Guide
March Madness bracket explained: how 76 teams get placed
The March Madness bracket explained properly: the 76-team field, the NET and quadrant records, the committee's seed list, and the rules that beat seed order.
By CricketTaken EditorialPublished Guide21 min read
Every year somebody looks at a finished bracket, finds a team two lines above where the metrics said it belonged, and calls it a mistake. It usually is not. The seed list and the bracket are two separate documents produced by two separate processes, and the second one is allowed to overrule the first.
That gap is the whole subject. Once you know why the committee is permitted to move a team off its true seed, and exactly how far, most of the annual outrage stops being interesting and the actually contentious decisions become easy to find.
This is the March Madness bracket explained as a procedure: who votes, what they vote on, what the numbers mean, and which rules take priority when two of them cannot both be satisfied.
The bracket is now 76 teams, and that is a recent change
The field has been 68 since 2011, with four opening games and a round of 64 arriving on the Thursday. That is over.
On 7 May 2026 the NCAA announced that the Division I men's and women's fields will each increase from 68 to 76 starting in 2027. The decision cleared the men's and women's basketball committees, both oversight committees, the Division I Finance Committee, the Division I Board of Directors and the NCAA Board of Governors, which is a longer chain of approvals than most rule changes require and tells you how much money was attached to it.
The arithmetic of the new field is clean. Every conference sends its tournament champion, which is 32 automatic qualifiers. The committee then selects the 44 best teams that did not qualify that way. Twenty-four of those 76 play in an opening round of 12 games, and the 12 winners join the 52 teams already through to make a round of 64. The men's championship is 75 games from first tip to final buzzer.
- 76Teams in the bracket from 2027
- 32Automatic qualifiers, one per conference
- 44At-large places decided by the committee
- 75Games in the men's championship
The NCAA framed the expansion in terms of access. Before the change, 18 per cent of Division I basketball teams reached the postseason, which the association described as the lowest share among major team sports. The expanded field takes that to 21 per cent.
The historical case is that the bracket has grown before, repeatedly, and always because Division I grew first. The men's championship began in 1939 with eight teams. It has expanded by at least eight teams on four previous occasions: to 16 in 1951, from 32 to 40 in 1979, from 40 to 48 in 1980, and from 53 to 64 in 1985. It stayed at 64 for sixteen years, went to 65 in 2001, and to 68 in 2011. Over the same period the number of Division I men's programmes rose from 282 in 1985 to 361. The women's championship started at 32 teams in 1982, reached 64 in 1994, and went to 68 in 2022, while the number of Division I women's programmes rose from 277 to 359.
Each point is the size in the year the NCAA names. The four increases of at least eight teams came in 1951, 1979, 1980 and 1985.
Show the numbers
| Item | Value |
|---|---|
| 1939 | 8 |
| 1951 | 16 |
| 1979 | 40 |
| 1980 | 48 |
| 1985 | 64 |
| 2001 | 65 |
| 2011 | 68 |
| 2027 | 76 |
Read that line and the argument against expansion loses some of its force. A 64-team bracket drawn from 282 programmes and a 64-team bracket drawn from 361 are not the same competition, and the second one is harder to reach.
Twelve people decide, and the recusal rules are the interesting part
The Division I men's basketball committee has 12 members serving five-year terms. One comes from each autonomy conference. Four represent the seven highest-ranked non-autonomy conferences, and four represent non-autonomy conferences ranked from 12th to 31st.
Every one of those people works for a school or a conference with something at stake. The NCAA's answer is not to pretend otherwise but to write the conflicts out in detail.
A member cannot be present during discussion or seeding of a team he or she represents as an athletics director or conference commissioner. Members may offer only general, factual information about teams from their own conference, and cannot vote for a team they represent. A conference office representative may not take part in discussing or voting for a team that intends to join that conference, which closes a loophole that would otherwise open every realignment cycle. Recusal also applies where an immediate family member is a player on the team, on the coaching staff, or a senior athletics administrator at the school.
