the official order. A computer, checked by a few neutral people, would then place the teams into the bracket according to all of the existing principles. That could actually create a pretty fun bracket reveal show, but I get the sense that the committee actually likes all the arguing and, more importantly, the power conference horse-trading.

That leads to the next point where I would love to see a committee consisting of people who we actually believe have the time to watch enough games to have some knowledge about all of teams involved instead of us knowing that they all have other important job duties. Of course, I cannot see the people with the most power relinquishing any of it to people with less active power. It would be great to have a committee with as little potential conflicts of interest as possible, but that is not happening either. We still get college football polls released soon enough after the last game of the week where there is just no plausible way that voters could see at least any significant amount of each game that week featuring team that would be in consideration for a Top 25.
 
My suggestion is fairly simple: any team that doesn’t play a meaningful non-conference game after Feb 1 gets their Net ranking moved down several spots. It would help rebalance the metrics between conferences.

-jk
 
I'll make it very easy for all sports. Committee members must not have any direct interest with any school under consideration -- AD, alumnus, etc. I'm sure a former Denison basketball player that is now the AD at Amherst (hypothetical) can judge the strength and quality of a record better than most ADs on the D1 level.
 
Adding on to the ideas. Separate the selection process of who gets in from seeding. The who gets in then gets input from the fans who watch a lot of bball. The group that does the seeding gets tweaked to include the media who also watch a lot of games. The seeding group also has no restrictions on who plays who and when. No rules about having played a team before or if teams are in the same conference or travel distance.
 
Along the lines of what the Moderator says, I would weight game results after 2/1 greater than prior to that. November and early December games mean very little compared to an early March game. This might enable teams to play more young players early in the season without fear of a loss coming affecting seeding that much.

Also, I would value neutral floor wins and away wins more than home wins. For example, our loss at Arkansas last year would never have happened at a neutral floor and frankly a team that comes into Cameron and wins should get extra credit for that win.
 
Ban Joe Lunardi from the face of the earth.

Don't have those weekly updates. They just confuse things and make it worse. And one would think these people have better things to do than flying weekly to wherever they meet (Dallas?). What a waste of time and money.
Here here.

Lunardi is a charlatan of the highest order.

"If the season ended today" is a crappy disclaimer to anything that follows. Especially if you utter that nonsense in December.
 
Don't follow recognizably bad data out the window (only inter-conference results in 1st ten games). It's the chicken's way out. Use judgment or heuristics (past results in NCAAs) to reach a sensible conclusion.

Oh, by the way, seedings are probably more important than bids.
 
Have more inter conference games later in the season, maybe?

If they could design a better algorithm, I would be ok with taking the human element completely out of the equation and let AI make the brackets.

If they want to keep the humans, I wonder what would happen lf they took the team names off and only allowed them to look at the team resumes in a blinded fashion, both for inclusion in the tournament and for seeding? When it was unmasked, I think they would be surprised by what they found.
 
Have more inter conference games later in the season, maybe?

If they could design a better algorithm, I would be ok with taking the human element completely out of the equation and let AI make the brackets.

If they want to keep the humans, I wonder what would happen lf they took the team names off and only allowed them to look at the team resumes in a blinded fashion, both for inclusion in the tournament and for seeding? When it was unmasked, I think they would be surprised by what they found.
I think doing an initial blind seeding would be a nice improvement. I expect the committee will want to weigh in using their opinion of the teams (eye test), but using an objective starting point would help force people to justify where their opinions differ.

Re: other opinions some people have posted, I would not support weighting games later in the year higher, as that disfavors the non-power conference teams, and they get screwed over enough as it is. Similarly, I don't think it makes sense to give any weight to past results in the NCAAs.
 
My suggestion is fairly simple: any team that doesn’t play a meaningful non-conference game after Feb 1 gets their Net ranking moved down several spots. It would help rebalance the metrics between conferences.

-jk
Just about anything to get teams to play meaningful non-conference games after Jan 1 should help.

My personal out-of-the-box idea would be to have a series of in-season tournaments during the off week between the NFL conference championships and Super Bowl. This would not only improve tourney selection but also help capture the sports world's attention before March. Perhaps the winners of these in season tournaments would get automatic NCAA tournament bids.

Another idea would be to ignore Quad 4 games when computing a (power conference) team's NET ranking. It shouldn't matter whether you beat Stonehill College by 25 points or 50 points, but it does in the NET. The Big 12 seemed to benefit from this "bug" last year.
 
I would get rid of the Committee entirely. With a field of candidates as large as they are evaluating, and the Committee made up of ADs or Conference Commissioners (largely non-basketball experts) who are doing this as a side hobby -- essentially no more qualified than anyone on this Board -- it makes no sense to rely on the Committee's subjective evaluations and deliberations. Rather than try to tinker around the edges with helping the Committee make better decisions, I would remove them entirely in favor of an objective, data-based, transparent measure that does "selection" of the 37 at-large teams for the field and ranks the full field of 68 in order and then stops there, with the Selection Sunday show then turned into a draft (in rank order) for "seeding"/"bracketing" (subject to rules to ensure some level of equality between the regions).

