What a format is trying to do
A tournament format has two jobs that pull in opposite directions. It should identify the strongest team as reliably as possible, and it should fit within a limited amount of time, venue space and broadcast schedule. Every format is a compromise between those aims.
The comparisons below assume results contain chance, so even the better side can lose any individual match. Without that, every format would produce the same winner.
Single elimination
In a single-elimination bracket, each loss ends a team’s run. With N teams, exactly N - 1 matches are played, since every match eliminates one team and only one remains.
That efficiency has a cost. Suppose the best of eight teams has a 0.75 chance of beating any opponent in a match. It must win three rounds, so its chance of winning the title is 0.75 x 0.75 x 0.75 = 0.4219, about 42%. With sixteen teams and four rounds, it falls to 0.75^4, about 0.316, or 32%.
Those figures treat the favourite as equally strong against everyone. In practice a top seed usually faces weaker opponents earlier, so the chance is somewhat higher, and the reverse can happen to teams that meet the strongest rivals early. Still, the pattern is clear: one bad match ends everything, and the more rounds there are, the more chances for it to happen. Using longer matches helps, and the effect is covered in the best-of series calculations: series length raises the favourite’s chance in each round.
Seeding also matters. A bracket arranged so the strongest teams meet late protects them from early collisions, but it also means that the final stages depend on an assumption that the seeding is right.
Double elimination
In a double-elimination bracket, a team must lose twice to be out. The structure is a winners bracket and a losers bracket, with the two finalists meeting at the end.
Every team except the champion is eliminated with two losses, so the total number of matches is typically 2N - 2, or 2N - 1 if the final requires a rematch because the team from the losers bracket wins the first meeting. For eight teams, that is 14 or 15 matches, double the single-elimination count.
The advantage is forgiveness. One poor match no longer ends a run, so the best team is more likely to survive to the end. The cost is time, and the price of the losers bracket is that a team coming through it plays more matches than a team that never lost.
Round robin
In a round robin, every team plays every other team. With N teams that requires N(N - 1) / 2 matches, which is 28 for eight teams, four times the single-elimination count.
The advantage is the volume of evidence. Each team plays seven matches rather than at most three. If the best team beats each opponent with probability 0.75, it is expected to win 7 x 0.75 = 5.25 of its seven matches. Because the final table is based on many results, luck has more room to average out, in line with the law of large numbers, and the strongest team is typically more likely to finish first than in a bracket of the same size.
The drawbacks are the schedule and the finish. The match count grows with the square of N, which becomes unrealistic for large fields. Ties in the standings need tiebreakers, and late matches between teams already decided can carry little meaning. Round robins are also exposed to team strength mismatch: a table of a few weak teams adds many matches that tell us little.
Swiss system
A Swiss format sits between the two. Teams play a fixed number of rounds, and in each round they are paired against others with similar records. Nobody is eliminated early in the pure form, though many esports events add an advance-at-X-wins, out-at-X-losses rule.
The number of rounds is usually small, often around log base 2 of the number of teams. For eight teams, three rounds of four matches gives 12 matches in total, less than half of a full round robin. After three rounds exactly one team can be 3-0, so it can separate a clear leader but cannot fully rank the rest.
Swiss has several useful properties:
- Early on, strong teams tend to meet weaker ones, and later rounds pair teams with similar records, so matches become more informative.
- Weaker teams are not eliminated after a single bad day, and they keep playing opponents near their level.
- The total number of matches stays manageable even for large fields.
Its weaknesses are pairing complexity and the fact that a final ranking can depend on opponent strength, which is awkward to explain to viewers. As with any ranking based on few games, the ordering in the middle of the table is noisy.
Comparing the formats
| Format | Matches for 8 teams | Treatment of one bad match |
|---|---|---|
| Single elimination | 7 | Ends the run |
| Double elimination | 14 or 15 | Survivable once |
| Round robin | 28 | Rarely decisive |
| Swiss (3 rounds) | 12 | Usually survivable |
Choosing sensibly
No format is simply best. Single elimination is quick and dramatic, and also the most dependent on luck. Round robin gives the most evidence per team but takes the most time. Double elimination and Swiss are practical middle paths. Often organisers combine them, using a Swiss or group stage to qualify teams for a final bracket with longer matches, reserving the most luck-sensitive part for the end where the audience wants it.
The takeaway for fans and analysts is modest. A champion from a short bracket is not necessarily the strongest team, and a table topper from a long league is somewhat more likely to be. Neither guarantees it. Even in the format most kind to the favourite, the outcome stays uncertain.