Ideal strategy for puzzles to get better at chess

@TLansman said in #1:
> I'm someone who basically never does puzzles, and I think doing them more would improve my game but I'm not sure how to approach them.
>
> What's the best choice for puzzle difficulty level, -600, -300, normal, +300 or +600?

-600 will punish your mistakes disproportionately, it is very useful to find your weak spots. If you have some especially weak area, it will keep the difficulty too low in the rest.

> Should I be trying to do puzzles quick or spend a lot of time calculating?

I think second option is better. Calculate until you understand every candidate move and opponent's replies.

Two ideas you could find useful:

* Solve a few checkmate themed puzzles every day. Quite often, the solution of a puzzle depends on knowing a checkmate threat. It will improve your game winning chances, too!
* Review your failed puzzles so you understand not only how you failed, but why (you missed an opponent's defensive move, an unexpected threat, etc.). Use that information to fill the gaps in your solving method.

Good luck and have fun!

@TLansman said in #1:
>

All of the above, but maybe with an experimental design that would adapt the knob regimes could are considering to your reality.

If you want maximal self-awareness to improve the proportions, I would start with some Occam's razor approach.

Do session with only one of your questions, alternative... One day. Spend a mental stamina session. 1 hour or 3 hours (diminishing learning returns within that range, and possibly the next day being ruined depends on your reality).

Then give it some days or not. Trying other regimes.

Don't forget the thematic information loop, not as an entry point (at first, if you don't have the self-observation skill developed), but as a non-flattering mirror after, say, 30 puzzles. The more in your dashboard, the better, because they will get spectrogrammed into a radar plot, and there might be 8 or 10 among the whole set that will end up there.

In order to give that radar plot individual theme components meaning as a decomposition of your good familiarity or your bad familiarity per theme, one would need each of them to have had a good sampling size (even if the statistical measures are not of frequentist framework, sample size still matters, at least more than how long ago the puzzle was played). This is why you would need more puzzles in the random entry mode (at various intended paces, part of the strategy tuning, long haul, I know, but chess is big). There is more than just replaying the puzzles you missed in history, you can use that radar plot, to then focus on the weak ones, then as entry to thematic mode (random offer within that theme having been tagged by the system).

Be sure to have finished all your prototyping within 3 months. That is, if you are interested in the theme feedback from your experience set. As Lichess forgets your thematic data after 90 days. To save on accumulation of data storage, or input size of live data for that Lichess feature set (hypothesis, trying to understand why they chop the data that way).

Have fun and be attentive to all 3 of the board, what the engine is NOT playing, and your own thinking. Lichess does keep all your puzzle results, though, you can also revisit your good and bad solution at first attempt.

I started five years ago with the same idea. Fast or slow. I did both.... But single session wise. To find my best and worse.

Later (year or so) I started to experiment more with the post-puzzle candies of the themes. As I might have hinted above.

The only problem, is my credibility when you read this. You might modulate my transparent reasoning suggestions, with what my ratings might be. But on the plus side, I gave you more than advice, I gave you tools to try yourself. That is what you asked, no?

Ideas of strategies. Good question. Given we only have one life, and chess so big, might as well throw a wide net once in a while.

I forgot to mention, whether you win or not the puzzle there is a post-puzzle charade deliberate work, (not just about the turn by turn not played, that you might have missed, even in success, that made your lucky one move at a time success more insightful, that is one mode of prototyping to test independently of the other Occam single direction above). I mean for the thematic work, you can learn tactical theme definitions according to Lichess and the population predecessors of yours having gone through those puzzles, interactively. Ask your self where exactly at which set of positions or position transition, did Lichess or the others see a theme having been tagged.

Eventually, you will decide that chess is not magics, and that reasoning from definition might be an acceptable way to learn about chess. You will find the courage to attribute tags yourself that you are then familiar with, and help others that way too. If in doubt, keep the track back information and ask here on the forums. That would be fun for all.

Chess is a game of reasoning first. I have spoken.

@MrPushwood said in #9:
> No, not at all. You carry along with you many an idea from the previous moves (at least, you do so unless you're a complete beginner). So I'd say that your notion is more the illusion. ;)

That is the reality of our psychology, the poster you were responding might have been talking about the ideal to aim at, but that might also be an illusion, because we are bound in our fallible ways by some chains...

These days I do 2 puzzle streaks (till I fail which happens usually between 35 and 55) and 5 difficult puzzles (my rating) a day.
I always take my time and try to calculate all the relevant variations. If I haven't seen a win in some of them, I look there even if the answer happened to be correct. If I failed I first check why my idea doesn't work (if I had one of course, sometimes I don't see anything concrete and just make a good looking move) then go to the actual solution.
Interesting to know other people puzzle routines or what you guys do instead.