How to Use Elo Ratings for Casual Games

To use Elo ratings for casual games, give every player the same starting rating (for example 1000 points). After each game, compare each player's expected score with the actual result and adjust both ratings by a fixed amount called the K-factor (32 suits most groups). Then use the current ratings to seed brackets or pair opponents so that games stay close. The system only needs a spreadsheet, one formula and a clear routine for recording results. That routine is what turns a loose group of friends into an organized event.

What is the Elo rating system for casual games?

The Elo rating system for casual games is a points-based method that estimates how strong each player is relative to the rest of the group, based only on wins, losses and draws against each other. It was designed for chess, but it works for any game with a clear winner between two sides: table tennis, video game duels, card games, darts or two-player board games.

Elo has three useful properties for a friend group:

• It rewards upsets. Beating a stronger player earns more points than beating a weaker one.
• It is self-correcting. A wrong starting guess fades as more games are played.
• It is relative. A rating of 1150 has no absolute meaning. It only says something when compared with the other ratings in the same group.

How does the Elo formula work?

The Elo formula works in two steps. First, it calculates each player's expected score from the rating difference. Second, it moves each rating toward the actual result by an amount set by the K-factor.

Expected score

The expected score is the result a player should average against a given opponent, on a scale where a win counts as 1, a draw as 0.5 and a loss as 0. For player A against player B:

Expected score of A = 1 / (1 + 10^{(Rating B − Rating A) / 400})

The two players' expected scores always add up to 1. Only the rating difference matters, not the size of the numbers. The table below shows what common rating differences mean.

Rating difference (points)	Expected score of the higher-rated player	Expected score of the lower-rated player
0	0.50	0.50
100	0.64	0.36
200	0.76	0.24
300	0.85	0.15
400	0.91	0.09

Rating updates and the K-factor

A rating update is calculated as New rating = Old rating + K × (Actual score − Expected score). K is the K-factor, the maximum number of points a single game can move a rating. A higher K reacts faster to recent results. A lower K gives steadier ratings that need more games to change.

When both players use the same K-factor, the points one player gains equal the points the other loses, so the group's total stays constant.

How draws are handled

A draw counts as an actual score of 0.5 for both players. Between equally rated players, a draw changes nothing. If the players have different ratings, a draw moves a few points from the higher-rated player to the lower-rated one, because the higher-rated player was expected to do better than half. If your game cannot end in a draw, simply use 1 and 0.

Worked example with K = 32

Ana and Ben both start at 1000. Each has an expected score of 0.50. Ana wins: she gains 32 × (1 − 0.50) = 16 points and moves to 1016. Ben drops to 984.

Later, Ana (1100) plays Cara (1000). Ana's expected score is 0.64 and Cara's is 0.36.

• Ana wins: Ana gains 32 × 0.36 ≈ 12 points (1112). Cara loses about 12 (988).
• Cara wins: Cara gains 32 × 0.64 ≈ 20 points (1020). Ana loses about 20 (1080).
• Draw: Ana loses about 4 points (1096). Cara gains about 4 (1004).

Keep the decimals in your spreadsheet and round only for display. Rounding every game can slowly pull the totals out of balance.

How do you set up Elo ratings for a friend group?

Setting up Elo ratings for a friend group means agreeing on starting ratings, a K-factor and the moment ratings are updated, before the first game is played. Follow these steps:

1. Choose one starting rating for everyone. 1000 and 1500 are common choices. The number is arbitrary because only rating differences matter. Equal starting ratings also avoid arguments about who is better.
2. Pick a K-factor. 32 is a practical default for small groups that play a few dozen games a year. Use a higher value, such as 40, during the first few games of a brand-new player so their rating settles quickly. Use a lower value, such as 16 to 24, if your group plays very often and wants steadier rankings.
3. Decide when ratings update. Updating after every game is simple and gives live feedback. Updating once at the end of an event, using everyone's pre-event ratings for all calculations, means the order of games does not affect the result. Pick one method and keep it.
4. Build a results sheet. Use one row per game with these columns: date, player A, player B, result for A (1, 0.5 or 0), rating A before, rating B before, expected score for A, rating A after, rating B after.
5. Name one scorekeeper. One person entering results prevents double entries and missing games.
6. Agree what counts as rated. Decide in advance whether practice games, games with house rules or unfinished games count. Byes and forfeits should usually not change ratings, because no game was played.

