All scores →
Loading…
Loading…
Loading…

How the NCAA tournament bracket is built: seeding rules and the S-curve

By 4Basketball Staff · · 14 min read

How the NCAA selection committee turns 68 teams into a bracket: the numbered seed list, the S-curve, the rules for separating teams, and why a seed is a label rather than a rating.

Every March, millions of people open a bracket and see a tidy tree: four regions, sixteen seeds in each, one line of matchups feeding the next until a single team is left. It looks as if it fell out of the sky already assembled. It did not. Somebody had to decide that a particular team would be a 3 seed instead of a 4, that it would play in one region rather than another, and that its likely path would run through a particular set of opponents. A bracket is a set of hundreds of small decisions made in a few days by a room of people, and those decisions shape who is likely to reach the Final Four long before a ball is tipped.

The method behind those decisions is more orderly than most fans assume. The committee does not simply sort teams into groups of four and draw regions out of a hat. It first ranks every team in the field from the best to the worst, then uses a pattern called the S-curve to spread the strength evenly across the four regions, and then adjusts for a short list of rules about geography, conferences and the schedule. Once you understand those three steps you can look at any bracket and see where the committee was following its own system, where it bent it, and where a team was helped or hurt by the shape of the draw.

This article walks through that process from the top: how the seed list works, what the S-curve does, how the matchups inside a region are fixed, how the First Four fits in, and why the whole machine is better at fairness than at prediction. It is about the men's tournament, which has a field of 68. The women's tournament uses a similar logic with its own hosting arrangements, which we cover in a separate piece on the format, regions and growth of the women's event. For how the 68 are chosen in the first place, see how the NCAA tournament field is chosen.

The raw material: 68 teams in two groups

Before any seeding happens, the field has to exist. Thirty-two of the places go to automatic qualifiers, one for each Division I conference that sends its champion, and the other thirty-six are at-large bids handed out by the selection committee. The road to those automatic places runs through the conference tournaments, a subject covered in our look at automatic bids and conference tournaments. The numbers have changed over the decades as the field grew, a story told in tournament expansion from 8 teams to 68, but the principle has been constant since the late 1970s: the committee's job is to place teams in a bracket by ranking them.

Seeding itself is a fairly recent idea. The earliest tournaments, starting in 1939, were small enough that teams were simply placed in regions, mostly by geography. The committee began assigning numerical seeds in 1979, and the modern 64-team format arrived in 1985. The field has since added a few extra spots, but the core bracket of 64 teams, sixteen to a region, has been the standard picture ever since. When people say that no 16 seed beat a 1 seed until 2018, they are describing something that only exists because of the seeding system, and only in the era of the 64-team bracket.

An important consequence of the two groups is that the field is not made of 68 equals chosen for the same reason. Some teams are in because they won a small conference's tournament and were never really in doubt for a seed line. Others were chosen on merit from the biggest conferences. When the committee builds the seed list, both kinds of teams go into a single ranking, and the champion of a small league can land anywhere from a very high seed down to the bottom of the field depending on how its record and ratings compare with everyone else's.

The seed list: ranking every team, 1 to 68

The central object in the process is the seed list, a single numbered ranking of the entire field. The committee does not think "who are the 3 seeds" in isolation. It thinks "who is the fifteenth-best team in the field, and who is the sixteenth", and the seed lines fall out of the ranking in blocks of four.

The first four teams on the list are the 1 seeds, and the team ranked first is the overall No. 1 seed, a phrase that matters because the system is designed to give the best team the easiest road. The next four, ranked fifth through eighth, are the 2 seeds. The next four are the 3 seeds, and so on down the lines. Each of the sixteen seed lines contains four teams, one for each region. The bottom lines are slightly different because of the First Four, which we come back to below.

How does the committee rank teams? It reviews a large package of information for each school, usually presented on a one-page summary called a team sheet. The sheet lists the record, the strength of schedule, results against top opponents, results away from home and at neutral sites, and a bundle of ratings. The NCAA's own ranking tool, known as NET, anchors the sheet, and results are sorted into four quadrants by the quality of the opponent and the location of the game. We explain that sorting in quadrant records and the selection committee's NET, and the family of public ratings that sit alongside it in college efficiency metrics: KenPom, NET and the ratings used to select the field. The committee also watches games and weighs injuries, so a team that was playing without a key player for part of the season may be judged on what it looks like at full strength.

The process of ranking involves repeated rounds of voting rather than a single list. Members name teams they believe belong in the field or on a given line, the group discusses, and the ballots are repeated until the ranking settles. Members step out of the room when their own school or conference is discussed. The goal is not to find a perfect order, which does not exist, but to produce a defensible one, because the seed list is what everything else rests on.

