How do you get 1337 at tactics?

So you think chess tempo works just as fine as a book? Maybe even better? I've seen that it keeps stats on your progress so you can see what kind of tactical themes you are good or bad at. How would you split up an hour of training time per day between fast solving on tactics trainer (pattern recognition) and slow solving on chess tempo (calculation)?

Problems that require analyzing the position and calculating are much more instructive since they imitate a real tense chess game. On the contrary, solving a few hundred "sac a queen and give a R+N mate"-type problems will hardly teach you anything new after you've done a few of them. 

One of the best methods is "step by step". You take a game full of tactics by some eminent GM, open it at a critical moment, give yourself some time on the clock and start actually playing! Write down your moves, just like in a real game, then see the move that has been played and the commentary. 

Bonus: if you love chess engines and have a good one, go to step 3 (after you've played the game out and checked it against the commentary) - check the analysis using an engine. Sometimes it turns out (with me and my friends) that we actually played a stronger move than the GM, and the book didn't mention it for some reason. 

Great hearing an answer from a GM about this! I thought GMs drilled tactics to get where they are. So training tactics by calculating let's say Tal's tactical games step by step is key to getting better at it in long timed otb tactical play? I'll definately look into this.

I see you do not think much of mating pattern books but what do you think of training tactics timed on tactics trainer and untimed on chesstempo?

Also how would you plan training sessions each day? Just solving tactical games or combine with tactics trainer for example? How did you personally do it to reach that far?

Books are great, but for me Chess Tempo is better.  The added motivation I get with the rating system and community spirit on Chess Tempo is significant.

My tactics training goes roughly as follows.  Every day I do one problem which takes me about 30 minutes (or more) to figure out, for calculation training.  And I do ~40 very easy problems just for pattern training.  (Note that a Chess Tempo premium membership is required to train with custom problem sets like "easy problems".)  The hard problem is rated, and the easy problems are unrated, so I'm generally willing to spend as long as it takes to get the hard problem correct (perhaps several hours), and this has really improved my chess.

This, by the way, is why I keep arguing that Tactics Trainer needs a rated untimed mode.  :)

I love it when this happens!  I am pretty awful at chess, so this doesn't happen with me very often, but I've noticed this a couple of times, especially with older books.  :)

I'm working my way through Ray Cheng's Practical Chess Exercises now, and I must say it's a breath of fresh air.  The reason these exercises are "practical" is that the best move in many exercises (you don't know which ones in advance) aren't really tactical at all, but rather aim to significantly improve more strategic aspects of your position.

I like this, because in order to improve "my tactics" I need to be able to identify them in positions where I didn't already know there was a winning tactical blow.  In real life, I certainly can't solve all my problems looking for clever ways to sac my Queen.  :)

Mating patterns are ok, but the most popular ones (like smothered mate, two bishop sacrifice, etc.) can be memorized quickly. So most puzzles involve calculations, some of the variations of which might end with a mating pattern. But not consisting of a single mating pattern, those are too simple. I have never read such books, actually...

Solving tactics untimed is not that good since in a real game your time resources are limited. However, this is often neglected (e.g. you're solving tactics on a plane, or blindfoldedly, one should be a training addict to time oneself).

One more advice: if you spend about 10-20 minutes on a problem, leave it, and return to it after some time. Never peek into the answers before actually solving it! If you do that, then you will hardly improve at all.

I don't have any training sessions each day. Sometimes I dedicate an hour to chess, sometimes (when I'm at a tournament) a few hours. Sometimes forget about studying chess at all and hang around at Chess.com forums and the like. Sorry for being so unprofessional. 

chesstempo practice endgame problems have helped me.  I don't like the timed tactics on chesstempo (and also not on Emrald).

Personally, I've learned so much from doing untimed tactics problems.  It seems to me that rushing hinders learning.

As an analogy, someone reading a book might require 20 seconds for page one, three minutes for page two, and 40 seconds for page three.  But if they were told, "You're allowed only 30 seconds per page," they're not going to get much from that book.

Yes this training schedule sounds reasonable.

Untimed TT would be nice but I get the feeling that a vast majority of the problems are way too easy to solve untimed.

This book seems very intresting. I considered getting it. Sounds promising.

Yes, it is probably no use repeating patterns you already know unless they are tangeled into eachother or appear several moves deep in a calculation. I might need to burn some mating patterns into my brain still. They are not quite 100% yet but after that I need more complex drills like you said.

Good point about leaving unsolved problems and going back to those the next day.

Thanks for the thorough answers.

The number of times I've played a winning move on Chess Tempo only to be told that it's not the best, and this has been legitimately annoying, has been very small — even after thousands of problems.  Most of the time if I'm getting this message it's for a good reason — I simply didn't find the best move.  Here, Chess Tempo could justifiably fail me, but she is kind enough to give me another try (since my move was winning).  :)

I disagree that spending 30 minutes on a problem is a waste of time.  This really exercises your calculation.  IIRC, dpruess has advised doing something like this, at least once a player gets to a certain rating.

As for your "ego" comment — I already mentioned above that I am bad at chess.  :)  I just don't have the natural ability.  So I try really hard to get better, but I still play some terrible games.  I have fun in the process, though, so it's all good.

If you play a winning move, but there's better, Chess Tempo says, "Good move, but not the best.  Please keep looking for a better move."  In this situation (of multiple winning moves), the #1 move has to be (if I'm not mistaken) something like +3 better than the #2 move.  So most of the time if you get this "try again" message, you simply didn't find the best move.  And the point of tactics puzzles is to improve your chess, so it's nice that Chess Tempo doesn't, at that point,  say, "Problem solved — you found a win (even if there was a better move)."

This system works well, but, as you rightly point out, there are times where this is annoying, as the #1 move (while being something like +3 better than the #2 move) is very complicated, and I would rather play the #2 move in a game.  However, as mentioned above, this is very rare in my experience.  Your post seems to imply that this happens 20% time, which is far from the case.

Still for many of us it is only of primary intrest to reach a winning position. If I find 2 good enough move in a position but can't find that 3rd critical and in some cases unintuative move that the computer wants it will eventually fail me. I don't see why I should fail eventhough I've come up with 2 winning moves. But in general chesstempo works fine.

Blitz can be good but it really screws up your thinking processes, you start to play on intuition rather than calculation.

As a rough guide starting from say 1400, 1000 problems equals 200 points, I have completed over 9000 problems on Tactical Trainer. Some of the other players are in the 50,000 .

I use the 1000 Checkmates by Reinfield and do 200 0r so in my head in about 2 hours. Usually scoring around 90%, good little book for warming up.  

Still fancy it ?

Have you noticed any improvement in your tactical ability in your games after so many tactics trainer positions solved?

I'm not sure that's true in any sense.

 Yes I have, my main problem in my games is that I get sloppy sometimes, I do not use falsification properly, blitz chess and timed tactics with forcing lines can give you this problem and it is hard to shake off.

I see. What do you mean by falsification?