The detail that shows the most thought is what happens on the way back into the room. A recused member returning to the table gets a neutral update from the NCAA vice president of men's basketball, not from the other committee members. The people who just argued about the team do not get to summarise the argument.
All voting is by secret ballot. If three consecutive rounds tie, the chair breaks it. And once the pool of teams still under discussion falls to 24 or fewer, conflicted members cannot participate at all, which is precisely the point at which a single vote starts to matter.
The NET is a sorting tool, and the committee says so
The NCAA Evaluation Tool arrived for the 2018-19 season and immediately acquired a reputation it does not deserve, which is that it decides the field.
It has two components. The Team Value Index is built from Division I results and accounts for the opponent and the location of the game; the NCAA describes it as an algorithm set up to reward teams that beat good teams. The second component is adjusted net efficiency, which is offensive efficiency minus defensive efficiency, again adjusted for venue and opponent quality. One half measures what you achieved. The other measures how well you played while achieving it.
The NET is not the only number in the room, and the committee publishes the rest.
BPI is a measure of team strength rather than team accomplishment, expressed as how many points better or worse per game a team is than an average Division I side. It blends a preseason rating with results, and the preseason component decays towards nothing as games are played. Each game accounts for points, possessions, opponent strength, game site, travel distance, days of rest and altitude.
KenPom is predictive, built from offensive and defensive efficiency per 100 possessions, and answers how strong a team would be if it played tonight, setting aside injuries and emotion.
KPI values every individual game. The best possible win is worth about +1.0, the worst possible loss about -1.0, and a neutral result sits at zero, adjusted for venue, opponent quality and the share of total points scored. Game values are summed and divided by games played. Because every game carries a number, a schedule can be sorted from best win to worst loss, and the index can be broken out by category.
Torvik shares KenPom's efficiency foundation and adds two twists: a game script that discards data after a game is mathematically decided, and a recency weighting under which games in the previous 40 days count fully, then decay by one per cent a day until they are 80 days old, after which they count 60 per cent.
Strength of record asks how hard it was to compile the record a team actually has, by calculating the probability that a typical 25th-ranked team would have matched or beaten it against the same schedule.
Wins above bubble works out how many games an average bubble team would have been expected to win against your schedule and subtracts that from your actual wins.
Six different attempts to answer the same question, deliberately not reconciled into one number. That is the design. A committee handed a single ranking would defer to it; a committee handed six that disagree has to think.
Around all of it sits a list of resources that has nothing to do with computers: season-long evaluation from watching games, conference monitoring calls, coaches' association rankings, box scores, head-to-head results, results against common opponents, imbalanced conference schedules, overall and non-conference strength of schedule, quality wins and losses, road record, and player and coach availability. That last item is why a team that lost six games without its starting point guard is not judged as though those were six ordinary defeats.
Quadrant records reward the road, and almost nobody schedules for it
The team sheet is the document the committee actually reads, and its organising idea is the quadrant.
Each of a team's games is filed into one of four quadrants according to the opponent's NET rank and where the game was played. Quadrant 1 is a home game against a team ranked 1 to 30, a neutral-court game against 1 to 50, or a road game against 1 to 75. Quadrant 2 runs 31 to 75 at home, 51 to 100 neutral and 76 to 135 away. Quadrant 3 runs 76 to 160, 101 to 200 and 136 to 240. Quadrant 4 is everything below that.
Those three bars are the single most exploitable fact in the selection process. Beating the 31st-best team in the country in your own building is a Quadrant 2 win. Beating the 75th-best team in theirs is a Quadrant 1 win. The gap between those opponents is enormous and the gap in how the result is filed runs the other way.
Athletic directors know this. It is still remarkably hard to act on, because a true road game at a mid-major requires the mid-major to agree to a series, and the guarantee-game economy pushes everything towards neutral floors and home dates that pay.
