Generally one can find good and interesting solutions among the top voted ones, but it is not necessary the case of the best voted one.

I suggest you follow multiple highly skilled coders (like akar-0) and instead compare your solutions to them.

Well, you have thousands of katas, with all sort of problems and difficulty levels. I don't know what to recommend, try a bit of everything and see what's the best for you and what you prefer. All I can say is I learnt a lot from looking at other people solutions.

Oh, you are a codewars warrior, nice :) Does it get better with lower kyu katas?
My skills are just not yet there to be able to solve 1-5 kyu katas, maybe some of the 5s.
Also can you recommend materials to improve algorithmic skills? I am mainly interested in JS. If you had any ideas, that would be great.

People vote anything "best practice". You should not care much about that, unfortunately it's the same on many katas.

Solutions with O(n^2) time complexity should never be the best practise in my opinion, if there is a better solution with the same readability(not mine, it looks bad without refactoring).
I see here very often, that not good solutions have been upvoted as hell.

It is kind of lame, because if you want to compare and fix your crappy solution you often only see other crappy solutions with slightly different syntax at the top.

Actually no, it doesn't because of the nested looping. Time complexity is the same: O(n^2), which is not ideal.