How the bot decides
Everything the bot does on a game night, in plain words. No admin, host or player can override any of it.
Your rating
Everyone starts at 1200. Win and your Rating goes up, lose and it goes down: a win never lowers it and a loss never raises it.
How much depends on the odds, and they are the same win chance the bot shows on the teams, from each team's total Rating. Beating the favourite pays more than beating a team you were expected to beat, and losing as the favourite costs more.
Once you have played 10 games, an even game moves you about 8 points, and one game never moves you more than 20.
Your first 10 games count extra, up to twice as much, so a new player finds their level quickly. Until then the board shows them as still settling instead of giving them a rank.
Everyone on a team gets the same amount from a game. Then the five are lined up by how they played, read for their role: on the winning team the best game counts ×1.2, down to ×0.8 for the fifth; on the losing team it flips, so the best game gives back least. The best on each side is the MVP and the ACE.
Points come from the other team. Once everyone in a game has played 10 games, what one side wins the other side loses, so 1200 stays the average.
Only your own games move your Rating. Nothing fades: take a month off and it is right where you left it.
The week has its own board. Every Sunday at 06:00 Cairo time everyone starts the week at zero points, and only that week's games count. Teams are always made from the all-time Rating.
Only Summoner's Rift games count. ARAM is tracked but never moves a rating. Nobody can edit a Rating by hand, and tapping any change on the site shows the sum behind it.
Picking the teams
When an admin taps Roll teams, the bot tries all 126 ways to split ten people into two teams of five.
For each one it puts everyone in their best lane and scores it: the rating gap between the teams, plus a cost for every player off their main role (bigger if they were filled last game), plus a nudge against repeating last game's teams. Lowest score wins.
It keeps its top three. The first is posted. A reroll moves to the second, then the third. There is no fourth, and nothing is random.
Reading a split
Every split comes with a receipt like this one. Here is what each part means.
Win chance
Blue 54 percent, Red 46 percent.
The center line marks 50–50
Close. Blue has a slight edge.
Next best: swap the top players, Hana and Omar. That's Blue 57%, with a bigger rating gap (170 vs 100 pts).
How the bot decided
The bot tried all 126 ways to split these ten into two teams of five. For each one it put everyone in their best lane and scored it: the rating gap between the teams, plus a cost for every player off their main role (bigger if they were filled last game), plus a nudge against repeating last game's teams. Lowest score wins. Here are its top three:
- #1In playBlue 54% · 46% Red
- Rating gap
- 100 pts
- Main roles
- 10/10
These teams.
- #2Blue 57% · 43% Red
- Rating gap
- 170 pts
- Main roles
- 10/10
Swap the top players, Hana and Omar.
Ranked lower: a bigger rating gap (170 vs 100 pts).
- #3Blue 48% · 52% Red
- Rating gap
- 220 pts
- Main roles
- 10/10
Swap the mid players, Karim and Nadia.
Closer odds, but ranked lower: a bigger rating gap (220 vs 100 pts).
Why win chance and rating gap can disagreeWin chance comes from everyone's Rating as it is. The rating gap is what the bot balances on, and it counts anyone off their main role as a bit weaker there, so the two can point slightly different ways.
Nobody picked these teams.Admins can set the mode and whether a game is rated, but only before its teams are rolled. Then they can tap Roll teams and Reroll (which moves to the next pick on this list), and nothing else. Nobody can hand-edit a rating; ratings only move when a game ends.
The bot's note: Blue favored 54%. Everyone on a main role. Gap 100. Next best: swap Hana and Omar, gap 170.
More on how it works- Win chance
- The headline number: how likely each side is to win, from each team's total Rating. It is the same win chance the rating uses when the game ends.
- Rating gap
- How far apart the two teams are, in rating points. This is what the bot balances on, counting anyone off their main role as a bit weaker there.
- Main roles
- How many of the ten are on the role they main.
- Bot's pick
- Which of the three splits is in play. A reroll shows here, so it is never hidden.
- How the bot decided
- Open it to see all three splits side by side, why the lower ones ranked lower, and the bot's own note.
Are the odds honest?
Odds are only worth something if they come true about as often as they say. So every group checks the bot against itself, under every split's How the bot decided: of the games where one side was favored, how often did that side win, and how often did the bot expect it to?
If the bot says 55% every game, the favorite should win about 55% of them, not all of them. Games where the teams changed in the lobby after the roll are left out, because those were not the bot's teams.
What admins can't do
- Pick the teams, or move anyone between them.
- Set, add to or edit anyone's rating.
- Change who won. The result comes from the end-of-game screen.
What they can do: set the mode and whether a game is rated, before its teams are rolled; tap Roll teams; and Reroll to the bot's next pick. The group's owner can also reset everyone's ratings at once, for a fresh start, but never one person's.