Originally posted by GolubI've read about this exercise:
Actually, if someone has ideas for how to improve calculation ability in other ways, I'm quite interested in it too..
And I should add I am NOT interested in combinations and tactics of that sort. I am interested in calculating many different variations (possibly including tactics).
Take a messy position from Kasp. or some other GM. Say a middlegame position with lots of pieces still on the board, no clear plan for either side and lots of imbalances.
Get out a piece of paper and start writing out all the variation you can see as deep as you can go. After that give the resulting position a value, white slightly winning etc...
Then check your variations with a chess programs.
I read that everytime this person did this, he claims his rating went up 200 points.
This was posted on a Senior Masters Website and the player who did the exercise is now a Master + .
I've read about similar exercises in different books.
So for example, take a Kasp game and go to move number 20 and then work out all the variations in your head and write them down. No matter how dumb they may seem.
So with 7 pieces for white, excluding king, you should have at least 7 variations plus all the different squares those pieces can move, plus pawn moves, plus different reply's for black.
Man you could have over 100 variations easily!
Other then that, do tactics and combinations in your head. That's how I learnt.
Originally posted by RahimKSounds interesting (and a lot cheaper). Have you tried it yourself? I feel that going up to 1700-1800 requires good calculation, and while I can get a win here and there, it is mostly my calculation and tactics that fail me in the end.
I've read about this exercise:
Take a messy position from Kasp. or some other GM. Say a middlegame position with lots of pieces still on the board, no clear plan for either side and lots of imbalances.
Get out a piece of paper and start writing out all the variation you can see as deep as you can go. After that give the resulting position a value, white ...[text shortened]... easily!
Other then that, do tactics and combinations in your head. That's how I learnt.
I tried out your method, and it was a quite interesting. The position I took was from a Gazza game (as white):
I wrote down all the lines I could think of on a paper in the order I thought of them. My first 5 variations or so of analysis was quite horrible, but the more I calculated the more accurate it got. Some of my earlier variations I could replace with stronger variations.
After I finished the work (perhaps 30 minutes, didn't time it), I played out all the lines on the board and I managed to confirm that most of my analysis was correct, but one, where I actually gave a position a huge plus which was a huge minus =) However, I managed to find two main variations which lead to an advantage (1. a3 and 1. Qc2).
For my 1. a3 line I was able to to nail the first four moves, but in the 1. Qc2 I missed an ever better move on the 5th ply. I guess that means I can see 2-4 moves ahead with decent accuracy. On average, I looked three moves deep.
But I think the method has some flaws. I got to write down all my variations which really simplifies the task of memorizing it all. In a match situation I can't write anything down, I have to rely on my memory. I think this is also the problem, because I so often go "back and forth", while with the paper I don't have to do that. What I really need is a structured way of coming up with the variations and memorizing/categorizing/valuing them.
I did notice an interesting thing though: just enumerating all the tactical (and potentially tactical) variations first seems to be a smart way of doing things because you then set out the constraints of the position. It took me 8 or 9 variations of calculation before I noticed I could even use the open diagonal to the black king. If I had first noticed this, I would have simplified my task alot (for those of you who have experience with AI, there is an analogy with partially ordered plans), however it is not that trivial to do, I guess.
Originally posted by Golublol. I'm rated 1,800 on a good day and can't calculate my way out of a paper bag. On this site, because it is correspondence chess, I can use the analysis board to compensate - but even so miss loads of stuff.
Sounds interesting (and a lot cheaper). Have you tried it yourself? I feel that going up to 1700-1800 requires good calculation, and while I can get a win here and there, it is mostly my calculation and tactics that fail me in the end.
Excelling at chess calculation by Agaard, which giantrobot mentionned above is a good book; but the basic technique for improving calculation is practice of the form recommended by RahimK in his post, and is essentially what Kotov says to do in Think Like a Grandmaster.
What would be useful is a series of increasingly difficult positions, initially with the plan pointed out, something like: "White would like to start an attack with move x is this possible?" The reason I say this is that difficult and unclear positions are difficult and unclear and you don't really learn anything from failing to solve them. Whereas if you start with relatively easy exercises, possibly with all the tactical motifs explained where you have to string them together, then you can build up confidence and mental arithmetic skills. When you learn (mathematical) mental arithmetic you don't start with square roots, you start with easier problems and work upwards - I don't see chess as being any different.
Originally posted by DeepThoughtI am sorry, it is exactly what I meant. I was refering to such a strength in "live" chess, I should perhaps have mentioned it.
lol. I'm rated 1,800 on a good day and can't calculate my way out of a paper bag. On this site, because it is correspondence chess, I can use the analysis board to compensate - but even so miss loads of stuff.
Originally posted by DeepThoughtEven though I think it somewhat flawed to write down variations, I think I am going to try that approach. At least I get some brain work-out, if nothing else.
Excelling at chess calculation by Agaard, which giantrobot mentionned above is a good book; but the basic technique for improving calculation is practice of the form recommended by RahimK in his post, and is essentially what Kotov says to do in Think Like a Grandmaster.
Originally posted by GolubMy playing style involves aiming for positions that I can keep control over and don't have to worry so much about the type of tactical complications that are going cause me calculational problems, I rely heavily on pattern recognition and intuition to compensate for not being great at calculation. It causes problems for me here as the analysis board only helps you so much, if you can calculate accurately in your head then you end up finding things on the analysis board that you wouldn't otherwise (again, a bit like arithmetic and using calculators). The main thing to bear in mind is that your opponent probably can't calculate very well either so any work you do to improve calculation will get more results than any amount of opening work and so on.
I am sorry, it is exactly what I meant. I was refering to such a strength in "live" chess, I should perhaps have mentioned it.
Originally posted by GolubWhy not just do it right? There are no shortcuts to learning how to calculate properly. At least follow those who's methods have been proven to work at the highest levels (e.g. the GMs mentioned in this thread) and not some nonsense like gaining 200 points per exercise (that initial article actually claims 100 points, which is still nonsense).
Even though I think it somewhat flawed to write down variations, I think I am going to try that approach. At least I get some brain work-out, if nothing else.