How to Choose the Right K-Factor for Your League

All guides

6 min read

The K-factor is the single most important setting in any ELO league, and also the most misunderstood. It controls how much one game can change a rating. Set it too high and your standings whip around after every match; set it too low and they take forever to reflect reality. This guide gives you concrete numbers to use.

What K actually does

In the rating formula, your new rating is your old rating plus K × (result − expected), so K is the most points a single game can move you. The actual move is a fraction of K set by how surprising the result was: an even matchup (expected ≈ 0.5) trades about half of K, while a near-total upset (expected ≈ 0) swings almost the full K. With the default K = 32, the biggest realistic jump is around 30 points; with K = 16 it’s half that. The starting rating — 1000 by default — doesn’t affect any of this; it only sets where everyone begins. Higher K means a more reactive, jumpier ladder; lower K means a more stable, slow-moving one.

The core trade-off: responsiveness vs. stability

  • High K (40+): ratings react fast. A hot streak shoots you up quickly. Great when you have few games to work with, or when you want the ladder to feel lively. The downside is volatility — one bad night can erase weeks of progress.
  • Low K (16–24): ratings are stable and hard to move. They reflect a large body of results rather than recent form. Great for established, long-running leagues. The downside is sluggishness — a genuinely improved player can stay underrated for a long time.
  • K = 32: the classic middle ground, and a sensible default for almost everyone.
← steadierjumpier →K=16StableK=32BalancedK=48Reactive
Lower K-factors keep ratings stable; higher ones make them swing faster.

Match K to how often you play

The right K depends heavily on your game volume. Think about how many matches a typical player logs in a season:

  • Few games (a weekend tournament, ~5–15 games each): use a higher K (32–40) so ratings can actually move enough to separate players before the event ends.
  • Moderate (an office ladder, a game or two a week): K = 32 is ideal.
  • High volume (hundreds of games, online clubs): drop to K = 16–24. With so many results, each game should contribute only a little, and you want stability.

How chess federations do it

FIDE uses a tiered K-factor that’s a useful template even for casual groups. New players get K = 40 until they’ve played 30 games. Most players sit at K = 20. Elite players (rated 2400+) use K = 10 so a single result can’t yank an established rating around. The lesson: K should fall as a player accumulates games and as the stakes rise.

A practical recipe

  1. Start everyone at the default rating (1000 in TrackMyElo, or 1500 if you prefer the chess convention).
  2. Use a higher K (around 40) for each player’s first ~10 games so new ratings settle quickly.
  3. Drop to K = 32 for ongoing play in a casual league.
  4. If you’re a high-volume, competitive group, lower it further to 16–24 once ratings have stabilized.

In TrackMyElo you set a single K-factor per league, so if you can only pick one number, 32 is the safe choice for most groups.

Symptoms of a wrong K

  • Standings feel chaotic, the leader changes every week → your K is too high. Lower it.
  • A clearly improved player stays stuck near the bottom for months → your K is too low. Raise it.
  • People stop caring because nothing ever changes → too low; bump it up to make wins feel meaningful.

You can always adjust K mid-season — existing ratings stay where they are, and only future games respond to the new value. If you’re still fuzzy on the underlying math, the ELO rating system explained guide walks through it with worked examples.