Chess Ratings Explained: Elo, FIDE and Online Scores

What does a chess rating actually measure?

A chess rating estimates playing strength within a particular system and group of opponents. It is useful for comparing results and finding suitable competition. It is not a universal score that means exactly the same thing on every website or in every form of chess. Lichess explicitly warns against comparing ratings across separate pools. ^[2]

As of November 4, 2026, the practical first question is therefore: which rating? Write the organization or website, the time control and whether the number is established or still changing quickly. A number without those labels leaves out much of its meaning.

For example, compare your own recent rapid results with your earlier rapid results on the same service. Do not combine a puzzle score, a blitz peak and an over-the-board result into a personal average. That would mix measurements designed for different tasks and groups. ^[5][6]

If you play from Iran, the same distinction matters when a friend describes a strong online account as having an official international rating. Ask which competition produced the number. Enjoying online chess does not require turning every result into a claim about a formal tournament record.

How does the Elo idea turn results into points?

The basic idea is to compare your actual score with your expected score. A result above expectation moves a rating upward; a result below expectation moves it downward. FIDE's standard regulations calculate the differences across the relevant games, multiply their sum by a development coefficient called K, and round the period's change. ^[1]

Here is a deliberately simplified arithmetic example, not a forecast for your account. Suppose your expected score is 0.5 and the chosen multiplier is 20. A win, scored as 1, gives 20 times (1 minus 0.5), or plus 10. A draw gives zero change in this example, and a loss gives minus 10. The assumptions are part of the example.

The important idea is surprise relative to expectation. A victory over a much stronger opponent carries different information from a victory over a much weaker one. Chess.com's explanation similarly connects changes with the gap between players, while also considering uncertainty in their ratings. ^[3]

Do not copy the example's multiplier into every rating calculator. Before using a calculator, choose the correct system and read its inputs. A clear calculation with the wrong assumptions still produces the wrong answer for the competition you actually played.

What makes a FIDE rating different from an online score?

FIDE standard ratings come from qualifying events and submitted results under its regulations. A first published standard rating requires at least five games against rated opponents and a resulting rating of at least 1400; the qualifying results can be pooled within the specified period of up to 26 months. These are publication rules, not a promise that any five games are enough. ^[1]

For a first event, ask the organizer whether it is FIDE-rated, which category it uses, how your player identity is recorded and where the submitted result can be checked. Keep the event name and your results. These questions help distinguish a casual club score from an official rating record.

Rapid and blitz have their own FIDE regulations. Those rules include eligible time controls and reporting requirements, so the label used by a website should not automatically be treated as the definition used for a FIDE-rated event. The two settings can organize play differently. ^[8]

The FIDE list is updated monthly under the regulations. A delay between finishing a game and seeing it in the official list is therefore not the same thing as an online service failing to update immediately. Check the relevant list and event report before deciding a result has disappeared. ^[1]

Why do online ratings jump so much at the beginning?

Chess.com uses Glicko, which includes rating deviation: a measure of uncertainty about the estimate. New accounts or a return after inactivity can lead to larger changes because the system has less confidence in the current number. The opponent's uncertainty also matters. ^[3]

Glickman's explanation separates the estimate of strength from its reliability. In his system, information from games reduces uncertainty, while time without competing increases it. The model does not assume that someone returning after a long break has necessarily stayed at precisely the same strength. ^[4]

Imagine two fictional players with the same displayed rating. One has been active regularly; the other is returning after a long absence. Identical displayed numbers need not carry identical confidence. That is why a rating system can sensibly react differently to their next results without treating one player unfairly.

Glickman also explains that the winner's gain and loser's loss do not generally have to be equal in Glicko. Their different uncertainty affects the adjustment. A screenshot showing unequal changes is therefore not, by itself, evidence that points were calculated incorrectly. ^[4]

For your own progress record, mark a new or returning account as unsettled. Avoid declaring a new personal level after a single dramatic change. Review the pattern of results and the quality of your decisions over a longer stretch of ordinary play.

