How does puzzle rating relate to chess rating

@brucebalmer
You shouldn't relate your game rating with puzzle rating. You just need to play and practise themed puzzles.
Don't focus on comparison, rather just focus on play.

@Sacmaniac the more puzzles you do, the harder they get... The puzzles at your level have about a 50 percent chance that you will solve them. My puzzle solve rate is 51-49. You can check yours in the puzzle dashboard.

if your puzzle rate 2000 your blitz rated = 1500.if your puzzle rated 2500 your blitz rated = 2000

@ShaneTrain #12 Is it just a coincidence or is there actual science behind it?

For example in my puzzle dashboard I'm overall 71% but in Endgame puzzles I'm 100%. And these stats can be skewed for multiple reasons to fit a theory too.

Another example is your standard puzzle rating is 2500 but your puzzle streak is only in the twenties. You are doing harder standard puzzles and a lot more of them then me, but my puzzle streak numbers are much higher. Why?

It suggests to me that I have a lot more patience and I'm more careful than you when doing puzzles. And I also don't do near as many standard puzzles as you.

This 50% number seems like an arbitrary number pulled out of a hat by stingerpuzzles in #8.

@Sacmaniac It looks like you don't do many standard puzzles. If you did them often I would expect the puzzle ratings increase and your success rate eventually drops to around 50. That's just how the rating system works. If you win, you get harder puzzles or games until they are at your level.

Whoops, my mistake I thought your standard puzzle rating was 2500 when it is only 2100.

And it's not like I have a small number of the standard puzzles with 1600+ completed. It's over a thousand so how do you explain my 71% success rate?

The issue is again, where does this 50% number stem from?

If you put a gun to my head and forced me to get better than 80% for the next 1000 puzzles starting at zero you must understand I am not under any time constraints and patience is a factor so I'm going to take several years to complete the task and this 50% number by stingerpuzzles in #8 is subjective.

@Sacmaniac said in #16:
> Whoops, my mistake I thought your standard puzzle rating was 2500 when it is only 2100.
>
> And it's not like I have a small number of the standard puzzles with 1600+ completed. It's over a thousand so how do you explain my 71% success rate?
>
> The issue is again, where does this 50% number stem from?
>
> If you put a gun to my head and forced me to get better than 80% for the next 1000 puzzles starting at zero you must understand I am not under any time constraints and patience is a factor so I'm going to take several years to complete the task and this 50% number by stingerpuzzles in #8 is subjective.

Lichess uses the Glicko 2 rating formula for puzzles. That is were the 50% number is coming from.
However, as you point out, the puzzle solver is controlling the conditions of solving the puzzle. They can do things, like move the pieces around on another board, that would not be allowed in a game. So whether the puzzle rating computed from Glicko 2 applies, depends on what the puzzle solver is doing.

@Sacmaniac

pretty sure shane and stinger are technically right. if your sucess rate is so high that means you will gain more and more rating unless you do alot of themed puzzles or easier ones:

ive heard that the top buckets of the puzzles dont contain as many easy motifs and you probably wont have many 2500 rated fork puzzles. so what happens is you might get very underrated puzzles in the healthy mix sometimes (where you lose alot more than you can win). thatswhy from certain rating you will not even out at 50% anymore because your doing puzzles way below your puzzle skill.

one more reason why there is less very hard puzzles in general might be that all puzzles are forcing, the lower the edge the more potential sidelines arise disqualifying the position from being a puzzle. while on the other hand if its too forcing that means its too easy, so you can see why you cant constantly do 2.5k puzzles: either the positions are not hard enough or not forcing enough.

your rating is probably just too high already at 24xx. im at 2100 and constantly climbing in rating with barely above 50%. but they are right on the math, benwerner (€: and jomega while i was typing this essay) confirmed puzzles use the same rating system based around a median in another thread iirc. its other reasons leading to the discrepancies in winrate needed to sustain the different ratings. its probably much like a high rated player having trouble to find action from players rated high enough. magnus played a group tournament 3 days ago and dropped alot of points coz the avrg rating was 2000 "only" and he had "only" 76%WR. then he got most of them back from naroditzky.

i dont know

@Rookitiki @jomega I've read both your posts and I'm still trying to wrap my head around this 50% number as it relates to puzzles because games are a completely different thing.

As jomega noted, "things can be done", to manipulate puzzle ratings that can't be done in timed games.

Now, that said, I've been on other chess sites where the puzzles are timed and the solver is awarded for speed and I can completely understand how they would do well to get over 50% because they are worried about a score or measurement.

However, There is a reason why puzzles and correspondence games don't have a leaderboard. It's too easy to cheat and get assistance without getting caught or stress under time pressure. Puzzles are the patience game.

Puzzles are by their nature are usually forcing moves and if a person is patient they can systematically work their way through
the moves of almost any puzzle... given enough time.

And I've also lost the plot on the point of this post by @brucebalmer too.

If anyone can find a link to an article that explains the math on (untimed)puzzle ratings using the Glicko2 system I'd appreciate it.