Start with the one to avoid
Two captains stand out the front and take turns picking. It's the method almost everyone learned, and it's the worst one on every measure that matters.
It's the slowest — two people deliberating in front of an audience takes minutes, not seconds. It produces the least balanced teams, not the most, because captains pick their mates and the people they've seen play before, not the best available player. And it does something no other method does: it publicly ranks a group of people, worst to best, in front of each other, and announces the result out loud.
That last part is the whole problem. Being picked last isn't a sporting disadvantage — it's a small public verdict, delivered by your peers, and people remember theirs for decades. There is no version of captains-pick that fixes it, because the ranking is the mechanism. You can reverse the order, pick in secret, or have the teacher be a captain, and at that point you've just invented a slower version of one of the methods below.
The one setting where it's defensible: a group of adults who all know each other well, playing something where nobody minds, with a rule that the captains alternate every week. Even then it's the slow option.
The six methods, compared
| Method | Time | Balance | Social cost | Best for |
|---|---|---|---|---|
| Random shuffle | Seconds | Luck | None | Almost everything |
| Count off 1-2-1-2 | Seconds | Luck | Low | Already in a line |
| Card draw | A minute | Luck | None | Kids, parties |
| Snake draft by rating | Two minutes | Good | Low if ratings stay private | Competitive sport |
| Split the top, randomise the rest | A minute | Decent | Low | One or two ring-ins |
| Captains alternate picks | Five minutes | Poor | High | Very little |
Random shuffle
Shuffle the whole list and deal it out one name at a time, alternating teams — that dealing order is what keeps the sizes within one of each other. It's the default for a reason: it takes seconds, it's transparently unriggable, nobody is ranked, and it changes every single week, which quietly solves the clique problem all on its own.
Its honest weakness is that random is random. Occasionally one side gets all four fast kids. If that matters, use one of the balanced methods below — but decide before you shuffle, because re-shuffling until someone's happy is captains-pick with extra steps.
Count off
"One, two, one, two" down the line. Instant, zero equipment, and every PE teacher's fallback. The catch is that it isn't actually random: people who want to be together work out where to stand roughly nine seconds after they learn how it works. If you use it regularly, shuffle the line first, or count off in threes and then combine.
Card draw
Deal one playing card each. Red versus black for two teams; the four suits for four teams. It's tactile, it's visibly fair, and small children find it genuinely exciting, which is not nothing. Also works with lolly wrappers, coloured pegs, or anything you have twenty of.
Snake draft by rating
This is the one that actually balances. Rate everyone 1 to 5 — you can do it yourself in thirty seconds, or ask people to self-rate, which is surprisingly accurate when nothing's at stake. Sort best to worst, then deal in a snake: down the teams, then back up, then down again.
Round 2 C B A
Round 3 A B C
Round 4 C B A
Straight round-robin dealing would give team A the 1st, 4th and 7th best players and team C the 3rd, 6th and 9th — a real gap. The snake evens that out: A gets 1st and 6th, C gets 3rd and 4th. It's the same trick that balances tournament pools.
One rule: keep the ratings to yourself. A published list of who you think is a 2 is captains-pick again, just written down.
Split the top, randomise the rest
The lazy version of a balanced draft, and often the right one. Name the two or three players everyone agrees are the strongest, put one on each team, then shuffle everybody else in. You get most of the balance for a tenth of the fuss, and you only had to make a judgement about the people who won't mind you making it.
The odd person out
Eleven people, two teams. Somebody's short a player, and this is where most team-picking goes wrong socially: whoever is "extra" gets treated as a problem to solve rather than a person who came to play.
Options that work, roughly best first:
- Just play 6 v 5. One player's difference matters far less than people think, especially in a casual game. Give the short side first possession, or the first serve, and get on with it.
- Rolling sub. Even teams, one player rotates on and off every few minutes. Everyone gets a rest and nobody is the odd one — but only if the rotation is a fixed order, not "whoever's tired".
- The all-time player. One person plays for whichever side has the ball — all-time quarterback, all-time keeper, permanent wicketkeeper. Genuinely fun, and it's usually the most-wanted job.
- Referee or scorer, rotated. Fine as long as it rotates every game and it isn't always the same kid. If it's always the same kid, you haven't solved anything — you've just made the exclusion official.
For pair work, the equivalent is a group of three, never a leftover of one. And for anything where numbers wobble weekly, split into teams of a size that divides both ways: teams of four out of a group that's usually fifteen to seventeen will always land somewhere workable.
