How US Chess Ratings Are Calculated
A US Chess rating is updated after every rated event by comparing the points you scored with the points your rating said you should score. That much is ordinary Elo. What makes the US Chess system different is how far a rating moves for a given surprise: it depends on how established the rating is, how long the event was, and whether you beat expectation by enough to earn a bonus. This guide walks through each piece with real numbers.
Step 1: the expected score
Against each opponent, your rating gives an expected score between 0 and 1:
Two equal players expect half a point each. The bigger the gap, the more the favourite expects:
| You are rated higher by | Your expected score |
|---|---|
| 0 points | 0.50 |
| 100 points | 0.64 |
| 200 points | 0.76 |
| 400 points | 0.91 |
Add those up over every game in the event and you have the score your rating predicted.
Step 2: how far the rating moves (the K-factor)
The difference between your score and that prediction is multiplied by a K-factor:
Here m is the number of games you played in the event, and N′ is the "effective number of games" behind your rating — how settled it is. N′ is the number of rated games you have played, but capped at a value that depends on your rating:
| Rating | Cap on N′ | K in a 4-round event |
|---|---|---|
| 400 | 8.4 | 64.5 |
| 800 | 10.3 | 56.1 |
| 1200 | 13.1 | 46.7 |
| 1600 | 18.1 | 36.1 |
| 2000 | 28.6 | 24.5 |
| 2355 and above | 50.0 | 14.8 |
The K values in the US Chess rating document's own table — for example K = 80 with N′ = 6 over 4 games — are reproduced exactly by the calculation this site uses.
A worked example
A player rated 1500 with 40 rated games plays a four-round event against opponents rated 1400, 1550, 1600, 1700. Their expected scores are 0.64 + 0.43 + 0.36 + 0.24, which is 1.67 in total. At 1500 the cap on N′ is 16.57, below their 40 games, so K = 800 / (16.57 + 4) = 38.89.
They score 2 points. That is 0.331 above expectation, so the rating rises by 38.89 × 0.331 = +12.9, to about 1513.
Step 3: the bonus for a big overperformance
When you beat expectation by enough, US Chess adds a bonus on top:
In a four-round event the threshold is 14 × √4 = 28 points of change. The bonus applies only in events of three or more games where no opponent was played more than twice. (Figures below are rounded to one decimal place.)
- Score 2.5 in the example above: the plain change is 32.3, which clears 28 by 4.3, so the bonus is 4.3 and the total change is +36.7.
- Score 4 out of 4: the plain change is 90.7, the bonus 62.7, and the total +153.3. This is how a fast-improving junior catches up with their real strength in one event rather than ten.
The bonus only ever adds points. Falling short of expectation costs exactly K × (score − expected) and nothing more. The Tournament Predictor guide shows what that one-sidedness does to a whole field.
New players: the special formula
A player with 8 or fewer rated games — or whose every previous game was a win, or every one a loss — is rated by a different method. Instead of adjusting the old rating by K, it finds the rating at which the player's results would have been exactly expected, counting their previous games as extra evidence.
It replaced an older formula that could punish a win. Under the old one, a player rated 1500 over 5 games who beat a 500-rated opponent dropped to 1400. The special formula leaves them at 1500. This site checks the guarantee more broadly on every build: across a spread of ratings, histories and opponents, turning a loss into a win never lowers a special rating.
Rating floors
No US Chess rating can fall below 100, and each player has a personal floor that rises slightly with games won and drawn and events played, up to 150. For example, 3 wins, 1 draw and 10 events give a floor of 124. Established players also earn a floor 200 points below their peak, rounded down to the hundred: a peak of 1941 gives a floor of 1700. Earned floors run from 1200 to 2100.
Dual-rated events above 2200
In events rated for both Regular and Quick (where the minutes plus the increment in seconds come to between 30 and 65 — G/45 with a 5-second delay is 50), players above 2200 have their Regular K reduced. At 2300 it is 75% of normal, and from 2500 it is a quarter, so faster games move a strong player's slow-game rating only a little.
Why no estimate matches the official number exactly
US Chess rates every player in an event twice. The first pass uses everyone's pre-event ratings. The second repeats the calculation using each opponent's first-pass rating, so your official result depends on how your opponents did in games you were not in. Any calculator working from your games alone — this one included — computes the first pass. It is usually within a few points, and further out when an opponent had an unusual event.
To try your own event, open the USCF Rating Estimator. The formulas here come from "The US Chess Rating System" (Glickman & Doan), as revised in 2020 and 2024.