How Tournament Brackets Work: Seeding, Byes, and Elimination Formats

A tournament bracket maps who plays, who advances, and how entrants leave championship contention. In single elimination, a completed N-player title bracket needs N − 1 matches. Byes fill a bracket's empty opening positions without being played matches; seeding determines who occupies the paths. Double elimination also needs an explicit grand-final rule, because a reset changes both the title procedure and the match count.

Tournament match counts at a glance

These totals assume a completed event and one loss per played two-entrant contest. Withdrawals, unplayed forfeits, extra placement matches, and other finals arrangements require separate accounting. A best-of-three series is one bracket match even when it contains three games.

For a practical single-elimination setup, follow the six-player studio workflow. The sections below explain the counts, placement rules, and the limits of each format.

Start with a real organizing problem

Six people register for a community gaming night. There are two stations and a limited evening to find a champion. An eight-slot bracket seems close enough, but two spaces are empty. Who gets to skip the opening round? Should last month's winner receive an easier route? And will someone who loses immediately still have a reason to stay?

Those questions belong together. The bracket's shape determines how many matches you need, while its placement rules determine who faces whom. The schedule decides whether that structure is comfortable to play through. A tidy diagram can conceal an awkward competition if those choices do not fit.

We will follow an illustrative six-player event from registration to its final, using small calculations you can check yourself. The aim is to build a tournament that delivers the experience its rules promise.

Start with the event you can actually run

Before choosing a bracket, decide what participants should get from the event. A quick championship, an afternoon of games for everyone, and a reliable ranking of the whole field require different amounts of play.

A single-elimination event can identify its champion with very few matches. It gives much less information about everyone else: losing to the champion in the opening round does not prove you were weaker than a semifinalist from another part of the tree.

For a social event, guaranteed participation may matter more than that fast finish. For a competitive qualifier, organizers may value credible rankings and protection against one bad result. Neither goal removes the need to count stations, officials, setup time, and recovery breaks.

Use the following comparison to choose a direction. Each “match” here is one contest between entrants; it may contain several games in a best-of series. Losing one game within a series is different from losing the bracket match.

Small fields make round robin manageable; large fields make its workload more demanding. Double elimination can suit an event that has room for recovery matches. Swiss or groups can provide more participation before a knockout finish. The useful choice depends on the match budget and what the entrants were invited to expect.

What one loss and a second chance change

In single elimination, a match winner advances and the loser leaves championship contention. The final winner becomes champion. A separate consolation activity can keep eliminated players involved, but it adds to the event rather than changing the championship bracket's rules.

In double elimination, everyone starts in the winners' bracket, also called the upper bracket. A first loss sends an entrant to the losers' or lower bracket. Winning there keeps the entrant alive; losing there supplies a second loss and eliminates them.

The lower bracket needs its own planned routes. When an upper-bracket match finishes, its loser must have a specified destination. Some later lower-bracket matches join a survivor with someone newly arriving from the upper bracket. Routing can reduce immediate rematches, but avoidance depends on the layout; it is not an automatic property of having two brackets.

Why the grand final sometimes happens twice

At the end of a traditional double-elimination event, the upper finalist has no losses and the lower finalist has one. If the upper finalist wins their meeting, the other entrant receives a second loss and the event ends.

If the lower finalist wins, both now have one loss. Neither has reached the two-loss elimination threshold. They play another bracket contest, commonly called the grand finals reset, to decide the champion. Evo's official rules describe this procedure in terms of sets: the lower-bracket finalist must win the first grand finals set to force a deciding set.

A reset restarts the final contest. It does not replay the tournament or erase earlier results. Also distinguish a new set from a deciding game already required within the original set.

Some events instead announce a single grand final, perhaps with a longer series or an advantage for the upper finalist. That can make scheduling more predictable. It also means the undefeated finalist can lose the title on their first bracket loss. Publish that exception clearly rather than assuming every double-elimination final works alike.

A second chance provides recovery from a loss, but paths remain unequal. An early loser may play more matches than an undefeated finalist and have less rest before their meeting. Reserve recovery time and a reset slot where applicable. Counting matches alone will not produce a considerate timetable.

Count matches before drawing boxes

The quickest way to count single-elimination matches is to count eliminations. Six entrants need five eliminations to leave one champion. Each played match removes one entrant, so the event requires five matches.

