Group Stage into a Knockout: How to Run Pool Play

Why the hybrid exists

A single-elimination bracket with 24 entries plays 23 matches, and twelve of those entries play exactly once. For a team that travelled to your venue, that is a bad day out, and it is the single most common complaint about knockout events.

A full round robin fixes that completely - everyone plays everyone - but the match count grows quadratically. Twenty-four entries in one round robin is 276 matches, which is not a tournament, it is a season.

Pool play splits the difference. Break 24 entries into four groups of six, and each group is a manageable 15-match round robin where everyone plays five games. That is 60 matches total, everyone gets a real day, and the top two from each group give you a clean eight-team bracket for the afternoon.

The other benefit is structural: the knockout is now exactly the size you chose, so the awkward field size problem disappears along with the byes it would have caused.

Choosing the group size

Group size is the main decision, and it is a trade between guaranteed games and total match count. A group of n plays n(n − 1)/2 matches and gives every entry n − 1 games.

Groups of four are the standard for large fields: three guaranteed games each, six matches per group, and the round robin completes in three rounds. Groups of five or six give more games and more confidence in the standings, at a cost that rises fast. Groups of three are possible but thin - two games each, and a single result can decide qualification.

Prefer equal group sizes. Unequal groups mean unequal numbers of games, which makes any cross-group comparison unfair, and it will be noticed the moment a qualification decision hinges on it. If your field does not divide evenly, it is usually better to adjust the number of groups than to run one group a size larger.

The other constraint is wall-clock time: groups run in parallel if you have the space. Four groups of four on four courts is three rounds of four simultaneous matches - around an hour if matches are 20 minutes. The same 24 matches on one court is a full day.

How many advance, and the trap in the middle

Advancing the top two from each group is the default for a reason: it produces a qualifier count that is a power of two whenever the number of groups is, which means a knockout bracket with no byes.

Four groups advancing two gives eight - a clean bracket. Eight groups advancing two gives sixteen. Two groups advancing two gives four. All tidy.

The trap is a group count that is not a power of two. Three groups advancing two gives six qualifiers, which needs an eight-slot bracket with two byes, and now you have to decide which two of your six qualifiers get a free pass. That is a real decision with real consequences and it is better made in advance, in writing, than at 4pm in a car park.

The usual fixes are to advance the three group winners plus the best runner-up, or to give the byes to the two best group winners on a stated criterion. Either works. What does not work is inventing the rule after the groups have finished - publish it before the first match, and the argument never happens.

• Top two from each group, with a power-of-two number of groups - the clean default.
• Group winners only - a smaller, sharper knockout; requires a group count that is a power of two.
• Top two plus best third-placed entries - flexible, but needs a published cross-group comparison rule.
• Whatever you choose, publish it before the first match, including how ties are broken.

Seeding the groups, then seeding the bracket

There are two separate seeding jobs in a hybrid event and they are easy to conflate.

The first is filling the groups. The goal is that no group is much stronger than another, which means distributing your ranked entries evenly rather than randomly. If you have any basis for ranking, snake the field across the groups - best entry to group A, second to group B, third to group C, fourth to group D, then reverse for the next four. Keeping entries from the same club apart is the other constraint here, and it is easier to satisfy at this stage than later.

The second is filling the bracket. The standard cross-over is that each group winner faces a different group's runner-up, so that entries from the same group cannot meet again until the final. Winner of A plays runner-up of B, winner of B plays runner-up of A, and so on across the field. It rewards winning your group and avoids a repeat fixture in round one.

Once you have the qualifier list in that order, the knockout is an ordinary seeded bracket: enter the qualifiers in rank order and the standard seed layout does the rest, keeping the strongest apart until late.

Running it in practice

Build each group as its own round robin, play it out, and read the final standings. The standings order - wins first, then draws, then the point-based tiebreakers you configured - is what produces your qualifier list, so the tiebreakers are worth setting deliberately before you start rather than discovering under pressure.

Then build the knockout as a separate bracket, entering the qualifiers in the cross-over order above. Keeping the two stages as two tournaments is simpler to run and much easier to explain to participants than any single all-in-one structure.

Two practical notes. Group matches are the only part of the day that parallelises freely, so schedule them across every court you have and accept that the knockout will be more sequential. And leave slack between the stages: the knockout cannot start until the last group match finishes, and it is always the last group match that runs long.

Roundra builds round robins up to 32 entries and knockout brackets for any field, free in the browser with no account - which is all the machinery a pool-play event needs.

Frequently asked questions

Keep going