What Is Elo in Chess?

"My Elo is only 900. Is that bad?"

I get this question almost every week, and it always carries the same quiet anxiety, as if the number were a verdict on intelligence or talent. It is not. Elo is a prediction tool, nothing more. It estimates the probability that you beat a given opponent, and it updates after every game. Once you understand what it actually measures, it stops being a source of shame and becomes what it was designed to be: a compass.

This guide explains where the rating comes from, how it moves your points, what counts as a good Elo, what the real average looks like, and why the three numbers you see on FIDE, Chess.com, and Lichess almost never agree.

Where the Elo rating comes from

Elo is named after Arpad Elo, a Hungarian-American physics professor and strong amateur player who designed the system in the 1960s. The world chess federation, FIDE, adopted it in 1970 to replace older, clumsier ranking methods. The word is a surname, not an acronym, which is why you write "Elo" and not "ELO".

The core idea was elegant for its time and still holds today. Instead of awarding fixed points for a win, the system asks a question before the game: given the gap between these two players, what result do we expect? You gain a lot when you beat someone the model expected to crush you, and almost nothing when you beat someone far below you. That single principle is the whole engine.

How Elo works, in one simple idea

Every rating implies an expected score. If you are rated 1200 and your opponent is 1200, the model expects a 50-50 match. If you are 1200 against an 1800, the model expects you to lose most of the time, so a draw alone would already be a small triumph for you and a small disaster for them.

After the game, the system compares what happened to what was expected. Beat the favorite and you take a large chunk of points. Lose to someone you were expected to beat and you drop hard. Deliver exactly the expected result and your rating barely moves. The size of each swing is controlled by a number called the K-factor: high for new and junior players so their rating finds its level fast, lower for established players so it stays stable.

This is why a long win streak against weak opponents raises your number slowly, while a single upset against a much stronger player feels so rewarding. The system is not measuring effort. It is measuring surprise.

What is a good Elo rating?

There is no universal answer, because "good" depends entirely on the pool you are comparing yourself to. Still, here are honest reference points that match what I see across hundreds of students.

• Under 800: you are learning to stop hanging pieces and to see one-move threats. Completely normal starting territory.
• 800 to 1200: you know the basic tactics and you are building real board awareness. The large majority of online players live here.
• 1200 to 1600: you calculate short lines, you have opening habits, you convert clear advantages. This is already a strong club-level amateur.
• 1600 to 2000: serious, consistent play. Reaching 2000 is a genuine milestone that most players never cross, and I wrote a full method for it in how to reach 2000 Elo in chess.
• 2000 to 2400: expert and candidate-master territory, usually the product of years of structured work.
• 2400 and above: international masters and grandmasters. The peak human rating ever recorded is Magnus Carlsen's 2882.

If you want the texture of what changes between levels rather than just the labels, read what 2000-rated players do differently.

The average chess Elo is lower than you think

Most beginners assume the average is somewhere around 1500, because that is the number platforms often hand to a brand-new account. It is not the population average. The true median online rating sits well below that, frequently in the 600 to 900 band depending on the site and time control, because the playing pool is dominated by casual and improving players, not by titled experts.

So if you are sitting at 900 and felt embarrassed at the start of this article, look again: you are very plausibly at or above the median of everyone playing. The feeling of being "below average" usually comes from comparing yourself to streamers and titled players, who are statistical outliers, not the norm.

Why your Chess.com, Lichess, and FIDE ratings never match

This is the single most confusing part for new players, so let me be blunt: a rating only means something inside its own pool. The same person can be 1500 on Chess.com, 1800 on Lichess, and 1400 in FIDE-rated over-the-board tournaments, all at the same time, and none of those numbers is "wrong".

Three reasons drive the gap. First, the starting point and calibration differ from site to site. Second, the playing pools are different, and a rating is always relative to the people in it. Third, time controls are not the same animal: your blitz, rapid, and classical ratings measure different skills and will not line up even on a single platform. Comparing a Lichess blitz number to a FIDE classical number is like comparing a sprint time to a marathon time.

The practical takeaway: pick one rating, in one time control, as your reference, and track that one over months. Switching the yardstick every week is how players convince themselves they are stagnating when they are actually improving.

What Elo can and cannot tell you

As a coach, here is how I use a student's rating. It tells me, roughly, the level of opposition they handle and the size of the gap to their next goal. It is a useful diagnostic input. What it does not tell me is why they are stuck, and that "why" is where progress actually lives.

Two players rated 1300 can be completely different cases. One hangs pieces under time pressure and would jump 200 points just by slowing down in the last five minutes. The other plays clean but never studies endgames and throws away won positions. Same number, opposite prescriptions. The rating flags that something can improve; it never names the lever. That is exactly the job a structured review, or a coach, does.

So treat your Elo as a thermometer, not a diagnosis. It tells you there is a fever. It does not tell you the cure. If your number has been flat for a while, the problem is almost never "study more", it is "study the right thing", which I unpack in how to improve at chess efficiently.

What to remember

Key takeaways

• Elo is a prediction system, not a verdict. It estimates your chance of beating a given opponent and updates on surprise, not effort.
• "Good" is relative to the pool. Crossing 2000 is a milestone most players never reach; the all-time human peak is 2882.
• The real average online rating is far below 1500, often in the 600 to 900 range, so beginners routinely underestimate where they stand.
• Your FIDE, Chess.com, and Lichess numbers will not match, and that is normal. Track one rating, in one time control, over months.
• Use Elo as a thermometer. It signals that something can improve; it never names the lever.

Your rating is a starting point, not a sentence. If you want to turn that number into a clear next step instead of a source of anxiety, begin with how to reach 2000 Elo in chess, or let me look at your games directly through JD Chess coaching.