For any field of N entrants, the same argument gives N − 1 championship matches. This assumes everyone completes the event, every contest produces one winner and eliminates one entrant, and there are no extra placement fixtures. A bye is not a played match. Withdrawals and unplayed forfeits can reduce the number actually played while still filling advancement positions in the record.

Rounds measure depth; matches measure work

With eight entrants, the opening round has four matches, the semifinals have two, and the final has one: seven matches over three rounds. Sixteen entrants need eight, four, two, and one: fifteen over four rounds.

Each round halves the surviving field. That is why powers of two—2, 4, 8, 16, 32, 64, and 128—fit a fully populated bracket. Drawn as a tree, entrant positions are leaves, matches join pairs of branches, and the final sits at the root. A balanced power-of-two field has log₂(N) rounds: the number of times you halve N to reach one.

Total time also depends on which matches can overlap. Our six-player event has two opening matches, two semifinals, and one final. If every contest has a reserved 20-minute slot, two stations can fit those rounds into three slots, or 60 minutes. One station needs five slots, or 100 minutes. These are illustrative playing schedules before adding check-in, breaks, delays, and awards; real match lengths may vary.

Dividing total match time by station count can underestimate duration. A final must wait for both semifinals, even if a station is free.

Double elimination counts losses

For strict double elimination, count two losses for every non-champion. Four entrants leave three non-champions, accounting for six losses. Because each bracket contest adds one loss, an undefeated champion means six matches. If the champion has one loss, the total is seven.

In general, that is 2N − 2 matches with an undefeated champion, or 2N − 1 with a one-loss champion. The latter is the reset scenario in the traditional upper/lower structure. Six entrants therefore require ten or eleven matches; sixteen require thirty or thirty-one.

These totals assume a completed two-entrant-per-match event, elimination at the second match loss, a reset when needed, and no additional placement contests. A conventional upper/lower event ending in a single, non-reset grand final instead uses 2N − 2 contests, regardless of which finalist wins. Alternative finals rules and formats that carry results across stages need their own count.

For four entrants, you can verify the traditional structure directly: three upper-bracket matches, two lower-bracket matches, then one grand final and a possible reset. That is six or seven. It is nearly twice single elimination's three matches, though parallel stations mean elapsed time need not double.

Fit the field with byes and balanced seeds

A bye advances an entrant without an opponent. In the conventional bracket used here, put the field into the smallest power-of-two capacity that can hold it, then use separate opening-round byes to fill the shortfall.

Six entrants fit into eight positions, leaving two byes. Ten entrants fit into sixteen, leaving six. Twelve fit into sixteen, leaving four. Once the opening matches finish, the remaining field fits the next round exactly.

For at least two entrants, the compact calculation is P = 2^ceil(log₂(N)), where P means capacity. “Ceil” means round upward: find how many doubling steps are needed, round up to a whole step, and use that power of two. Then byes = P − N.

If each bye has a real recipient and no opening pairing contains two empty positions, the number of played opening matches is N − P/2. For ten entrants, two opening matches remove two people; their winners join six bye recipients to make eight remaining entrants.

This is one practical construction, not a rule that every competition must use. Some formats deliberately introduce entrants at later stages. Here, keeping byes in the opening round makes the rest of the fixed bracket easy to follow.

A seed number needs the right position

A seed is an entrant's assigned position in an ordering, usually based on prior performance. Standard balanced placement spreads the leading seeds across the tree so that they cannot meet too early.

For eight positions, one valid order from top to bottom is 1, 8, 4, 5, 2, 7, 3, 6. Adjacent entries play each other: 1 versus 8, then 4 versus 5, then 2 versus 7, then 3 versus 6. Winners of the first two matches meet in one semifinal; winners of the last two meet in the other.

Seeds 1 and 2 are in opposite halves. If higher seeds always win, the semifinals are 1 versus 4 and 2 versus 3, followed by 1 versus 2. Mirroring a branch can produce an equivalent layout.

Simply listing 1 versus 8, 2 versus 7, 3 versus 6, and 4 versus 5 in that order gets the opening pairings right but puts seeds 1 and 2 on course to meet in a semifinal. Check the paths as well as the pairs.

For six ranked entrants, use the eight-position layout and leave notional seeds 7 and 8 empty. Seeds 1 and 2 receive the byes, in opposite halves. That rewards their ranking through one fewer required win. It is a deliberate advantage, not equal work for everyone.