The S-curve: spreading strength across four regions

Once the list exists, the committee has to put the teams into regions. If it placed the four 1 seeds in order, one per region, and then did the same for the 2 seeds going in the same order of regions, the region that received the best 1 seed would also receive the best 2 seed, the best 3 seed, and the best 4 seed. That region would be brutal, and the opposite one would be easy. The S-curve prevents that by reversing the order on alternate lines, so the pattern snakes back and forth, like the letter S.

Here is how it works. The No. 1 overall team goes to the first region. The second-best 1 seed goes to the second region, the third to the third, and the fourth to the fourth. On the 2 line the order flips: the best 2 seed, ranked fifth overall, is placed in the region of the weakest 1 seed, ranked fourth overall, and the weakest 2 seed, ranked eighth, goes in the region with the best 1 seed. The 3 line flips again and runs in the original direction. In other words, each region gets one team from each line, and the regions are balanced by pairing strength with weakness.

Overall rank Seed line Region it joins
1 1 seed Region A
2 1 seed Region B
3 1 seed Region C
4 1 seed Region D
5 2 seed Region D
6 2 seed Region C
7 2 seed Region B
8 2 seed Region A

The table shows only the top two lines, and the real bracket continues the same back-and-forth pattern all the way down to the 16 seeds. In its purest form, the S-curve gives the top overall team the weakest 2 seed, and balances that by giving it the strongest 3 seed, then the weakest 4 seed, and so on down the lines. A simple way to see the logic is to add up the overall rankings of the four teams a region holds from the top four lines. Region A holds the teams ranked 1, 8, 9 and 16, and Region D holds 4, 5, 12 and 13, and both add up to 34. Regions B and C come out the same. That equality is the whole point.

The S-curve is also why the committee thinks of the bracket as a system of incentives. Finishing first in the ranking is worth something concrete, the best draw available, and the difference between the first and fourth 1 seed is real even though the seed number is the same. It is also why a team that is ranked fifth feels lucky when it is a 2 seed and ranked eighth feels less so, even though both are called the same thing.

Inside a region: why 1 plays 16 and 8 plays 9

After the S-curve has put sixteen teams in each region, the matchups inside the region are fixed by a rule that never changes. In the first round, the seed pairings add up to 17: 1 plays 16, 2 plays 15, 3 plays 14, 4 plays 13, 5 plays 12, 6 plays 11, 7 plays 10 and 8 plays 9. The positions in the bracket are arranged so that, if the favorites keep winning, the matchups in the later rounds are also predictable: the 1 seed would face the 8 or 9 in the second round, then the 4 or 5 in the Sweet 16, and then the 2 or 3 in the regional final.

Top to bottom in the printed bracket, the order of the first-round games is 1-16, 8-9, 5-12, 4-13, 6-11, 3-14, 7-10 and 2-15. If you look at the structure, a 1 seed's quarter of the region holds the 8 and 9 seeds in its second-round game, and the 4, 5, 12 and 13 seeds in the group it meets in the Sweet 16. The 2 seed's half mirrors it with the 7 and 10, the 6 and 11, and the 3 and 14.

This fixed shape has a few consequences that are easy to miss.

  • The 8-9 game is the real coin flip of the first round. Neither team is a favorite in any meaningful sense, and the winner earns a second-round game against a 1 seed. A team that is ranked a few places above the 9 line gets nothing special for the effort except that opponent.
  • The 5-12 and 6-11 games hold more potential for surprises than the 4-13 game, partly because the gap in quality between the lines is smaller. We take that apart in the article on the 12-5 upset.
  • A 1 seed that gets through its first two games faces, on paper, a very different Sweet 16 opponent from a 2 seed, depending on how the rest of the region fills out.

The bracket never changes the structure. What the committee decides is who goes where. The matchup pattern itself is as fixed as a chess opening.

The extra rules: geography, conferences and the calendar

The S-curve is the main system, but the committee is bound by principles that sometimes force it to move a team a seed line or a region from where the pure S-curve would put it. The committee has described a few of these over the years, and while the details are revised from time to time, the underlying ideas are stable.

Keeping conference rivals apart

The committee tries to avoid early rematches between teams that play each other during the regular season. The most visible form of this is a rule that teams from the same conference should not meet until late in the bracket if they have already met multiple times that season. The details, including how many regular-season meetings trigger the protection and how late in the tournament teams may meet, have been adjusted by the committee over the years. The practical result is that you rarely see two teams from the same conference matched in the first two rounds, and the committee will move a team to a different region, even if that means a small departure from the S-curve, to make it work.

Geography and the early rounds

The first two rounds are played at eight sites around the country, and the committee tries to place the strongest seeds close to home when it can. The No. 1 overall seed typically gets the most favorable geography, though it is not guaranteed. This matters because crowd support can matter in a single-elimination game. The tournament is not a home-court event, but a team with a short drive and a lot of fans in the building does have an edge that a team flying across the country might not.

The committee also handles individual school policies, such as the long-standing practice of BYU not playing on Sundays, by placing the school at a site where it is scheduled to play on days that fit. These accommodations are small, but they show that the bracket is built by people balancing competing demands rather than by an algorithm that spits out the result.

