Ten friends, one custom lobby, three maps before people start drifting off to bed. The games themselves take care of themselves. What eats the first twenty minutes is the setup: who’s on which side, which map, and whether the joke rules everyone agreed to last time are still on. This guide runs a whole CS2 night with three of the site’s tools in the order you’d use them, the Balanced Team Generator, the Random Game Map Picker and the Random Strat Roulette Generator, with the odds for each step so the host knows what each click is really doing.
The night: ten players, 5v5, three maps, one host who runs every tool from one browser and shares the screen on Discord. The last part matters more than it sounds, and we’ll come back to it.
Before anyone joins: tag the ten players honestly
Random teams from ten friends of mixed ability are lopsided often enough that someone always asks for a redo. The balancer avoids that by spreading a tag evenly instead of a name. Each line is a name with a tag in square brackets at the end, like Mo [A]. The tool groups people by tag, shuffles each group, and deals each group round-robin across the teams. The detail that matters here is that it carries its position on from one tag to the next, so if one team gets the extra player of one tier, the deal starts the next tier on the other team.
How you size the tiers decides how even the teams come out. Score the tiers A = 3, B = 2, C = 1 for a rough strength total and compare two ways of tagging the same ten people:
| Tiers (A / B / C) | Team 1 gets | Team 2 gets | Strength gap |
|---|---|---|---|
| 3 / 4 / 3 | 2 A, 2 B, 1 C (11) | 1 A, 2 B, 2 C (9) | 2 points |
| 4 / 4 / 2 | 2 A, 2 B, 1 C (11) | 2 A, 2 B, 1 C (11) | none |
With odd-sized tiers, team sizes still come out 5 and 5 because of the carry-over, but one side has the extra top player and pays for it with an extra weaker one. That’s roughly even, not mirrored. If you can honestly group your ten into tiers of 4, 4 and 2, every tier splits down the middle and both sides get the same shape. When the honest split is 3 / 4 / 3, accept the small gap rather than promoting someone to A to make the numbers work. A mislabelled player does more damage to balance than an odd tier.
The tags print next to names in the result. If you’d rather not put “C” next to a friend on a shared screen, use neutral tags like [red], [blue], [green]; the tool only cares which people share a tag. The list saves in your browser, so you tag everyone once and only edit it when the group changes.
Map one: teams, map, roulette
Teams. Set Number of teams to 2 and click Build balanced teams. If anyone in the call suspects the host of arranging it, type a phrase into Seed before generating. The same seed with the same list in the same order gives the same teams, so anyone with the list can check it. The Share link button copies a URL with the list and the options in it, which is the quickest way to hand them that check.
Map. In the map picker, choose CS2 and leave the pool on Active / Competitive, which is the seven maps the tool lists as active. Switch to All and six reserve maps join them for thirteen. Pick a map draws one, uniformly.
Rules. In strat roulette, choose CS2, and tick No repeats before the first draw. Each team’s captain calls for one challenge for that map, and the host clicks Get a challenge twice, once per side, on screen.
Read each challenge for its scope before play starts. Most of the 30 CS2 entries apply to the whole match (“Only use SMGs for the whole match”), but a few are written for one half (“Deagle only for the entire half”) or for specific rounds (“Play both pistol rounds with your knife only”). Agree the house rule now: a half-scoped challenge either covers the first half only, or that team draws a fresh one at the switch. Either is fine. Changing your mind at 6-6 is not.
Maps two and three: the parts that need a decision
Don’t play the same map twice. The map picker draws independently every time, with no memory of earlier picks. Over three maps from a pool of seven, the chance that at least one map comes up twice is 39%. For a best-of-five from the same pool it rises to 85%. Two easy fixes:
- When a repeat comes up, click again. That’s what the re-roll is for.
- Or, before the night starts, paste the seven map names into the Random List Shuffler and play its order top to bottom. A shuffle can’t repeat an entry, so the whole night is settled in one click and nobody has to judge whether a re-roll was fair.
Rebuild the teams, or don’t. Clicking Build balanced teams again between maps keeps the balance and changes who’s with whom. With a tier of four split two and two, two particular players from that tier land on the same side one time in three, so the chance that the same pair is stuck together for all three maps is 1 in 27, under 4%. If the group prefers fixed teams for a series, keep the first result, and use the seed so the teams can be reproduced if someone’s game crashes and the lobby has to be rebuilt.
Let No repeats do its work across nights. Two challenges a map for three maps is six draws a night. With No repeats on, the tool removes each challenge from the pool once drawn, and only resets when all 30 have been used. Six a night divides 30 exactly, so five nights go by with no challenge ever repeated, and the reset falls cleanly between night five and night six. Leave it off and any six draws from 30 contain a repeat 41% of the time, which is how a group ends up playing “no crouching” three weeks running.
That memory lives in the browser that made the draws, separately for each game. This is why one host should run the tools. If a different person runs roulette next week from their own browser, their pool is full again and the five-night guarantee is gone.
When eleven people turn up
An eleventh player breaks 5v5, and someone has to sit each map. Who sits is a fairness problem of its own, and it’s the one the Fair Rotation / Substitution Tool was built for. Paste all eleven, set On-field per period to 10 and Periods / quarters to 3, and click Build rotation. Thirty seats across eleven players means eight play all three maps and three sit out one map each. The tool always fills each map with the people who have played least so far, so nobody can sit twice while someone else hasn’t sat at all.
Then give that map’s ten to the balancer. Keep an eye on the tiers. If the person sitting out is from a tier of four, that tier is now three for this map, and you’re back to the two-point gap in the table above. That’s acceptable for one map. If the same strong player keeps landing on the same side because of it, rebuild and move on.
The host’s checklist
- Before the night: tag all players, aim for even tier sizes, and paste the seven active maps into the list shuffler for a no-repeat order.
- Each map: build balanced teams (seed it if anyone asks), take the next map from the shuffled list, and draw two roulette challenges with No repeats on.
- Read each challenge’s scope aloud and apply the house rule for half-scoped ones.
- With eleven players, build a three-map rotation first, then balance each map’s ten.
- Same host, same browser, every week.
At the end of the night, export the teams as CSV from the balancer and copy the roulette results out of Recent results, which keeps the last ten draws per tool. If the group keeps a running scoreboard of who won under which handicap, that’s the raw material, and it settles the inevitable argument next week about whether “never buy armor” was really that bad.
None of the three tools knows about the other two, so the night works because one person runs the sequence the same way each time. The single-tool posts go further on each piece: balanced teams, map picks and why roulette suits casual lobbies. Put together, they take setup from twenty minutes to about two, which is twenty minutes more of actually playing.