“Snake” or serpentine seeding often describes distributing rankings back and forth across groups. It is clearer to call the elimination arrangement above balanced bracket placement, and show exactly how it connects.

Decide what you mean by fair

Start with the quality of the rankings. Recent results against comparable opponents may justify protecting top seeds. Unfamiliar entrants with no shared competition history may be better served by a published random draw. With partial information, seed bands offer a compromise: separate a trusted leading group, then draw within the permitted positions.

Publish how rankings are established, how tied seeds are resolved, and what restrictions apply before drawing. Avoid repeatedly redrawing until the organizer likes the result. A constrained draw also needs care: a procedure that uses randomness does not necessarily make every permitted bracket equally likely.

It helps to separate the promises:

• Structural fairness: the opportunities, loss allowances, and advancement paths the format provides.
• Procedural fairness: whether published rules are applied consistently, including the draw and corrections.
• Competitive balance: how evenly matched opponents or groups are, which may compete with a policy of protecting top seeds.
• Random assignment: how positions are selected; a uniform assignment gives each entrant equal selection probability for a particular position.
• Championship probability: the chance of winning the whole event, given the draw, opponent strengths, and match assumptions.

These are working distinctions for organizing decisions, rather than one universal fairness standard. A research review by Devriesere, Csató, and Goossens documents differing fairness objectives and explains why a seeding policy's merits depend on what the tournament is meant to achieve.

Equal players can receive unequal title chances

Consider six equally matched players in our fixed bracket. Assume every contest is an independent 50–50 result, there are no draws or forfeits, and no reseeding occurs. A bye recipient needs two wins: half of a half is 1/4, or 25%. An entrant playing from the opening round needs three: half of a half of a half is 1/8, or 12.5%.

The totals check: two entrants at 25% plus four at 12.5% add to 100%. The bye changes the route even though everyone has identical match ability.

Now uniformly assign the six people to those six occupied positions before play. Each has a two-in-six chance of receiving a bye. Before the draw, each person's overall title probability is (2/6 × 1/4) + (4/6 × 1/8) = 1/6. After the draw, their known paths have different probabilities. Real differences in ability would change the calculation further.

Random placement therefore provides an agreed opportunity to receive a position; it cannot promise identical realized opponents or equal title chances. Ranked placement makes a different promise: prior performance earns protection. Neither guarantees that the strongest player wins, and neither is the right policy for every community.

When everyone needs more matches

If sending someone home after one match would defeat the purpose of the evening, consider a preliminary format that keeps people playing.

In a single round robin, everyone meets everyone else once within the defined field or group. Four entrants produce six pairings, and each entrant plays three times. Count N choices of entrant, each with N − 1 opponents, then divide by two because that counted each pair twice: N(N − 1)/2 matches. Six entrants need fifteen; sixteen need 120.

In a Swiss system, pairings develop between rounds, generally bringing entrants with similar scores together. The field does not have to meet every opponent, and an ordinary loss does not eliminate an entrant. FIDE's basic Swiss-system rules provide a formal example covering similar-score pairings, restrictions on repeat opponents, and pairing-allocated byes. Those are chess regulations, not automatic rules for every game using a Swiss-style event.

For an even, unchanged field where everyone plays once per round, each round has N/2 matches. Across R rounds, that is RN/2. Thirty-two entrants over five rounds need eighty matches. A top-eight single-elimination finish adds seven, giving 87. With odd numbers, withdrawals, or forfeits, count the actual scheduled and played contests separately.

Set the number of Swiss rounds and the qualification cutoff before play. Five rounds in that example are an organizing choice, not proof that eight qualifiers will be unambiguously separated. Entrants can tie at the cutoff, and the pairing and tiebreak rules must explain what happens.

Groups and the route into a knockout

An eight-player event could instead use two four-player round-robin groups. Each group needs six matches, so pool play uses twelve. Advance the top two from each group into semifinals, then a final: three more matches, or fifteen overall. Every entrant gets three group matches if everyone plays as scheduled.

Publish the semifinal mapping: Group A's winner faces Group B's runner-up, while Group B's winner faces Group A's runner-up. Previous group opponents can then meet again only in the final. A third-place match would add one contest.