The Committee currently receives "team sheets" that contain 7 ranking metrics, composed of:
1. NET Rankings.
2. Commercially-available predictive (efficiency, adjusted for schedule and home/road game location) metrics: KenPom, ESPN's BPI and, added this year, Torvik.
3. Commercially-available results/resume (relative wins/losses performance compared to how an average team would have done against your schedule) metrics: KPI, ESPN Strength of Record, and, added this year, Torvik Wins Above Bubble.

Why not just use an average of those 7 rankings as the ranking, and it would be published every day so teams would know exactly where they stood in terms of what they'd need to do to qualify for an at-large bid if they don't win their conference tournament?

Or, if you think there's different, better objective rankings to use, pick those (or get rid of the Net Rankings entirely --although the thing most criticized about the Net Rankings, the "quadrants," actually has nothing to do with the rankings themselves, only how they are currently used), or modify some features of these 7 if they can be improved in some way. As long as it is agreed to prior to the season and is then transparent, so everyone can see how it works, it would be better than a Committee's subjective eye test.

The Torvik predictive metrics have some components, described further here https://www.on3.com/news/ncaa-annou...ainst-bubble-metrics-for-tournament-seedings/, that address some of the points others have made above re: weighting results later in the season more heavily and re: not wanting to incentivize blowouts too much:

For example, in Torvik's efficiency-based metrics "[a]ll games in the last 40 days count 100%, then degrade 1% per day until they’re 80 days old, after which all games count 60%.” (I don't think this is unfair to the smaller conference teams because the efficiency metrics for any given game are adjusted by opponent/location -- it only means each given game result counts for slightly more the more recent it is).

And, in addition, "once a game is well in hand for the much better team, Torvik doesn’t view that 'garbage time' as quite as important as the more competitive parts of the game, and therefore discounts the numbers from garbage time."
 
That leads to the next point where I would love to see a committee consisting of people who we actually believe have the time to watch enough games to have some knowledge about all of teams involved instead of us knowing that they all have other important job duties.
And who would be an example of a person with that kind of time and focus? Joe Lunardi. I think we would all agree that he and his ilk are not who we want allocating bids or doG forbid seeding teams. (a little snark, didn’t mean to rag on you specifically)

This is a tough but good question. I like the idea of blind evaluations; it has the air of objectivity. Likewise compelled recusal for people with any conflict of interest. DQ’ing the CHeats makes sense too, obviously.

The inherent problem, however, is that it is literally impossible to determine the actual best 68, or whatever number they go with, teams and in what order. The tiny sample size of games played, the lack of head to head competition, the fact that we’re talking about kids that could have an off day that totally undercuts a whole season’s worth of statistics, among other reasons, truly renders this task the stuff of barroom arguments, not accurate algorithms. Maybe that’s why the tournament is so popular

Go Duke! 9f!
 
And who would be an example of a person with that kind of time and focus? Joe Lunardi. I think we would all agree that he and his ilk are not who we want allocating bids or doG forbid seeding teams. (a little snark, didn’t mean to rag on you specifically)

This is a tough but good question. I like the idea of blind evaluations; it has the air of objectivity. Likewise compelled recusal for people with any conflict of interest. DQ’ing the CHeats makes sense too, obviously.

The inherent problem, however, is that it is literally impossible to determine the actual best 68, or whatever number they go with, teams and in what order. The tiny sample size of games played, the lack of head to head competition, the fact that we’re talking about kids that could have an off day that totally undercuts a whole season’s worth of statistics, among other reasons, truly renders this task the stuff of barroom arguments, not accurate algorithms. Maybe that’s why the tournament is so popular

Go Duke! 9f!

An example of a person with that kind of time and focus would be Sam Vecenie, currently of the Athletic. This is a guy who watches film and tape and is breaking it down seemingly 24/7, and he is a very, very smart basketball analyst. He has more ability to analyze basketball players and teams in his pinky toe than these committees made up of ADs have in their entire bodies. By a lot. And he has no dog in the hunt. A room of guys like him would be great. And I'm not talking about college basketball reporters like Seth Davis or TV guys like Seth Greenberg. Those guys communicate with coaches a lot, but they are generalists, knowing a little bit about a lot of teams and their coaches, to provide when it's their turn to talk for a few minutes. They don't get in the weeds like would be ideal for the committee positions we're talking about.
 
More out of conference games later in the year, perhaps a dedicated week in early February, so that relative conference rankings aren't set in stone after December 31.
 
Back
Top