Can you convert Lichess, Chess.com and FIDE ratings?

There is no dependable universal conversion such as subtracting the same number from every Lichess rating. Lichess uses Glicko-2, while Chess.com describes its regular ratings as Glicko and FIDE uses an Elo-based approach. Differences in the player pools also prevent a direct numerical comparison. ^[2][3]

A survey of people using two services can describe that particular sample. It cannot turn a rough relationship into a personal certificate. Before trusting a conversion chart, ask when it was collected, which time controls it compared and whether the accounts actually belonged to the same players.

If you want an over-the-board benchmark, obtain results in that setting. If you want to measure improvement online, stay with the same platform and format. These are practical choices for making your comparison easier to interpret, not a claim that one community's players are inherently better.

Even shared software would not eliminate the problem. Lichess notes that separate pools remain separate comparisons. A number's meaning comes from the results and opponents used to produce it, not just the name of the mathematical formula. ^[2]

Why are rapid, blitz and puzzle ratings different?

Chess.com tracks ratings separately for its time controls. Its help page explains that rapid gives more time for calculation, while faster formats place different demands on decisions and time management. Improving one established rating does not automatically raise the others. ^[5]

That does not mean you should choose the format with the most flattering number. Choose the kind of chess you want to practice. When keeping a progress note, record the time control beside the rating so a switch in format is not mistaken for a sudden improvement or decline.

Puzzle scores also need their own context. Chess.com says it introduced a new puzzle rating system in October 2025 and recalculated ratings. A changed scale can move a displayed score without representing the same change in a person's solving skill. The service now describes its puzzle calculations as Glicko-style. ^[6]

For example, keep a note about the type of mistake you made: missed defensive move, overlooked check or incomplete calculation. Such notes remain useful when the scoring system changes. Use the score as one signal about your practice, not the entire description of what you learned.

Does an analysis score prove your true playing strength?

A post-game report is different from your ongoing match rating. Chess.com's Game Review provides engine-based move classifications, an accuracy score, key moments and opportunities to retry positions. These are tools for examining that game, not a substitute for the history of results used in a rating system. ^[7]

Start with one turning point. Before revealing the engine's answer, write down what you were trying to do and which reply you missed. Then compare the alternatives. This is a suggested review routine; it does not promise a specific increase in points or intelligence.

The same service warns that its game-rating estimate is not your Chess.com rating or an overall measurement of playing strength. A flattering estimate from one game should not be advertised as a new official level. Look at the moves it highlights and the explanation behind them instead. ^[7]

Separate feedback from accusation as well. A strong result or a high accuracy score alone is not enough for you to determine how another person played. If you have a concern, use the service's reporting process and avoid public claims based only on a number you do not fully understand.

How should you use ratings without chasing every point?

Choose a manageable record: date, platform, time control, rating and a short lesson from the session. Keep peaks separate from your usual level. If you change sites or return after a break, mark that change so the next comparison does not hide it.

Uncertainty is part of the system, not an insult to the player. Glickman's model treats reliability explicitly. Read an unsettled score as a provisional estimate that needs more information rather than as a verdict you must defend immediately. ^[4]

For puzzles, focus on understanding the complete solution before checking the next score. Chess.com's updated puzzle guidance distinguishes difficulty settings and explains that adjustments reflect the challenge relative to the solver. A missed easy puzzle and a solved difficult one do not carry identical information. ^[6]

Pouyam's chess page is a place to explore the game. If you compare positions on a leaderboard, read the rules that define its score first. Check whether a leaderboard rewards accumulated points, game results or estimated playing strength before comparing it with an official chess rating.

At the end of a session, ask a concrete question: did I notice the opponent's threat, manage the clock or check the ending more carefully? Those questions give you something to work on even when the rating moves the wrong way. A useful progress record should help you decide what to practice next.