For G equal groups of s entrants, group play requires G × s(s − 1)/2 matches. If Q qualifiers enter a single-elimination title bracket, add Q − 1. Unequal groups require counting each group's pairings separately and deciding how any cross-group comparison will work.

Group allocation deserves the same attention as bracket seeding. Ranking bands can distribute known strength; a random draw avoids pretending to know rankings that do not exist. Neither makes every group equally difficult in practice.

Decide qualification and knockout placement together. Unusual comparisons between groups or a fixed route to opponents can create incentives that the organizer did not intend. Reseeding survivors is an option if announced, but it sacrifices fixed future paths and does not establish a universal guarantee against strategic losing.

Run the six-player event from registration to champion

Return to our fictional gaming night. The organizer closes check-in with Asha, Ben, Cora, Dev, Emi, and Finn present. A previously published club ranking assigns seeds 1 through 6 in that order. The announced format is single elimination, with byes rewarding seeds 1 and 2. Matches are best-of-three games; the first player to win two games advances.

The organizer publishes the roster, match identifiers, station sequence, and winner destinations before play. The two halves connect as follows. “Winner of M1” means whoever wins that contest, not a prediction.

Here is one possible completed event, match by match:

1. M1: Dev beats Emi, two games to one. Dev advances to M3; Emi leaves championship contention.
2. M2: Cora beats Finn, two games to zero. Cora advances to M4; Finn leaves championship contention.
3. M3: Asha beats Dev, two games to zero. Asha advances to M5; Dev is eliminated.
4. M4: Cora beats Ben, two games to one. Cora advances to M5; Ben is eliminated. Protecting Ben's seed did not guarantee a win.
5. M5: Asha beats Cora, two games to one. Asha is champion; Cora is runner-up.

Five bracket matches contained thirteen individual games: three, two, two, three, and three. Asha won two matches after a bye; Cora played three. Each non-champion lost one bracket match. The records do not settle a unique order between Dev and Ben, or between Emi and Finn, unless an additional placement policy was announced.

The organizer can now mark the bracket complete and share the final version. A casual play area could have given eliminated entrants more games during the event; those games would be separate from the five-match title bracket and would need their own station capacity.

Build the six-player bracket in the current studio

The Sandbox bracket studio supports single- and double-elimination layouts, seeded rosters, series scores, and bracket sharing/export controls. Use the six-player single-elimination example above for a first run:

1. Open Studio Controls, choose Single Elimination, and set Series Format to Best of 3.
2. Under Seeded Participants, put Asha, Ben, Cora, Dev, Emi, and Finn on separate lines in that seed order, then select Apply Teams. Roster order determines seeds; do not shuffle if the published ranking should determine placement.
3. Turn off 3rd Place Match if your event should contain only the five title matches. Confirm that seeds 1 and 2 receive byes and trace both semifinal paths before play.
4. Use Slot Click Action to choose between recording game wins and Instant Advance for a complete series result. Record what actually happened; Simulate is for rehearsal rather than entering real event results.
5. Use the match inspector for score details and check the downstream opponent after each completed match. Save or share only the version you have checked.

The current double-elimination studio ends with one grand final and no automatic grand finals reset. A lower-bracket finalist can therefore win the title in that final even though the upper finalist has suffered only one bracket loss. If your published rules require a reset-based two-loss championship, use a separate authoritative record or a tool that supports that procedure; the studio's general Reset action restarts progression and is not a grand finals reset.

Swiss pairings and round-robin group standings described earlier are alternative organizing formats, not controls supplied by this bracket studio. Share URL creates a snapshot link rather than a promise of live synchronized standings. Review the exported PNG or JSON before distributing it, and retain the event's own rules and result record.

For a large-screen event, the projector setup guide covers rehearsal and readability checks. A bracket that fits your screen still needs readable participant names and clear next-match instructions.

Publish the rules that boxes cannot show

Agree on the handling of tied matches separately from tied standings. A knockout contest needs one advancing winner, perhaps through an additional game, overtime, or another published procedure appropriate to the activity. A group or Swiss table needs a method for ranking equal records or deciding which tied positions may remain shared.

Standings criteria might include head-to-head results, a table restricted to tied entrants, score differential, total scoring, opponent strength, or a playoff. Head-to-head alone cannot resolve every cycle: A can beat B, B beat C, and C beat A. Strength-of-schedule measures can help when opponents differ, but their treatment of byes and unplayed rounds needs rules too.

