How Paddle Rack's Rotation Algorithm Works

Paddle Rack's rotation algorithm chooses who plays next by a fixed priority order: first put whoever has played the fewest games on next (in testing, 98% of everyone-off open-play sessions end with every player within one game, and none more than two apart), while never leaving a court empty, then minimize repeat partners, then minimize repeat opponents, then optionally balance skill, and it produces the same schedule every time for the same inputs.

Updated October 2, 2026 · 6 min read

Every open play organizer has heard it: "I've sat out three times and he hasn't sat once." Usually nobody did anything wrong. A paddle rack tracks order, not totals, and human memory is worse than either. This page explains, in plain words, how Paddle Rack decides who plays next, so you can see why the results are fair, and check them yourself.

Definitions

• Bye: a round in which a player is scheduled not to play.
• Sit-out: the same thing in open play; a player waiting off court while others play.
• Mixer: a doubles format where partners rotate every game and results are tracked per player.
• Repeat partner: playing on the same team as someone you have already partnered in the session.
• Repeat opponent: playing against someone you have already faced in the session.
• Games-played spread: the difference between the most and fewest games played by any present player. A fair rotation keeps it at 0 or 1.
• Seed: a number used to break ties consistently, so the same inputs always produce the same schedule.

The fairness rules, in priority order

When a court opens (in open play) or a new round starts (in a round robin), the rotation picks the next four players and splits them into two teams. It follows these rules in strict order. A lower rule only decides between options that are equally good on every higher rule.

1. Fewest games goes next, and no court sits empty

Among everyone currently present, nobody should get two games ahead of anybody else. So the players who have played the fewest games come on first, and the players who have played the most sit first. When several players are tied, the one who has waited longest since their last game goes first.

This single rule is what makes sit-outs even. In a round robin, games played (and sit-outs) differ by at most one, always. In open play, courts finish at different times and a court never waits for the "right" four players, so the spread can briefly reach two or three while games are running; the next rotation corrects it. In 378 simulated sessions (4–24 players, 1–6 courts, every rotation rule), no court ever sat empty while four players were free; 98% of everyone-off sessions ended with every player within one game, and none ended more than two apart. If you need "within one" at every moment, turn on strict fairness: courts may then wait briefly for the right players.

2. Minimize repeat partners

Among the choices that satisfy rule 1, pick the teams that create the fewest repeat partnerships. If Ana has already partnered Ben, Ana and Ben go on opposite sides or on different courts when possible.

3. Minimize repeat opponents

Among choices that tie on rules 1 and 2, pick the matchups with the fewest repeat opponents, so players face as many different people as possible.

4. Optional: balance skill or keep bands

If the organizer turns on skill balancing, teams are paired so the combined ratings on each side of the net are close (a 4.0 with a 3.0 against two 3.5s, for example). If the organizer uses skill bands, each band rotates on its own courts. Skill balancing never overrides games-played fairness. See skill-level open play.

How "everyone off" and "winners stay" fit in

The rules above describe the default, where all four players come off after a game. If the organizer chooses winners stay (capped at two consecutive games) or a challenge court, the staying players are kept and the rotation fills the open spots using the same rules. The cap matters: it is what stops winners-stay from breaking rule 1. Paddle stacking rules covers these variants.

Late arrivals and early leavers

Late arrivals. A player who arrives in round 5 has played zero games. If the rotation treated zero literally, they would play every game for the next half hour while others sat. If they went to the back of a queue, they might wait two rounds. Neither is fair. Instead, a late arrival joins with a fair starting count: credited as if they had played about as many games as the players who have played the least so far. From there, rule 1 treats them like everyone else.

Early leavers. When a player checks out, they are simply removed from the pool of present players. Their games stay in the results, but they no longer count toward the "within one" spread, and the rotation never waits for them.

Manual overrides. The organizer can always swap a player in or out (someone needs a break, a pair needs splitting). The rotation adapts from the new state.

Same input, same schedule

The rotation is deterministic: given the same players, the same settings and the same seed, it produces exactly the same schedule every time. Ties are broken by the seed, not by chance at the moment you tap a button. That means:

• A printed schedule can be regenerated and checked by anyone.
• The free web generator uses the same engine as the app, so the schedules match.
• Disputes can be settled by looking at the record, not by memory.

The fairness report

At the end of a session, Paddle Rack shows a fairness report: the range of games played, sit-outs per player, and counts of repeat partners and repeat opponents. A typical one reads: "Everyone played 6 or 7 games, 0 repeat partners." Organizers share it with the group as proof that the rotation was even. Every round robin schedule page on this site includes the same summary.

The math limits (what no rotation can do)

Some unfairness is built into the numbers, and an honest rotation tells you so.

• Someone must sit. With 5 players on 1 court, one person sits every game. Each player sits one game in five.
• Repeat partners become unavoidable. Each player has N minus 1 possible partners. With 6 players, that is 5. Once a player has played more than 5 games, at least one partner must repeat. The rotation can only keep repeats as few and as spread out as possible.
• Uneven totals. If playing slots do not divide evenly among players, some play one more game than others. That is the "within one" in rule 1 for round robins.

A worked example: 6 players, 1 court

Players A to F. Each game, 4 play and 2 sit. Here are the first six games:

Check the fairness:

• Games played: after game 3, everyone has played exactly 2. After game 6, everyone has played exactly 4. The spread is never more than one after any game.
• Sit-outs: after game 6, everyone has sat exactly twice, and nobody sat twice in a row.
• Repeat partners: zero. Each player has had 4 different partners out of 5 possible (A partnered B, E, C and D; F partnered B, D, C and E).
• Looking ahead: each player has only one unused partner left. By their sixth game, repeats begin. That is the math limit, not a flaw.

You can generate this exact case, or any other, on the 6 players, 1 court page or in the generator.

Why publish this?

Because fairness you cannot check is just a promise. Organizers deserve to know how the next four are chosen, and players deserve to see that nobody is being favored. If you spot a case where the rotation seems off, the deterministic schedule makes it easy to reproduce and report.

For running the session itself, see how to run pickleball open play and the round robin formats guide.

Keep reading