Win Probability Calculator
Enter two player ratings to see the statistical probability of each outcome - win, draw, or loss. Based on the Elo expected score formula with draw probability estimates derived from historical master-level data.
Professor Archer says: Statistics are interesting, but remember - any single game can go either way. I have seen 1200-rated players beat 1800s when they play with clarity and purpose. The probability just tells you what tends to happen over many games. Your job is to play the best moves on the board in front of you, regardless of the numbers.
Features
- Visual win/draw/loss probability bar
- Based on Elo expected score formula
- Historical draw probability estimates
- Compare any two ratings from 100 to 3000
- See how rating gaps affect outcomes
What Rating Gaps Actually Mean
The Elo system was designed so that specific rating gaps translate to specific expected scores. A 100-point favorite expects to score about 64%. A 200-point favorite expects about 76%. At 400 points the favorite expects roughly 92%, and beyond 600 points an upset becomes a genuine rarity, on the order of a few percent.
Expected score is not the same as win probability, because draws exist. A 76% expected score might be 60% wins, 32% draws, and 8% losses at master level, but 72% wins, 8% draws, and 20% losses between club players, where draws are rarer. This calculator splits the expected score into separate win, draw, and loss estimates so you see the full picture.
How the Draw Estimate Works
Draw frequency depends on two things: how close the players are in strength, and how strong they are in absolute terms. Games between equally matched players produce the most draws, and draws become more common as the level rises: elite grandmaster games are drawn more than half the time, while beginner games almost never are.
This tool models the first effect with a decay formula, P(draw) = 0.5 x e^(-|R1 - R2| / 200), fitted to historical game data. It is an approximation, and your own draw rate will vary with style and time control, but it captures the essential pattern: the closer the ratings, the more likely the half point.
What the Numbers Are Good For (and Not)
Use win probability to set realistic expectations. If you are 150 points below your opponent, you are not supposed to win most of these games, so a loss is information, not a catastrophe, and a win is genuinely earned. Tournament players use these numbers to estimate performance ratings and to decide when a draw offer is mathematically reasonable.
What the numbers cannot do is predict a single game. Preparation, fatigue, time control, and style matchups all move the real odds. Treat the calculator as a long-run average, then go play the position in front of you. If you are curious where your own number comes from, take our chess rating test, and read how chess ratings work for the mechanics behind the math.
Frequently Asked Questions
How accurate are chess win probability predictions?
Win probability estimates based on Elo ratings are statistically accurate over large numbers of games. However, individual games are unpredictable. Factors like preparation, time control, psychological state, and playing style all influence the actual outcome. Use these probabilities as a general guide, not a guarantee.
How is draw probability estimated?
Draw probability is estimated using the formula P_draw = 0.5 * e^(-|R1 - R2| / 200), which models the observation that games between equally-rated players are more likely to end in draws than games with large rating gaps. This approximation is based on statistical analysis of professional chess games.
Does the rating difference tell you everything about a matchup?
No. Rating difference gives you the statistical expectation, but it does not account for playing style matchups, opening preparation, time control preferences, or current form. Some players consistently perform better against certain styles of play, regardless of the rating gap.
What are the odds of beating a player 300 points higher?
A 300-point underdog has an expected score of about 15%, which works out to winning roughly one game in ten with a few draws mixed in. It absolutely happens - upsets are a normal part of chess - but over a long match the favorite pulls ahead reliably.