The committee, to its credit, publishes the warning as well as the tool. Its own guidance points out that a road win over a top-five team sits in the same quadrant as a road win over a team ranked 38th or 75th, that a home loss to a team ranked 80th sits in the same quadrant as a home loss to a team ranked 160th, and that a home win over a team ranked 165th appears in the same column as a road win over a team ranked 353rd. The quadrants sort. They do not evaluate. Anyone reading a team sheet as a set of four win-loss records has thrown away most of the information on it.
Phase I: how 44 at-large teams are actually voted in
The selection process is three phases run in order, and the balloting inside the first two is more machinery than most coverage suggests.
Each committee member receives an alphabetical list of every eligible Division I team and submits an initial ballot in two columns. Column 1 holds up to 24 teams the member believes should be at-large selections. Column 2 holds any number of teams that deserve consideration.
Two thresholds come out of that first ballot. A team receiving all but three of the eligible votes in Column 1 is an automatic at-large selection and leaves the discussion immediately. A team receiving at least four votes in either column goes onto the under-consideration board, as does any team that won or shared its regular-season conference title. Verbal nominations are allowed, so nothing is trapped by the paperwork.
The board is then worked through in rounds. In each round members pick the best teams remaining: eight when more than 20 are under consideration, up to six when 14 to 19 remain, up to four when 13 or fewer remain. The eight teams with the most votes form the next ballot. Members rank those eight, with the best team scored at 1 point. The four teams with the lowest point totals, meaning the four ranked highest, join the at-large field. The other four carry over.
Then each member names the best eight teams still on the board, the four with the most votes join the four carried over, and the cycle repeats until all 44 places are filled.
Three safety valves run throughout. A team can be pulled back out of the field by all but three votes and returned to the board. A team that fails to be selected on two consecutive ranking ballots goes back to the board rather than lingering. The chair may adjust how many teams are eligible in a round or how many are selected per round.
What this produces is not one vote on the bubble but dozens of repeated head-to-head comparisons of small groups. A team does not need to be somebody's 44th-best team in the country. It needs to keep surviving eight-team ballots against whoever else is left, which is a much better test of a genuine consensus.
Phase II: the seed list, and what scrubbing actually does
The committee then produces a seed list ranking all 76 teams from strongest to weakest. This is the true seed, and once final it does not change.
The balloting machinery is the same: groups of eight, six or four depending on how many teams remain, ranking ballots scored with the best team at 1 point, four teams added to the list per round and four carried forward. The same conflict rule applies once the pool falls to 24 or fewer.
Two features of this phase are worth pulling out.
The first is that the committee does not have to build the seed list from the top down. It may seed different sections at different times, which matters because the hardest arguments are usually in the middle, and a group that has spent two days on the bubble has already done most of the thinking about where those teams belong.
The second is scrubbing. Once a team is on the seed list, it can be moved up or down by simple majority. Not a supermajority, not the all-but-three threshold used for selection. A majority. Scrubbing applies only to teams already in the field, so it cannot be used to smuggle anyone in through the back door, and its purpose is to fix the ordering errors that the ballot process creates. Balloting in groups of eight is efficient and slightly lossy: a team that entered the list in round three might be clearly better than one that entered in round two, and scrubbing is how that gets corrected without rerunning everything.
- Automatic qualifiers32
- At-large selections44
The automatic side is decided on the floor by 32 conference tournaments. The at-large side is decided in a room.
Show the numbers
| Item | Value |
|---|---|
| Automatic qualifiers | 32 |
| At-large selections | 44 |
Phase III: the bracketing principles that beat seed order
Now the seed list meets the bracket, and this is where placement stops matching ranking.
The bracket is 16 seed lines across four regions. Teams are placed as predetermined groups rather than individually: seeds 1, 16, 8 and 9 travel together, as do 4, 13, 5 and 12, then 2, 15, 7 and 10, then 3, 14, 6 and 11. Each group is assigned to a first and second round site, with two pods at each site that may feed into different regionals. When people complain that a team was "sent" somewhere, the thing that moved was usually the group.