For team games, a simple approach is to use each team's average rating as its rating and apply the same change to every team member. For games with more than two players at a table, you can treat the final placings as a series of head-to-head results between every pair. Divide K by the number of opponents so that one game does not swing ratings too much.

Which event format works best with Elo ratings?

The best format depends on how many players you have and how much time and space are available. Elo ratings help in every format, either by seeding the bracket or by pairing similar players.

Format	How Elo is used	Games needed	Best for	Main drawback
Single elimination	Seeds the bracket so top players meet late	Players − 1 (8 players = 7 games)	Short events, a clear final	Half the players are out after one game
Round robin	Updates ratings after every game; standings by points	n × (n − 1) / 2 (6 players = 15 games)	4 to 6 players, a full afternoon	The number of games grows quickly
Swiss	Pairs players with equal scores by rating	Players / 2 per round, usually 3 to 5 rounds	8 to 30 players, everyone plays every round	Pairing takes a few minutes between rounds

How do you build an Elo-seeded bracket?

An Elo-seeded bracket ranks players by current rating and places them so the strongest players can only meet in later rounds. This keeps the final competitive and avoids early mismatches between favorites.

1. Sort players by rating. The highest rating is seed 1. Break ties with a coin flip or with the number of games played.
2. Round the bracket up to a power of two. Use 4, 8, 16 or 32 slots.
3. Give byes to the top seeds. The number of byes is the bracket size minus the number of players.
4. Pair highest against lowest. In an 8-slot bracket, the first-round games are 1 v 8 and 4 v 5 in the top half, and 2 v 7 and 3 v 6 in the bottom half.
5. Write the bracket where everyone can see it before the first game starts.

Example: 8 players

With 8 players there are no byes. Round 1 has 4 games, the semifinals have 2 and the final has 1, for a total of 7 games over 3 rounds. If everyone plays to their seed, seeds 1 and 2 meet in the final.

Example: 6 players

With 6 players, use an 8-slot bracket with 2 byes. Seeds 1 and 2 skip round 1. Round 1 is 3 v 6 and 4 v 5. In the semifinals, seed 1 plays the winner of 4 v 5 and seed 2 plays the winner of 3 v 6. The event still needs 5 games in total (players − 1).

To keep eliminated players involved, invite them to play rated casual games on free tables while the bracket continues. Their results still feed into the Elo ratings.

How do you run a Swiss event with Elo pairings?

A Swiss event pairs players with the same tournament score in every round, using Elo ratings to decide the order. Nobody is eliminated, and players are quickly matched with opponents of similar strength.

1. Set the number of rounds. A common guide is enough rounds to separate one clear winner: 3 rounds for up to 8 players, 4 for up to 16 and 5 for up to 32.
2. Pair round 1 by rating. Split the rating list into a top half and a bottom half and pair them in order. With 8 players, that means 1 v 5, 2 v 6, 3 v 7 and 4 v 8.
3. Score the event separately from Elo. Use 1 tournament point for a win, 0.5 for a draw and 0 for a loss.
4. Pair later rounds by score, then rating. Group players with the same tournament score, sort each group by rating and pair the top half against the bottom half. If a group has an odd number of players, move the lowest-rated player down to the next group.
5. Avoid rematches. If a pairing repeats an earlier game, swap with the nearest player in the same group.
6. Agree on tiebreaks in advance. Simple options are head-to-head result, then the total tournament score of each player's opponents.

How do you plan a realistic schedule?

A realistic schedule is built from the length of one round slot multiplied by the number of slots, plus fixed time for check-in, breaks and the closing. The common mistake is to plan only for game time and forget changeovers.