Bracket integrity

The committee gives priority to what it calls the integrity of the bracket: teams on the same seed line should not be moved far from where the seed list placed them, and the adjustments for rivals and geography are kept to a minimum. Fans sometimes see a team as an unlucky 4 seed because of a hard draw, but most of the time the draw was the S-curve's natural output.

The First Four and the bottom of the bracket

Since 2011 the field has had 68 teams, and the four extra teams play the First Four in Dayton, Ohio, to determine who fills the last spots in the 64-team bracket. Two of those games involve the four lowest-ranked automatic qualifiers, which compete for two places on the 16 line. The other two involve the last four at-large teams to make the field, usually the 11 or 12 line, and decide two more places. We explain why the games exist and how they have worked out in the First Four: why there are play-in games.

For bracket building, the First Four has an important consequence. The seed list runs to 68 rather than 64, and the committee must decide which teams will be in the opening-round games before it assigns the rest. The bottom four automatic qualifiers on the list are the ones who play for the 16 line, so a small-conference champion with a weak record can end up in Dayton on Tuesday or Wednesday rather than in a first-round game on Thursday. The last four at-large teams are the ones on the very edge of the field, the bubble teams who made it in by the thinnest margin. Our piece on the bubble explains how that edge is judged.

The winners of the First Four are slotted into the bracket in the spot reserved for them, and they play their first-round game against the team the pairing says they should. Since the First Four teams have already played, they sometimes show up in the second weekend with the extra game behind them, and a few have advanced a long way.

A seed is a label, not a rating

Anyone who fills out a bracket learns quickly that a seed number tells you less than it seems to. A seed describes where the committee ranked a team in a list of 68 and where it put the team on the bracket. It is a label, and a label compresses a great deal of information. Two 5 seeds can be very different teams, one that was ranked 17th and one that was ranked 20th, and the difference could be the difference between a team with a fair chance to win and one that is a coin flip against a good 12.

The more interesting problem is in how the seed list relates to a bracket's difficulty. The S-curve balances the total strength of regions in the abstract, but the path a team faces depends on specific opponents and on the shape of the bracket around it. A 1 seed in a region with a very strong 2 seed will likely face a tougher road than a 1 seed in a region where the 2 seed is the weakest on the line. A 4 seed that avoids a dangerous 5 seed is better off than the number suggests. And because the field includes both conference champions and at-large teams, the actual spread of quality on a line can be wide: a 12 seed from a strong small league may rate higher than a 10 seed from a larger conference.

This is where a thoughtful fan can see something that casual viewers miss. The committee builds a bracket to be fair, not to make the tournament easy to predict. Fairness means that the best team is rewarded, that no region is stacked, and that teams are not treated unfairly because of geography. None of that is the same as accuracy. The tournament is a single-elimination format with games decided by a few possessions, and the best bracket in the world cannot remove the randomness that makes the event what it is. We look at why that randomness is built into the sport in why upsets happen.

Reading a bracket like a committee member

Once you know the system, you can read a bracket the way the committee does. These are the questions worth asking when the field is released.

  1. Who are the four teams on the 1 line, and in what order did the committee rank them? The region of each tells you roughly how the S-curve was applied.
  2. Is any region obviously strong or weak? If so, look for the reason: a late conference-tournament result, a rival-separation rule, or a real judgment call by the committee.
  3. Which seed lines were affected by the First Four? The bottom of the field holds the information about how the committee viewed the bubble.
  4. Where are the favorites playing geographically? Short trips usually mean a high ranking.
  5. Which teams look underseeded or overseeded compared with public ratings? Those are the teams most likely to attract talk of upsets, and they are the subject of any bracket strategy, which we cover in bracket-picking strategy.

You can also check the bracket against the record of how top seeds have fared. The long-term pattern is explored in the history of the one-seed, and the way the event has grown into a full cultural phenomenon is discussed in the history of the NCAA basketball tournament. For a view of the whole field on the site, the March Madness page and the men's bracket show the structure in practice.

What to take away

A tournament bracket is built in three layers. First, a numbered ranking of all the teams, produced by repeated votes and a large pile of data. Second, an S-curve that spreads those teams across four regions so that the strongest do not cluster and the best team receives the easiest path. Third, a small set of rules that adjust for geography, rivals and the calendar. The matchups inside a region are fixed, with seeds always adding to 17 in the first round, so the committee's real power is in deciding who occupies which slot.

For a fan, the lesson is that a seed is a ranking position with a bracket attached, not a prediction. It tells you how the committee ordered the teams and it tells you the path, but it does not tell you who will win. The best use of the seed list is to see what the committee believed, to compare that belief with other ratings, and to notice where a team's route is easier or harder than its number suggests. That habit will make the next bracket more interesting to read, and it is the foundation for every good pick you make in it.

Search more on this

Keep reading