Conference separation comes first and is not negotiable in the ordinary run of things. The first four teams from any conference placed on the top four seed lines must go to four different regions, unless that conference has five or more teams on those lines, at which point the rule cannot be satisfied and stops applying. Beyond that, how early two teams from the same conference may meet depends on how often they have already played. Three or more meetings across the regular season and the conference tournament, and they cannot meet before the regional final. Two meetings, and not before the regional semi-final. One meeting or none, and they may meet as early as the second round.
Those restrictions relax if a conference has nine or more teams in the tournament, which is now a real possibility rather than a hypothetical, and any principle may be relaxed if two or more of a conference's teams are among the last 12 at-large teams in the Opening Round.
Bracket integrity is the second constraint, and it is the one that produces the movement people notice. The committee may move a team up or down one seed line from its true seed, and in rare cases two, to make all the principles fit. That is the entire licence. A team sitting 27th on the seed list is a true 7 seed, can appear on the 6 or 8 line, and will only appear on the 5 or 9 line in an unusual bracket.
There is also a balance check. After the top four lines are placed, the committee compares the total seed value in each region and tries to keep the imbalance small, preferably with no more than six points between the highest and lowest total. Four regions each holding roughly the same weight of quality is what stops a "region of death" being built by accident.
Geography is third. Teams should stay as close to home as mileage allows. If two teams from the same area want the same bracket slot, the higher seed stays local. A team cannot be sent to a venue where it has already played more than three games that season, excluding exhibitions and the conference tournament. Host schools cannot play at the site they are hosting, though they may play on the same days, and a team may play at a site where its conference is the host. The committee also tries not to move the same programmes out of their natural region year after year, which is a fairness principle operating across seasons rather than within one bracket.
The placement order itself is fixed.
- Phase I, the initial ballotEach of the 12 members names up to 24 teams as at-large selections and any number more as worth considering. All but three votes in the first column and a team is in immediately.
- Phase I, the under-consideration boardEverything else that got four votes, plus every regular-season conference champion. Members repeatedly pick the best eight remaining, rank them, and the top four join the field.
- Phase II, the seed listThe same balloting machinery orders all 76 teams from strongest to weakest. Scrubbing then moves individual teams up or down by simple majority to correct the ordering.
- Phase III, conference separationBefore anyone is placed, the rivalry constraints are applied. Three previous meetings means no earlier than the regional final. Two means no earlier than the regional semi-final.
- Phase III, the top four linesThe four No. 1 seeds are assigned to regions, and the overall No. 1 chooses its region and site. The 2 seeds follow in true-seed order, then the 3s and 4s, then a balance check across the regions.
- Phase III, everyone elseBracket groups are assigned to first and second round sites in true-seed order, then seeds 5 to 16 are placed. A team may shift one line, rarely two, to make the principles fit.
Three phases, run strictly in order. Nothing in Phase III can add a team to the field, and nothing in Phase III changes the seed list. It only decides where the seed list gets put.
The four No. 1 seeds go in first, which fixes the Final Four semi-final pairings, and the overall No. 1 seed chooses its preferred region and its first and second round site. The 2 seeds follow in true-seed order, with enough flexibility to avoid putting the strongest 5 seed in the overall No. 1's region, though conference separation still cannot be compromised to achieve it. The 3s and 4s go in by the seed list. Then the balance check. Then bracket groups are assigned to sites in true-seed order, and seeds 5 through 16 are placed.
One scheduling rule sits underneath all of it and explains a placement that looks arbitrary every year. First and second round sites split into Thursday-Saturday and Friday-Sunday pairs. If all four No. 1 seeds would otherwise land at Thursday sites, the fourth of them is moved to the nearest Friday site, because Opening Round winners have to play on the Friday and somebody has to be there to play them.
Rematch avoidance is a preference rather than a rule, and the hierarchy is explicit. The committee tries to avoid non-conference rematches in the Opening Round and the first round, and tries again in the second round, and avoids rematches from the previous year's tournament where it can. But teams will not be moved off their true seed line purely to prevent a rematch. Seeding beats novelty.
