Elo vs Simple Ranking Systems

If you are tracking results for a group of friends, you have three realistic options. Here is what each gets right and where each falls over.

Option 1: counting wins

Count wins, maybe subtract losses, write it on a whiteboard. It is the default because it needs no explaining.

It fails in one specific way: every win counts the same. Beating the best player in the group is worth exactly as much as beating someone on their first night. Over a season that turns the leaderboard into a record of who showed up most, because the person who plays twice as often banks twice as many wins whatever the standard of them.

Fine if everyone plays roughly equally often and nobody takes it seriously. It stops being fine the moment attendance is uneven, which in a casual group it always is.

Option 2: points per placement

Common in trivia and club ladders — three for a win, two for second, one for turning up, totalled over a season.

This is genuinely better than a win tally for one reason: it handles more than two sides, which a win/loss record cannot. But it still ignores who you were up against, and it rewards attendance by design. That is sometimes exactly the point — a season-long competition often should reward the people who commit to it.

Where it misleads is when people read the top of the table as "best player" rather than "most points". Those are different questions, and a points ladder only answers the second.

Option 3: Elo

Elo compares the two ratings before a match, works out what each side was expected to score, and moves both ratings by the gap between expectation and result. Beating someone well above you moves your rating a long way; beating someone well below you barely moves it. The full explanation is here.

The trade-offs are real:

• It is not a running total, so there is nothing to accumulate by turning up. Some groups find that demotivating — the reward for playing is that the rating is correct, not that it goes up.
• A new player's number is unreliable until they have played enough matches for it to settle.
• You cannot do it in your head. A whiteboard tally is transparent to everyone in the room; Elo needs something to do the arithmetic.
• Classic Elo is built for two sides. Multi-player and team results need a deliberate extension rather than an obvious one.

What we do with it

Custom Elo Games uses the standard Elo formula, then handles the parts a group of friends actually runs into.

• Up to ten sides in one result. The result is broken into every head-to-head pair inside it, each pair scored by the normal formula, with the K-factor divided so that a six-way exposes you to the same total movement as a one-on-one.
• Teams. A team competes at the average rating of its players, and every player on it takes the full rating change — so a rating always means strength per person.
• Ties anywhere in the finishing order, including an all-square draw, because trivia and party games produce them constantly.
• One leaderboard everyone can see, rather than a number one person maintains. A result takes seconds to log and the standings update as soon as it is in.

So which one

• Nobody is keeping score seriously. Count wins. Honestly — do not install anything.
• A fixed season with a champion at the end, where turning up should count for something. Points per placement.
• You want to know who is genuinely best, with people playing at wildly different frequencies. Elo.

The third case is the one we built for. If you are in the first, a whiteboard is a better tool than we are.

You can see exactly how a single result would move two ratings on the Elo calculator, without an account.