FIDE's tiebreak regulations illustrate an ordered list published before competition. They do not prescribe a universal hierarchy for community tournaments. Choose rules suited to the game and explain which matches count, whether criteria restart after a tie splits, and what happens if the list is exhausted. Random selection is an option only where the event's rules permit it.

A small tiebreak you can check

In an illustrative three-player group, A beats B 10–8, B beats C 10–7, and C beats A 10–9. Each finishes with one win and one loss. Suppose the announced rules compare wins, then overall score differential, then overall points scored, applying each successive criterion only to entrants still tied.

A scores nineteen and concedes eighteen: +1. B scores eighteen and concedes seventeen: +1. C scores seventeen and concedes nineteen: −2. Differential places C third. A and B remain tied, so nineteen points scored beats eighteen, placing A first and B second.

That result follows this example's rules. A different published hierarchy could work differently. Differential also creates incentives to increase winning margins; organizers can consider a declared per-match cap where appropriate, but must calculate the standings using that cap consistently.

Absences, withdrawals, and corrections

Publish a check-in deadline, a no-show grace period, and who can authorize a forfeit. Before the first match, a changed roster may justify rebuilding the draw under an announced policy. Once play starts, quietly reseeding everyone changes paths people have already competed on.

If someone withdraws, mark the affected match as unplayed or forfeited under the event rules. Record the advancement without inventing a played score. In double elimination, a withdrawal also needs explicit handling of any remaining lower-bracket route; an absent entrant should not reappear through an automatic destination.

Give participants a way to challenge an incorrect result and a deadline for doing so. Correcting an early result can affect later opponents, so pause affected matches, let the designated organizer resolve the consequences, and publish the amended record. Preserve the earlier entry and the reason for correction rather than leaving two unexplained versions in circulation.

Make the bracket usable throughout the event

At the venue, the most urgent questions are usually personal: who am I playing, where do I go, and what happens afterward? Show participant names, match IDs, round or stage, status, and winner destinations clearly. Add loser destinations for double elimination. Keep identifiers stable even when a displayed name is corrected.

Use text such as “winner,” “eliminated,” and “bye” alongside visual emphasis. Color alone cannot carry those meanings, especially on monochrome printouts. W3C's color guidance explains why additional cues matter.

Large brackets need different views for different jobs. A projector can show the active round at readable size; a phone-friendly round list can explain the next matches without shrinking an entire tree. Keep connector lines consistent where a tree is shown. Make all navigation operable from a keyboard, and use readable contrast. W3C's reflow guidance recognizes that some diagrams need two dimensions, while encouraging reduced scrolling and reflowing surrounding text.

Maintain one authoritative result record, whether it is on paper, in a spreadsheet, or in tournament software. Give score sheets a match ID, entrants, result fields, and an organizer contact. Keep contact details and other unnecessary personal information off public standings.

When software supports exports, check the actual output before promising a printable bracket or downloadable file. A clear print layout, a large-screen display, and a social image have different reading distances and purposes. Label public snapshots with a version or update time; replace outdated wall copies and provide a clear route to the current standings.

Before opening registration, be able to answer these questions:

• How many matches can the event accommodate, including any reset or placement matches?
• What does each entrant receive: a title chance, a minimum number of games, or both?
• How are seeds and byes assigned, and which advancement routes are fixed?
• How are tied matches, tied standings, absences, and disputed results resolved?
• Where will people find their next match and the authoritative results?

A bracket works when those answers agree with what happens at the stations. The mathematics gives you the workload and the paths. Clear rules and careful result handling make those paths usable for the people playing.

Sources and research notes

The match counts, six-player results, bye calculations, and probability examples above are independent derivations or explicitly fictional illustrations. The probability example assumes independent, equally likely match outcomes; it is not a forecast for a real game or a claim about any digital picker. All title-bracket totals exclude additional placement matches unless stated.

The tournament-design review supports the discussion of competing objectives; it does not establish one universally fairest format. Evo supplies a specific reset-enabled example, and FIDE supplies specific Swiss and standings rules. Their regulations are cited where used, without treating one competition's policies as rules for every event. For the different job of selecting prizes by chance, see our guide to transparent community giveaways.

The linked sources and the current studio's format, roster, scoring, and export controls were reviewed on 21 October 2026. The worked example uses single elimination; the studio's single-grand-final double-elimination variant is identified separately above.