How does the NCAA basketball Selection Committee choose the March Madness bracket? Here's what you need to know to understand the selection process.
Right now as you’re reading this, 10 men and women—a mix of athletic directors and conference commissioners—are hard at work building college basketball’s defining event. The decisions that group makes on Sunday, March 11, will shape the next three weeks and ultimately influence the history books.
The NCAA men’s basketball Selection Committee has a daunting job. The least we could do is try to understand the process.
How does this whole thing work?
Yes, that’s the perfect place to start. The committee cannot seed and bracket the teams without first selecting which ones will participate. In any given year, it’s relatively easy for the committee to include, say, between 62 and 65 of the 68 teams that make the field. First, the 32 conference tournament winners get automatic bids. Every tournament champion from the mighty ACC to the not-so-mighty MEAC gets to put on the dancing shoes, without question.
From there, the committee must select what it deems to be the 36 best at-large candidates. Most of these are slam-dunks. For example, neither Purdue nor Michigan State won the Big Ten tournament, but both are just as assured of a bid as Michigan, which did cut down the nets in New York last week. Weeks before Selection Sunday, the committee likely already knows at least one-third of the teams that will get at-large bids, if not more. By time Championship Week arrives, chances are there are just a handful of remaining spots up for grabs.
Those remaining spots are where things get tricky. Bubble teams play a high-stakes game of musical chairs during Championship Week, with X number of teams fighting for Y number of at-large bids. Yes, the Y number is always fewer than the X number. The committee weighs multiple criteria, debating the teams on the bubble before it settles on what it believes to be the final entries to round out the 36 at-large bids and, in turn, the NCAA tournament field. After that, it gets to seeding and bracketing.
What are the criteria?
The committee will pore over any number of stats and metrics to try to make their team-by-team comparisons as close to apples-to-apples as possible. Some of those will be simple, such as overall record. Some will be a bit more complex, such as computer rankings generated by respected observers of the game. This includes kenpom.com, Sagarin and BPI. Some—such as conference record— will be part of the discussion but carry little weight. Some will act more as tiebreakers, such as non-conference strength of schedule. Again, this is not a job the committee takes lightly. For as much flack as it gets, the committee leaves no stone unturned.
The most important criteria are a mix of old standbys and new constructs, in an effort to improve the process while not completely ditching what the committee believes works. They are as follows:
RPI: Like it or not, the Ratings Percentage Index remains a big part of the selection process. The RPI has been around a long time, and while everyone admits it isn’t close to perfect, it’s still one of the most important parts of a team’s resumé.
Strength of schedule: There are 351 Division I basketball teams, all of which play varying non-conference schedules before separating into their 32 conferences for league play. No one schedule is anything close to alike. SOS helps the committee compare schedules as fairly as possible.
Quadrant 1 and 2 record: This is a new one. It deserves its own section—see more below.
Losses outside the top-100 RPI: Teams separate themselves from the pack, in a positive way, by racking up impressive wins. They do it in a negative way by losing to bad teams, which the committee defines as those with an RPI of 100 or worse. Want to be a high seed? Don’t lose to sub-100 teams. Want to be on the right side of the bubble? Don’t lose too many games to sub-100 teams. It’s pretty simple.
What’s up with these new Quadrants?
The committee needs a way to group wins and losses in a way that helps it compare like teams against one another. Previously, it did so by separating win-loss records against the RPI top 25, top 50, top 100 and top 150, when necessary. However, that system did not account for game location. Is a home win over the No. 25 team in RPI better than a road win over the No. 30 team? Previously, the answer would have been yes, even though anyone who has watched any amount of college basketball knows the road victory is almost certainly the better one.
The Quadrant system aims to fix that. Every game a team plays will fall into one of four quadrants, based on opponent RPI and game location. They are as follows.
Quadrant 1: Home games vs. top 30, neutral games vs. top 50, road games vs. top 75
Quadrant 2: Home games vs. 31-75, neutral games vs. 51-100, road games vs. 76-135
Quadrant 3: Home games vs. 76-160, neutral games vs. 101-200, road games vs. 136-240
Quadrant 4: Home games vs. 161+, neutral games vs. 201+, road games vs. 241+
No matter if you’re Virginia with a guaranteed No. 1 seed or the last team in the field, you shouldn’t be losing many Quadrant 3 or 4 games. The committee will be putting a ton of weight on Quadrant 1, or Q1, and Q2 records. You will hear this discussed throughout Selection Sunday. It could very well be the difference between making the field, and settling for the NIT.
Are there any other bracketing principles we should know about?
Ah, yes, we’ve come to my favorite part. Of course there are!
Once the committee selects the 68 teams, it gets to the seeding and bracketing process. This starts by seeding all the teams, one through 68. However, it isn’t as simple as slotting the teams into the bracket based on those seeds and calling it a day. In order to account for travel, spread out conferences and avoid rematches, the committee follows certain bracketing principles. We’ll go over the most important ones here.
Geography. Are you a No. 1 or No. 2 seed? Congratulations, the committee is going to do whatever it can to make sure you stay as close to home as possible through the Elite Eight. Villanova could very well end up as the No. 1 or 2 overall seed, and Duke could be no worse than the No. 5 or 6, but they might share the East Region, even though a pure seeding process would pit Villanova against the No. 7 or 8 overall seed. That’s because the East Region, hosted in Boston this year, is one of the closest of the four regional spots to Duke’s campus in Durham. That same geographic preference trickles down to teams on the 4- and 5-lines in the first and second rounds.
“But wait,” you say. “The South Region is in Atlanta this year. That’s much closer to Duke’s campus than Boston. Why wouldn’t the Blue Devils end up there?” That brings us to our second crucial bracketing principle. Each of the first four teams from one conference must be in different regions if they are on the first four lines. Duke’s first geographic preference may be Atlanta, but so is Virginia’s, and they cannot share a region. As a No. 1 seed, the Cavaliers get dibs. As such, Duke cannot be in the South Region. That pushes the Blue Devils to their second geographic preference and, depending on what happens this weekend, North Carolina to its third.
You’ll feel that spirit across the bracket. Teams from the same conference cannot meet before the Sweet 16 if they played twice during the regular season, including the conference tournament. They cannot meet before the Elite Eight if they played one another three times in the regular season. They can meet in the second round if they only played once, but the committee will do what it can to avoid such matchups. The committee will also avoid non-conference rematches in the first round. They are allowed if bracketing principles force it, but they are incredibly rare.
Finally, the committee may move teams up or down one seed line—or two seed lines in extreme cases, according to the NCAA’s principles and procedures—to satisfy all bracketing principles.
You still with me? Good, because that’s it. You’re now an expert on the NCAA tournament selection, seeding and bracketing process. As for actually filling out the bracket and whittling the 68 teams down to one, that’s a completely different story.