Comfort

Comfort

Three dials, four switches. Set once, remembered on this device only.

60%
10 or below holds everything still. 100 = full physics. Collisions still land within 50 ms.
0%
Starts at 0. Unlocks on your first tap, never on its own.
75%
Strength of the six inks. 0 is grey. Each chip's name is in its label.

Random team generator

Fair random teams from any list.

12 names. Bullets and numbering get stripped. For a rating, write Aisha, 8.
Team names, leftovers and rating balance

Ratings come from the names list. Write a line as Aisha, 8 and that 8 is her rating.

Keep people apart or together

No rules yet. Every arrangement is checked properly, so if a set of rules cannot work you get told instead of getting a quiet wrong answer.

    Saved rosters

      Rosters live in this browser on this device only. Nothing is sent anywhere. Clearing your browser data clears them.

      How a random team generator splits a list fairly

      It shuffles the whole list into a random order, then deals people out across the teams like cards. That is the only method that treats everyone the same, because every possible ordering of your list is equally likely before the dealing starts. This page does the shuffle with crypto.getRandomValues, the browser's own cryptographic random source, and a Fisher Yates pass. Nobody has a better chance of landing with a friend than anyone else.

      Most free team pickers get this part wrong in the same way. They sort the list with a comparator that returns a random number, which looks clever and is not a shuffle at all. A sort needs consistent answers to work, and a random comparator breaks that, so the result depends on the sorting algorithm hiding inside the browser. In practice people barely move from where they started. We ran both methods 200,000 times over a five item list. The proper shuffle hit all 120 orderings within 6% of even. The random comparator was out by over 1,000% on some orderings, landing on one of them 18,651 times where 1,667 was fair.

      What happens to the leftover people

      They go one each to the earliest teams, and the page says out loud which teams got them. Thirteen people in four teams becomes 4, 3, 3, 3, and the line under the result names the team that ended up bigger. That matters when someone is counting heads and feels hard done by. If you split by people per team instead, you choose what the leftovers do: join the other teams one each, or form their own smaller team. Neither option is hidden from you.

      Two other cases get said plainly rather than fudged. Ask for more teams than you have people and the spare teams come back empty with a note explaining why. Ask for teams of eight when only three people turned up and you get one team of three, and the page tells you that is what happened.

      Keeping two people apart, or together

      Mark a pair as keep apart and they never share a team. Mark a pair as keep together and they always do. Keep together is treated as contagious, so if you glue Ana to Ben and Ben to Cem, all three move as one lump. The generator then searches every possible arrangement of those lumps across the teams, hardest ones first, and stops on the first arrangement that breaks no rule.

      Because the search is exhaustive, running out of options is proof that nothing works, so the page will tell you a set of rules is impossible only when it genuinely is. Three people who all refuse each other cannot fit into two teams, and you get told that in one sentence instead of watching a spinner. The same goes for a pair told to keep apart and keep together at once, and for a glued group bigger than any team. This kind of puzzle is NP hard in general, so on a wildly tangled set the search can run out of its step budget. If that happens the page says it gave up, which is a different claim from impossible, and it never quietly drops one of your rules to make the numbers work.

      Balancing teams by rating

      Put a number after a name, like Aisha, 8, turn on balance by rating, and the teams come out with close totals instead of random ones. The method is longest processing time first: sort everyone by rating, biggest first, and give each person to whichever team has the lowest total so far and still has room. It is fast, it is the standard approach, and it is a heuristic, not a perfect answer. Splitting numbers into groups with equal sums is the partition problem, which is NP hard, so nothing short of brute force guarantees the best possible split.

      What you get instead is the real spread printed under the result: the gap between the heaviest and lightest team. On a skewed set of twelve players rated 0 to 100 split four ways, this lands around a spread of 6, where ignoring ratings leaves a gap of 100 or more. Some sets cannot be balanced at all. One player rated 500 among nine rated 1 will always leave one team miles ahead, and the spread tells you that rather than pretending otherwise.

      Why your class list is still there next week

      Saved rosters go into this browser's own storage under one key, on this device, and that is the end of the journey. There is no account, no server and no upload. Save a roster once and the names, the ratings and the keep apart rules all come back next Monday. Clearing your browser data clears them, and so does using a different device or a private window, which is the honest trade for never asking you to sign up.

      Good for a classroom, a five a side, or a standup

      Teachers use it to break up the same four people who always work together, which is exactly what keep apart is for. Coaches use ratings so one side is not stacked. Managers paste a standup list and split it in half. Copy as text gives you something to paste into a message, and copy as table gives tab separated columns that land as a proper grid in Sheets or Excel.

      One honest limit: the shuffle is random every time, with no memory. Press shuffle again and the same two people can land together twice in a row, because that is what random means. If you need people to rotate through different teammates week after week, keep apart rules are the tool for it.

      Related tools

      Everything here runs in your browser. Nothing you type is sent anywhere or stored on our side.