What the Opening Round changes, and who it is really for
The First Four is gone. From 2027 the opening games are branded the March Madness Opening Round, and there are 12 of them.
The 24 teams involved are the 12 lowest-seeded automatic qualifiers and the 12 lowest-seeded at-large teams. Six games pair the automatic qualifiers against each other, six pair the at-large teams against each other, and the pairings are made by adjacency on the overall seed list, so the two teams at the very bottom of the list meet, then the next two, and so on. Exceptions are allowed to avoid a regular-season rematch or for travel, and the existing conference-separation principles are applied to keep conference rivals apart in the Opening Round and the round of 64 wherever possible.
The men's games run on the Tuesday and Wednesday after Selection Sunday, which in 2027 means 16 and 17 March. Rather than two games a night in Dayton, there will be three each night in Dayton and three each night at a second site; the committee's own published procedures name Wichita, Kansas alongside Dayton. Opening Round teams get priority placement to whichever first-round site is closest to where they played, with other teams on the same line placed afterwards to tidy up the travel. The round of 64 stays on Thursday and Friday, the round of 32 on Saturday and Sunday, and the regional and Final Four dates do not move. On the women's side the Opening Round is played on the Wednesday and Thursday on the campuses of 12 of the top 16 seeds.
The sporting argument for it was made by the men's committee chair, Keith Gill, and it rests on a specific claim: across the last two years of the men's and women's tournaments, teams seeded 15 or 16 went winless in 32 first-round games. Under the new structure, nearly half of the 28 men's and women's teams on the 15 and 16 lines will win at least one tournament game, because they will have played somebody their own size first.
Whether that is a better tournament depends on what you think the 16 line is for. It is unarguably a different one.
There is a second effect that gets less attention and probably matters more to the people affected. Six conference automatic qualifiers, most likely from outside the autonomy conferences, will now play at least two March Madness games, which guarantees their conference at least one additional basketball fund unit. Under the old format only two teams got that. For a one-bid league, a unit is a meaningful share of a budget.
The bracket is a revenue document as well as a sporting one
Expansion was not sold to the membership on competitive grounds alone.
The NCAA said it will be able to award more than $131 million in new revenue distributions to member schools over the remaining six years of its broadcast agreements, with the value of the rights deal rising by an average of $50 million a year across those six years. The basketball fund grows by 16 additional units on the men's side and 16 on the women's. Part of the deal opens previously restricted product categories to the corporate partner programme, including beer, wine, spirits and hard seltzer, and expands in-game advertising in both linear and streaming coverage.
That money now lands in a different world from the one the old bracket was built for. Schools are permitted to share revenue directly with athletes, so an additional unit is not only a line in an athletic department budget; it can be scholarships, or a distribution, or both. The economics running underneath college basketball have changed faster than the competition format, and the format has been dragged after them. Anyone following the same forces through the rest of college sport will recognise the pattern from the way name, image and likeness deals reshaped recruiting and from the movement of players between programmes each spring.
All 75 games of the men's championship air across TBS, CBS, TNT and truTV. Every game of the women's championship airs on ESPN networks.
The complaints that miss, and the one that lands
Three criticisms come round every March. Two of them are answerable from the NCAA's own documents.
The committee just reads the NET. It does not, and the published resource list runs to more than a dozen inputs of which the NET is one. The committee's own guidance on quadrants is a warning against reading them as a ranking. If anything the risk runs the other way: a committee with seven conflicting metrics can find support for almost any conclusion it has already reached, which is a problem of a different shape.
Small conferences never get the benefit of the doubt. The NCAA published the counter-evidence with the expansion announcement. Since the men's field last changed in 2011, the publicly revealed "First Four out" across those 15 tournaments included 60 teams. Thirty-five came from the five autonomy conferences, including the old Pac-12. The other 25, which is 42 per cent, came from 13 different conferences: four each from the Mountain West and the American, three each from the Big East and the Atlantic 10, two each from the Horizon League and the Missouri Valley, and one each from the Colonial, Conference USA, the Ivy League, the MAAC, the Southern Conference, the Summit League and the West Coast Conference. On the women's side, in each of the five seasons since the 2022 expansion, at least one team from outside the autonomy conferences has been among the First Four out.
