Elo Ratings Explained: From Chess to the NBA
Arpad Elo designed his rating system for chess in the 1960s. Today variants power rankings in soccer, American football, basketball, esports and more, because the core idea is elegant: every result moves ratings by an amount proportional to how surprising it was.
The win-probability formula
For teams A and B with ratings RA and RB:
P(A wins) = 1 ÷ (1 + 10(RB − RA) / 400)
A 100-point gap gives the stronger side about 64%; 200 points ≈ 76%; 400 points ≈ 91%.
Updating after a game
New rating = old rating + K × (actual result − expected result), where the result is 1 for a win, 0.5 for a draw and 0 for a loss. K controls how fast ratings react: high K for short seasons (NFL), lower K for long ones (NBA, MLB).
Sports-specific tweaks
- Home advantage: add a fixed bonus to the home side's rating before computing probability (often 50–100 points depending on sport).
- Margin of victory: scale K by a function of the score margin so blowouts count more, with an autocorrelation adjustment so favourites are not over-rewarded.
- Season reversion: pull ratings partway toward the mean between seasons to reflect roster turnover.
Try it with our Elo probability calculator, including a home-advantage input.
Strengths and limits
Elo needs only results, so it works in any sport with a schedule. It self-corrects and is easy to explain. Its limits: it does not know about injuries or trades until results arrive, and it treats every game as equally informative unless you adjust K. Good analysts combine Elo with player-level data for a fuller picture.
Keep learning
Next: Home Advantage: How Much Is It Worth in Each Sport?
Get one 3-minute lesson every Sunday in the free Edge Report.
Get the Edge Report