Different rooms, different rules
Classrooms
Random is the safest default, because it protects the two children who most need protecting: the one who is always picked last, and the one nobody picks at all. Deliver the teams as a finished fact — read them out, don't negotiate them. If there are pairs you know shouldn't be together, sort that out before the lesson and present the result as the draw; nobody needs to know a hand was on the scale.
The exception is deliberate grouping for learning — reading groups, mixed-ability project teams — where the whole point is that you chose. Just don't call that a random draw, and don't do the choosing in the room.
Sport and training nights
Balance matters more here, because a 9-0 game is nobody's night out. Use the snake draft, or split the top and randomise. And keep the better tool in your back pocket: handicaps instead of stacking. Let the stronger side play a player short, or give the weaker side a goal start, or make everyone use their weak foot. Adjusting the game is almost always better than agonising over the teams.
The other move worth knowing is the half-time swap: if one side is three goals up, swap two players across. Announce that rule at the start so it reads as a format, not a judgement.
Trivia and pub quizzes
Self-selection is the enemy — friends cluster, and friends often know the same things. Random tables mix the knowledge as well as the people, and a team with one person who knows sport and one who knows musicals will beat four people who all know the same decade. Four to six per team is the sweet spot; past six, the quiet ones stop offering answers and the loud ones stop hearing them.
Work
"Find a partner" is the worst instruction in workshop facilitation: it's fast for the confident and awful for the new starter. Random assignment takes the same ten seconds and strands nobody. For breakouts where you want cross-functional mixing, split the obvious clusters deliberately — one from each team per group — then randomise inside that.
When the same people keep ending up together
Every regular group develops this problem, and there are only three real fixes:
- Re-shuffle from scratch every session. Not "same teams as last week unless someone objects" — that's how a Tuesday kickabout turns into a closed shop over about six weeks.
- Write last week's teams down. If more than half a team is repeating, shuffle again. One agreed re-roll rule, applied by a written condition rather than by whoever complains loudest.
- Ban the veto. The moment "I'll swap with you" is allowed after the draw, the draw is decorative. If swaps happen, they happen before the shuffle and for real reasons — an injury, a lift home, a kid who needs to be with their sibling.
And a note on cliques that isn't about the method: a group where people lobby to be together is usually a group where losing feels worse than it should. Shorter games, more of them, and re-picking sides between each one does more for that than any algorithm.
FAQ
What's the fairest way to split into teams?
For fairness of process, a random shuffle — every arrangement is equally likely and nobody is ranked in public. For fairness of outcome, a snake draft on private ratings, which spreads the strongest players evenly. Most groups want the first and think they want the second.
Is it bad to let captains pick teams?
It's the slowest method, it produces less balanced teams than a rated draft because captains pick friends and familiar faces, and it publicly ranks everyone in front of each other. Being picked last is a verdict people remember for years. Any of the other five methods is better.
How do you handle an odd number of players?
Uneven by one is usually fine — play 6 v 5 and give the short side first possession. Otherwise use a rolling substitute in a fixed rotation, or an all-time player who plays for whichever side has the ball. If someone referees instead, rotate it every game so it's never the same person.
How do you make teams even in skill?
Rate everyone 1 to 5, sort them, and deal in a snake — down the teams, then back up, then down again — so the first pick and the last pick land on the same side. Keep the ratings private. Or adjust the game instead: handicaps, a goal start, or the stronger side playing a player short.
How do you stop the same people ending up together every week?
Re-shuffle from scratch every session rather than defaulting to last week's sides, write the teams down so you can spot repeats, and don't allow swaps after the draw. A fresh random shuffle each week solves it on its own, as long as nobody is allowed to renegotiate the result.
What team size works best?
For trivia and discussion, four to six — past six people stop contributing. For sport, whatever the game needs, but more smaller-sided games beat fewer big ones because everyone touches the ball more. When in doubt, more teams and shorter games.
Or just paste the list
None of this needs a tool — a deck of cards and a rule you stick to will do it. But if you'd rather not shuffle names in your head every Tuesday, the team picker takes your pasted list and deals it into even teams with a proper unbiased shuffle: pick a number of teams, or a team size, and it works the rest out. Nothing is sent anywhere — the shuffle runs on your own device — and it's free with no signup, which matters if the list is a class roll.