1. Estimate one game. Use the typical length of a full game in your group, not the fastest one.
2. Add changeover time. Allow time for reporting the result, resetting the game and finding the next table. Five to ten minutes is a reasonable starting allowance.
3. Count slots per round. Divide the number of games in a round by the number of tables or setups and round up.
4. Add fixed blocks. Include check-in before round 1, one break roughly every two rounds and time for the final standings.
5. Set a firm round time. Agree in advance what happens if a game runs over, for example finishing the current turn or point and then counting it as a draw.

Example: 8 players in a 3-round Swiss

Each round has 4 games. Suppose a game takes 30 minutes and changeover takes 10, so one slot is 40 minutes.

• With 4 tables: each round is one slot of 40 minutes. Three rounds take 120 minutes. Add 15 minutes of check-in, a 15-minute break after round 2 and 10 minutes for results, and the event takes about 2 hours 40 minutes.
• With 2 tables: each round needs 2 slots, or 80 minutes. Three rounds take 240 minutes, and the whole event runs past 4.5 hours. In that case, shorten the games or cut to 2 rounds plus a final.

Example: 6 players in a round robin

A round robin for 6 players has 15 games, played over 5 rounds of 3 games. With 3 tables and 40-minute slots, the games take 200 minutes. With check-in and two breaks, plan for about 4 hours.

Where to put breaks

Schedule breaks between rounds, never in the middle of one, so that all ratings and standings are up to date when play resumes. The organizer can use the break to enter results and prepare the next pairings.

How do you make the next round clear to everyone?

The next round is clear when every player can see, without asking, who they play, where they play and when the round starts. A visible board and one announcement per round achieve this.

1. Number the tables or setups. Use table 1, table 2 and so on, and put the top pairing on table 1.
2. Post one pairing sheet per round. List table number, player A, player B and a blank result column. Use a whiteboard, a printed sheet or a shared screen.
3. Post current standings next to the pairings. Show tournament points and current Elo ratings so players understand why they were paired.
4. Announce the start time. Say it aloud: Round 2 starts at 14:20; check the board for your table.
5. Collect results in one place. Ask the winner, or both players after a draw, to write the result on the sheet immediately after the game.
6. Do not post the next round until every result is in. Pairings and seeding depend on complete results.

How do you handle changes on the day?

Changes on the day, such as late arrivals, dropouts and odd numbers of players, are easiest to handle when the rules are agreed before round 1 and announced at check-in.

• Odd number of players: one player gets a bye each round. In a Swiss event, give it to the lowest-ranked player who has not had one yet. A bye usually counts as a win in the tournament standings but does not change the Elo rating.
• Late arrival before round 1: add the player to the rating list and re-seed. If the bracket is already posted, give them the lowest seed or, in a Swiss event, let them join from round 2 with 0 points.
• Dropout during a Swiss event: remove the player from future pairings. Their completed games stay rated.
• Dropout during a bracket: their next opponent advances by forfeit. A forfeit is not rated.
• Running behind schedule: cut one Swiss round or shorten the game format for later rounds, and announce the change at the next break.
• A new player joins the group long-term: give them the standard starting rating and a temporarily higher K-factor so they reach their true level faster.

What mistakes should you avoid with casual Elo ratings?

The most common mistakes with casual Elo ratings come from inconsistency, not from the math.

• Changing the K-factor mid-season: this makes older and newer rating changes impossible to compare.
• Treating early ratings as final: ratings after only a handful of games are rough estimates, so present them as provisional.
• Rating unfinished games or forfeits: this moves ratings without any real result behind them.
• Mixing different games in one rating: keep a separate rating list for each game, because skill in one rarely predicts skill in another.
• Letting several people edit the sheet: this leads to duplicate or missing games and makes later rating updates wrong.

With equal starting ratings, a fixed K-factor, a single results sheet and a posted plan for each round, an Elo-rated game night runs smoothly. Each event also makes the ratings, and the next set of pairings, a little more accurate.