The criticism that does land is about dilution, and it is not really about the 76th team. It is about the bubble. Every expansion moves the argument down the seed list, and the argument is the product. A tournament that admits everyone with a case removes the thing people spend February arguing about, and the marginal at-large team added by expansion is, definitionally, the team the committee was least sure about. The counter-argument, that the field has grown before and the sport grew faster, is a good one. It is not a complete one.
How to read a bracket the way it was built
Four checks, in this order, will tell you almost everything about a finished bracket.
Compare placement line to true seed. Every team on the 7 line is not a true 7. The committee releases the full seed list, and the interesting cases are the teams whose placement and true seed disagree. That gap is not an opinion about the team. It is the fingerprint of a constraint.
Look for the conference collisions. Find the conferences with four or more teams in the field, count how often each pair met during the season, and you can reconstruct most of the movement in the top half of the bracket. The three-meetings rule is the most restrictive constraint in the document and it does more work than the seed list in a year with a dominant league.
Check the venue history. A team that played four games in a building cannot be sent back to it. That single rule redirects at least one bracket group most years, and it is invisible unless you go looking for it.
Add up the regions. Sum the seed numbers in each region. If one region is materially heavier than the others, either the committee accepted an imbalance to satisfy a harder constraint, or the seed list itself was lopsided at the top. Both are worth knowing, and neither is a mistake.
None of this requires an opinion about who deserved the last at-large place. It requires the seed list, the schedule and the published principles, all three of which the NCAA hands out. If you want to see how differently another American sport answers the same problem of choosing a postseason field, the committee-selected football bracket is a useful contrast, and so is the professional game's approach of settling the final places on the floor rather than in a room. The rest of the archive lives in the basketball section, and the qualifying route to the international game is set out in the Olympic pathway.
Common questions
How many teams are in the March Madness bracket?
From 2027 the Division I men's and women's fields are 76 teams each, up from 68. The NCAA announced the expansion on 7 May 2026. Thirty-two places go to conference tournament winners as automatic qualifiers and the remaining 44 are at-large selections made by the committee.
What is the NET ranking and how does the committee use it?
The NCAA Evaluation Tool has two components: a Team Value Index built from game results, opponent and venue, which is designed to reward beating good teams, and adjusted net efficiency, which is offensive efficiency minus defensive efficiency with the venue and opponent quality factored in. The committee does not select or seed by NET order. It uses NET to sort a team's results into quadrants on the team sheet, and it says plainly that not all wins inside the same quadrant are treated equally.
What are quadrant records in college basketball?
A quadrant record splits a team's wins and losses by how good the opponent was and where the game was played. A Quadrant 1 result means an opponent ranked 1 to 30 in the NET at home, 1 to 50 on a neutral floor, or 1 to 75 on the road, with the bands widening through Quadrants 2, 3 and 4. The point of the venue adjustment is that a road win is harder than a home win over the same opponent, so the same opponent sits in a different quadrant depending on where you played them.
Why do teams get moved off their true seed line?
Because the seed list is only the input to the bracket. The committee may move a team up or down one seed line, and in rare cases two, to satisfy the bracketing principles: keeping conference rivals apart, keeping teams near home, avoiding a venue where a team has already played more than three times, and keeping the four regions roughly equal in total seed value. A team that finishes 27th on the seed list will not always appear on the 7 line.
What happened to the First Four?
It has been replaced. From 2027 the opening games are branded the March Madness Opening Round and there are 12 of them rather than four. Six pair the 12 lowest-seeded automatic qualifiers against each other and six pair the 12 lowest-seeded at-large teams against each other, with the men's games split across Dayton and a second site.
Filed under Basketball·college basketball · march madness · ncaa · seeding · format