I flicked through some papers that described how one can obtain an ELO estimate based on a game. (Basically we were comparing players' moves to the best moves suggested by a strong chess engine. My question is, is there any online/offline resource when I can do this?

  • Unfortunately, I doubt people will invent anything tangible at the moment... Good luck though... +1 It would be very handy to test your ELO strength that way instead of paying so much money for rated tournaments and traveling expenses... Jun 22, 2014 at 22:48
  • 1
    Perhaps this paper can be useful: web.tecnico.ulisboa.pt/diogo.ferreira/papers/… Aug 17, 2014 at 15:24
  • @AlwaysLearningNewStuff if knowing your rating is the reason you'd go to a live tournament, then yeah, don't go!
    – David
    Jul 7, 2020 at 7:46
  • I am really interested of what is the latest we have regarding this post. Is the only reason I play chess is to check my own mental faculty
    – daparic
    Nov 18, 2020 at 12:55
  • My idea for an estimator would be take thousands of games, run them against stockfish to calculate the deviation from perfect move, then build a statistical model to estimate against true elo. Actually that's pretty much what the linked papers do.
    – qwr
    Aug 3, 2023 at 3:29

6 Answers 6


It's a matter of how much data there is. When you first get a rating in a tournament, it is a provisional rating until you reach X number of games. In the US I think you need 20 games. The provisional rating fluctuates more because of a single loss or win.

However, I recently obtained a copy of Houdini Pro 4. In the help contents is this: "Unix inventor and computer chess pioneer, Ken Thompson, has developed an algorithm which allows one to create an Elo rating list out of an arbitrary set of games".

There is also a step-wise process to generate an Elo list. So, perhaps it can not be done based "on a game", but maybe not too many more are needed using Ken's algorithm? Not sure, I never tried using that tool.

  • I should also mention the performance rating. It can approximate rating given a small amount of data. This is an excellent site that goes into details on the subject of ratings calculation: chessmetrics.com/cm/CM2/Formulas.asp
    – ezaspi
    Jun 26, 2014 at 11:06
  • Does the algorithm you mentioned consider only the results of the games or the actual moves? The quoted description sounds like it only considers the results.
    – JiK
    Jun 30, 2014 at 9:09
  • I haven't seen the algorithm but ELO is not based on moves, just results. A response here by sco-ish shows a link where a rating is given based on moves. But there is no "algorithm". Simply view source on the page and you'll see that rating is assigned based on moves - arbitrarily. To give a true rating based on a single move, IMO, would require a formidable amount of data, to the extent that Chess itself is solved - a very large amount indeed.
    – ezaspi
    Jun 30, 2014 at 10:48

Jack Welbourne created a program like this called Chess Game Report.



Since about April 2023, chess.com's game review has an estimated rating feature. Unfortunately, unless you have a membership you only get one review a day. I can't vouch for its accuracy either.


Software based on a game, probably not to be honest, the best way to get your strength 'measured' is by playing on some internet chess server where you can obtain an online rating.

Although this website can give you elo rating estimate based on a series of exercises which you do. http://www.chessmaniac.com/ELORating/ELO_Chess_Rating.shtml


You are lucky, because I have just published an Android App that does exactly what you need. It is called "Analyze your games" and you can get it here for free:


  • 2
    Nice. Perhaps you'd care to share how you derive ELO in your App?
    – ezaspi
    Jul 4, 2014 at 10:03
  • Thanks. Look very promising. Unfortunately it is not supported by my phone (xperia mini st15i) but I will try it on some other phone for sure. Or can you somehow extend the support for more phones? Plus I'm quite interested, how do you infer the ELO? Is there some machine learning behind that? How accurate it is, you believe? Because lately I was considering to make something like this my master thesis - but after this I guess I wouldn't serve much :) Jul 5, 2014 at 15:08
  • 1
    I am the one who published the app. There is no magic in the calculus of ELO rating, because there is a correlation between ELO and Quality based on thousands of games, as explained here: chess-db.com/public/research/qualityofplay.html
    – user1657
    Jul 6, 2014 at 11:21

Elometer.net estamates your ELO rating on finding the best moves in 76 game-situations.

  • The site is down currently
    – qwr
    Aug 3, 2023 at 3:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.