How to Set Starting Ratings for New Players

To set starting ratings for new players, give every unrated player the same default number, or a number based on a simple skill tier, and mark it as provisional. Let that provisional rating change faster than established ratings for the first few games. Placement matches, which are short early games that only exist to sort players, can make the number more accurate before seeding matters. For a casual group of friends, this approach is enough to seed a fair bracket, keep early rounds competitive and correct mistakes within a few games.

What Is a Starting Rating for a New Player?

A starting rating, also called an initial rating, is the number a rating system gives a player before it has any results for them. It is a best guess. Every rating system has to start somewhere, but a new player's number carries high rating uncertainty: nobody yet knows how strong they really are.

The scale itself is arbitrary. Many Elo-style systems use 1000 or 1500 as the default, and the Glicko systems use 1500 as their default rating. What matters is the gap between players. In the standard Elo formula, a player rated 200 points higher is expected to score about 76% against the lower-rated player, whether the two numbers are 1100 and 900 or 1600 and 1400.

The practical goal is not to guess perfectly. You want a start that puts new players close enough to their true level that the first rounds are not one-sided. You also want a system that fixes the rest quickly.

Which Method Should You Use for Starting Ratings?

For most friend groups, the best method is a single default rating combined with provisional status. Add skill tiers if the group's abilities clearly differ. Add placement matches if you have time before the main event. The table compares the common options.

Method	How it works	Best for	Main drawback
Same default for everyone	All unrated players start at one number, for example 1000	First-ever event where nobody has a rating	Strong and weak newcomers look identical at first
Self-assessment tiers	Players pick beginner, intermediate or experienced, and each tier gets a different start, for example 900, 1000 and 1100	Groups with obvious skill differences	People overrate or underrate themselves
Organizer assessment	Someone who knows everyone assigns starting numbers	Small groups where one person has seen everyone play	Can feel unfair or biased
Placement matches	New players play a few unranked or lightly weighted games before seeding	Events with spare time before the bracket starts	Uses time and tables
Provisional rating	The new player's rating moves faster until they have played a set number of games	Every situation; combine with any method above	Early results can swing the number a lot

These methods are not exclusive. A common, simple combination is: tier-based starting number, then provisional status for the first five games.

How to Set Starting Ratings Step by Step

Setting starting ratings takes five decisions: the scale, the default, the tiers, the provisional rules and the placement plan. Make them before the event and write them down so everyone sees the same rules.

1. Choose a scale. Pick one round number as the centre, such as 1000. Use it for the whole group and do not change it later, or old and new ratings will not compare.
2. Set the default for unrated players. Give every unrated player the centre value unless you use tiers. If your group already has rated regulars, a fair alternative is to start newcomers at the group's median rating, so a newcomer is assumed to be typical.
3. Decide whether to use tiers. If you do, keep it to three tiers and space them modestly, for example 100 points apart. Wide gaps cause big errors when someone picks the wrong tier.
4. Define provisional status. Mark new players' ratings with a P, for example "1000P". Decide how many games remove the mark. Five to ten games is a reasonable choice for a casual group; it is a rule you set, not a fixed standard.
5. Decide on placement matches. If time allows, schedule one to three short games per new player before the bracket. Use the results to adjust their rating before seeding.

How Provisional Ratings Work in Practice

A provisional rating is a starting rating that the system adjusts more aggressively because it is uncertain. In an Elo-style system, you do this by giving provisional players a larger K-factor, the multiplier that controls how much one game changes a rating.

The Elo update is: new rating = old rating + K × (actual score − expected score). The expected score is 1 ÷ (1 + 10^((opponent rating − player rating) ÷ 400)). A win counts as 1, a draw as 0.5 and a loss as 0.

Example: A newcomer rated 1000P uses K = 40. An established player rated 1100 uses K = 20.

• The newcomer's expected score is about 0.36.
• If the newcomer wins, they gain 40 × (1 − 0.36) ≈ 26 points and move to about 1026P.
• The established player loses 20 × 0.64 ≈ 13 points and moves to about 1087.

The newcomer moves twice as fast, so a wrong starting guess is corrected quickly. The established player is protected from big swings caused by an unknown opponent. When the provisional period ends, switch the newcomer to the normal K-factor.

Systems such as Glicko handle this automatically by tracking rating uncertainty as a separate number. In Glicko-1 and Glicko-2 this number is called rating deviation, and new players start with a high value. If your app or spreadsheet supports Glicko, you do not need a separate provisional rule. A manual Elo sheet with a higher K for new players gets most of the same benefit.

How Placement Matches Improve Starting Ratings

Placement matches are games played before or at the very start of an event to estimate a new player's level before seeding. They reduce the chance that a strong newcomer knocks out a top seed in round one, or that a beginner faces the strongest player immediately.

Keep placement simple:

• Pair each new player first against someone near the middle of the group.
• If they win, pair them next against someone stronger. If they lose, pair them next against someone weaker.
• Apply the provisional K-factor to each result.
• Seed using the rating after placement.

Two or three placement games per newcomer is usually enough for a casual event. Use shorter time limits for these games if your activity allows it.

How Do Starting Ratings Shape the Bracket?

Ratings decide seeding: the highest-rated player becomes seed 1, the next becomes seed 2, and so on. A good bracket keeps the top seeds apart until the later rounds. Unrated players are seeded using their starting or post-placement rating, just like everyone else.

Seeding an 8-Player Single-Elimination Bracket

Sort players by rating and pair the strongest with the weakest:

• Seed 1 vs Seed 8
• Seed 4 vs Seed 5
• Seed 2 vs Seed 7
• Seed 3 vs Seed 6

Place the 1 vs 8 and 4 vs 5 matches in the top half of the bracket and the 2 vs 7 and 3 vs 6 matches in the bottom half. If the higher seeds keep winning, seeds 1 and 2 meet only in the final.

