How real is the rating of puzzles with respect to the ELO rating?

How can the rating of puzzles on lichess be converted in the ELO rating, if possible?

It would be good if Lichess could do a statistical analysis and say what the correlation is. It may be low. They could tell you what your expected rating would be. It would be interesting.

Since the lichess's puzzles are untimed the correlation should be very weak.

Im totally with @marcus624 But in my opinion the ELO puzzle is inflated. I achieved near 2400 in lichess elo puzzle and im a 1600 fide and around 1900/1950 on lichess blitz elo. And i tell you im not even remotely near of a player that can have the tactics of a 2400 fide.

Puzzle rating means..........absolutely nothing. Unfortunately. My puzzle rating is over 2500 but I haven't beaten an 1800+ rated player at 30/20 for a long time.

Hopefully I will start to play a little better soon, when it gets cold and I can't do outdoor stuff.

I agree that puzzles indeed have very small connection to real game.

But at the same time, in my opinion, puzzles, especially untimed, are also paradoxically the best part of chess. It's just problem solving, pure and pleasant.

There is no point to having a rating system for puzzles. The point of puzzles it to learn something new about chess, not to get a rating. The measure of what I've learned is my rating for playing games. I actually dislike the fact that I get rated for my puzzle performance. This makes me not want to do puzzles on this site, so over time I will likely do this less often or not at all. Is there a way to turn the ratings off? If not *****HINT***** make one!

@jhehrun Puzzle rating does have a point: Finding appropriate puzzles for a huge variety of players' strength.

The skill of playing chess has several components. Tactical skill alone is not enough to determine the strength of a game. Moreover, the time factor and style of play also affect the coefficients for converting one rating to another. There is a correlation, but the calculation error is too large. I think it is more correct to look for the range in which the real rating falls. Try taking 80% of rating of puzzle. This is a very rough approximation.

Thing is...during your game you don't suddenly get a big sign come up saying 'white to play and win'...
puzzles in that respect are a little artificial...