What is Elo?

Elo is a rating system invented by Arpad Elo for chess. Its core idea: a rating is a number such that the difference between two players' ratings predicts the result of a game between them. Equal ratings predict 50/50; a 200-point gap predicts roughly 76/24 for the stronger player; 400 points, roughly 91/9.

After every game, ratings move toward whatever the result implied. Beat a player you were expected to beat and you gain a little; beat one you were expected to lose to and you gain a lot. The size of the update is set by a constant called K. High K makes ratings responsive but noisy; low K makes them stable but slow to notice change.

Elo's strengths are why it spread far beyond chess — it needs nothing but results, it is simple to compute, and it is well calibrated for head-to-head competition. Its main weaknesses: it has no concept of uncertainty (a new player's 1500 and a veteran's 1500 look identical), and it handles inactivity poorly (a rating from two years ago is treated as current).

Those two weaknesses are exactly what Glicko and Glicko-2 were designed to fix, which is why LUDUX uses Glicko-2 rather than plain Elo. The intuition transfers directly: LUDUX ratings still mean "difference predicts result" — they just carry error bars as well.