If two newcomers both start at 1000P, break the tie with a coin toss or by placing them in different halves of the bracket. Spreading uncertain players apart limits the damage if one of them is much stronger than their number suggests.

Handling Numbers That Are Not a Power of Two

If the number of players is not 4, 8, 16 or 32, give byes to the top seeds. A bye means a player advances to the next round without playing.

Example with 6 players: the next power of two is 8, so 2 byes are needed. Seeds 1 and 2 get byes. Seeds 3 to 6 play the first round: 3 vs 6 and 4 vs 5. The winners then face seeds 1 and 2.

Because byes go to the top, starting ratings matter here too. If a newcomer's rating is a guess, avoid giving them a bye unless placement matches have confirmed their level.

Which Format Suits Uncertain Ratings Best?

When many players are unrated, a format with more games per player works better than a short knockout. More games give the ratings time to correct themselves.

• Single elimination: fastest, but one bad seeding guess can end a player's day after one game.
• Swiss system: everyone plays a fixed number of rounds, and each round pairs players with similar scores. This works well when ratings are uncertain, because mismatches in round one are corrected by round two.
• Round robin: everyone plays everyone. Seeding barely matters, but it takes the most time.

How to Plan a Realistic Schedule

A realistic schedule starts from three numbers: how many matches the format needs, how many tables or courts can run at once, and how long one match takes including changeover. Multiply these out before choosing a format.

Use this formula: total time = number of time slots × (match length + changeover time) + breaks. A time slot is one period in which all available tables are in use at the same time.

Example: 8 players, 2 tables, 20-minute matches and 5 minutes of changeover, so 25 minutes per slot.

Format	Matches needed	Time slots with 2 tables	Playing time
Single elimination	7 (4 + 2 + 1)	4 (2 + 1 + 1)	About 100 minutes
Swiss, 3 rounds	12 (4 per round)	6 (2 per round)	About 150 minutes
Round robin	28	14	About 350 minutes

The round robin needs nearly six hours of play, which is unrealistic for one afternoon. A 3-round Swiss fits in about two and a half hours of play and gives every player at least three games. That is enough to move provisional ratings meaningfully.

If you run placement matches, add their slots before the first round. For example, with 2 newcomers each playing 2 placement games on separate tables, you add 2 slots, or about 50 minutes in this example.

Where to Put Breaks

Schedule one short break after every two to three rounds, and one longer break around a meal time. Breaks also give the organizer time to enter results, update ratings and post the next pairings.

A sample afternoon for the 8-player Swiss:

1. 13:00 – Check-in and rules explanation (15 minutes)
2. 13:15 – Round 1 (50 minutes)
3. 14:05 – Round 2 (50 minutes)
4. 14:55 – Break and rating update (20 minutes)
5. 15:15 – Round 3 (50 minutes)
6. 16:05 – Final standings and updated ratings announced

Add a buffer of 10 to 15 minutes per round if games often run long. It is easier to finish early than to rush the final round.

How to Handle Late Arrivals, Dropouts and Odd Numbers

Plan for changes in advance: decide how you will handle an odd player count, a late newcomer and a player who leaves. Then announce these rules at check-in.

• Odd number of players: in Swiss or round robin, one player sits out each round with a bye. Give the bye to the lowest-ranked player who has not had one yet, and record it as a win or half-point according to your announced rule. A bye does not change anyone's rating, because no game was played.
• Late newcomer: give them the default or tier rating with provisional status. In a Swiss event, they can join from the next round with zero points. In single elimination, they usually cannot join once the bracket starts.
• Dropout mid-event: in Swiss, remove them from future pairings. Their completed games still count for ratings. In single elimination, their next opponent advances.
• Unfinished game: decide in advance whether it counts as a draw, is replayed or is excluded from ratings. Excluding it from ratings is the simplest option.

How to Make the Next Round Clear to Everyone

Post pairings for the next round in one visible place, with table numbers and current ratings, before calling players to start. Confusion between rounds is the most common cause of delays at small events.

1. Collect all results as soon as each game ends, using a sheet at each table or one central scoresheet.
2. Update ratings, using the provisional K-factor for players still marked P.
3. Update standings by points, not by rating. Ratings decide seeding and pairings; points decide who is winning the event.
4. Create the next pairings according to your format.
5. Post a list showing: table number, both players' names, their current ratings and the round start time.
6. Announce the round out loud and start a visible timer if matches are timed.

A clear pairing line might read: "Table 2: Alex (1026P) vs Sam (1087) – Round 3 starts 15:15." Showing the P tells everyone which ratings are still settling, so a surprising result from a newcomer is not seen as a mistake in the system.

Common Mistakes When Rating New Players

Most problems with starting ratings come from overconfidence in the first number or from changing the rules mid-event. Avoid these:

• Starting newcomers at the very bottom. A strong newcomer then beats everyone and gains many points from lopsided wins, while established players lose rating unfairly. Start at the default or median instead.
• Using wide tier gaps. A 300-point spread between tiers means one wrong self-assessment causes several mismatched games.
• Treating provisional ratings as final. Do not award titles or fixed seeds based on a rating marked P.
• Changing the K-factor or scale during the event. Keep the announced rules until the event is over, then adjust for the next one.
• Mixing ratings with standings. Players should know whether a list shows skill estimates or event points.

With a consistent default, a short provisional period and, if time allows, a few placement matches, starting ratings for new players stop being a source of arguments. They become a simple tool for fair seeding, sensible pairings and an event that runs on time.