Strong fluctuations in tactical skills...and I still don't know exactly why

learning process is usually uneven; one skill set is polished; then you get to another and make mistakes; thus fluctations in tactical rating

The Tactics Trainer at chess.com is goofy in that you can get a dozen problems in a row correct, then miss one on the the third move and end up with a lower rating than you started.

I work tactics problems every day but haven't done them here for a while. We're not supposed to compare chess.com with other sites, so I'll stick to chess.com here and say the algorithm needs to be changed and the problems need to be more consistent with the tactics rating of the person.  Some easy mate-in-one problems are rated higher than some difficult several-move problems.  The idea that some problems should be solved in 5 seconds is beyond ridiculous.

The fact that you can set it up so you work problems within one tactical motif is valuable in building pattern recognition.

I agree that the tactics can be challenging at times, but I like the idea of getting 12 in a row right, but one wrong and losing all the points you gained. That is exactly how a real game of chess works. You can make 12 good moves in a row, then one bad move loses the whole game. I also agree the idea of solving some problems in 5 seconds is ridiculous. Because that's rarely how real chess works. If it takes me 2 minutes to find the solution I dont feel too bad because that's how long it can take me in a real game.

I think tactics training is about the best time spent on chess. If you have a half hour a day for chess, do tactics.

chesstempo.com is a good learning website

On the other hand, most positions in a game do not require flashy tactics. I was watching a video of a famous coach who explained that if you really want to improve, you have to work hard on your games, analysing them carefully, learning from these mistakes and using these games to learn openings and endgames. Not first learn openings and endgames, but the opposite: going through the games and checking the played moves with your repertoire.

I always knew that the analyses (not with an engine!) are fundamental to make progress. But guess what: I am still postponing this work, doing things that are easier (like learning openings and endgames and tactics).

Another thing the guy said: play correspondence chess, just one game as black and one as white, and as soon the automatic opening moves are over spend three hour to play your move. Exactly: three hours! I like his ideas and I think I will follow them to try to improve. Now I am working on a book on tactics, and I will focus on knowing these combinations well (a total of 1001 exercises, including things for beginners like mate in one or two).

But I am happy to see that I am able to improve a lot on TT by investing more time on the positions, jumping quickly from 1500 to 1700 (okay, I dropped a lot of point again playing like a zombie).

The same thing happens in games, they don't make their plays easier or harder because you're a (insert rating here) player. Sometimes you will be stuck with an 1800 level "puzzle" in a game, board states can be complicated or simple at various levels of play, The level of the player only gives you some leeway on how much you miss and get away with, not how hard it is to spot the game-changing tactic.

 

As for the idea that there are tactics you should see immediately, these do exist, you can't sit for a minute and go "oh, I'll just do this quick little move" unless you intend to lose on time. You have to be able to recognize which moves to take time on and when to make a quick and clear gain without needing to spend much time on it, to save for the previous set of moves you do.

Tactics is to learn! Do not worry about rating but what you learn from each problem!

Same here. My OTB rating is around 2000, my correspondance rating over 2100, but my TT rating is about 1400. Not because I miss tactics, but because I'm slow.

I know this is a relatively old post, but perhaps I can help anyone that stumbles upon this with my insights. For the fluctuations, you're probably encountering certain puzzles with tactical motifs you're weak in-- say 100-200 points weaker than other tactical motifs you're much stronger and more accustomed to passing or doing well in-- so you'll fluctuate constantly because the puzzles are constantly rotating between your strong and weak points.

If anyone has this problem, as I know I do, just check your puzzle statistics and see what motifs you're weak at and create practice problem sets that target these weaknesses until you feel very strong in them-- do not return to rating mode until you've worked over your weaknesses!