To fix this problem, I started to analyze my own games. As you can see in my previous post.
But I soon realized that my own games lack something. Certain things simply cannot be learned by studying your own games with the aid of a chess engine. And so I decided to study positions from grandmastergames. The first few books from Buckley, Larsen, Yakovlev, Polgar and Dvoretsky were very disappointing. The positions in those books are not computer checked. If Houdini finds 15 better moves than the move proposed by the author, and the difference is >0.30 points, it is too hard to keep your faith in that you are on the right track. And in these times no author has an excuse to not check his analysis by computer.
Luckily I remembered the book Grandmaster Chess Strategy from Kaufeld and Kern with analyzed games from Ulf Andersson. Sofar I found that the analysis is thouroughly checked by computer. I hope that it stays that way. Andersson has a crystal clear and logical positional style. His games do contain the things I need to learn but cannot learn from my own games.
That is in no way an easy task!
Look for instance at the following diagram:
Andersson played here g4! which is backed by Houdini and Ivanhoe as the best move.
I can't say I really understand this position. That I could find such move myself. But I recognize it's importance, so I give it a try.
The pawn on b6 is a weakness. It can be attacked with Rc6. Hence the knight is bound to the defence of c6. It is difficult to develop the knight and keep b6. b5 just shifts the problem to a6.
The move g4 is designed to undermine e5 by attacking f6. Or, after f5, gxf5, gxf5 the h-pawn can become weak. So g4 is essentially about inflicting a second weakness. You can find the whole game here.
These positions contain quite new patterns and idea's. An whole other way of looking at the position is needed. I'm very curious for more!
I did start to "copy" several positional books into Anki. Anki is a program for spaced repetition.
ReplyDeleteI scan the diagram as question ( front side of the card ) and i scan the explanation of the writer as answer. With the snipping tool of windows i copy these pictures from the sreen and insert them on the front and backside of a new Anki-Card. All done in a few seconds. But i will skip all examples which are not computer proof. And this is a little more timeconsuming. Often there are pgn's in the internet to make an automated check of several positions. If there are no pgn's t find, i use this program https://docs.google.com/document/d/1W__MfTKrRTZsjEWZUPcFZ_2EdFYKBRstBf7wS-iwlvQ/edit as (little) aid to create fens from the scanned picture.
But without repetition i will forget sooner or later and this is real waste of time.
Thanks to anki i can keep informations "for ever" and its not too much work.
I wonder how much you still know from your chessimo (PCT) Strategy training. 6 repetitions might be enough to keep the information for ~ one year( i am afraid its less). I think its possible that you are now, several years later, at the ~ same level again. This work you have done might be ~~ lost.
http://gmsmirnov.com/what-obstructs-most-players-2/
Aox, this is really true.
ReplyDeleteLook at my current CT blitz rating. I stopped repeating my SR sets ("puzzles I failed")for about 4-6 weeks now. And my CT rating dropped. Co-incidence? First I thought: yes. (I had some excuses: the duplicate-avoid formula served me for 1-2 weeks much more difficult puzzles on average, but meanwhile it is back to normal, and I have no excuse and the drop of rating even accelerated.)
Since 3 days or so I started to do my scheduled puzzles again. Lots of them are waiting for me. I will need 1-2 weeks until I at the point were no puzzles are scheduled.
But I realized how much I had forgotten. My set-ratings fell sharply (every set has its own standard rating).
My feeling tells me, that as soon as I have finished all puzzles that are scheduled - that will be the day where my CT rating is back in the 1900's. (such as 1910 or so).
The level I had in September, and it did not feel so difficult for me to hold this level. But now it seems very far away for me.
Repetition is important, and even though I did these puzzles 5-7 times, it took just 4++ weeks forget many of them and experience a sharp drop in strength.
Smirnov does not say exactly how to study correct, but I started to post the patterns I learned, similarly as I would write down notes if I studied for an examination. I write down my knowledge, and try to sort similar patterns into certain categories.
I created new tags such as "Qeen+knight vs. king tactics" (I call that q&N_mate, but it isnt only for mates, but mate threats, too.).
Or "r&N" for rook and knight tactics.
Or "hook_mate". Or "take last square"-pattern.
Many new tags help me sorting puzzles. Let's see if that aids me in making these patterns "part of my personality" as Smirnov would describe it.
Tempo said: "Or, after f5, gxf5, gxf5 the h-pawn can become weak".
ReplyDeleteNo, that is not the reason. Because if f5, then white takes the e5 pawn.
If you miss, that f5 is not possible, of course g4 makes you wonder.
But now that you know it, you see, that g4 threatens g5. And after the g-pawn arrived at g5, f5 is still not possible, and f6xg5 wins the g5 pawn, but loses the e5 pawn.
Black could play e5-e4 and attack the white knight, but I guess this is all it is about: provoking e5-e4 so the knight can find an active spot in the center. For instance it jumps to d4.
So what can black do against the threat g4-g5?
He could play h7-h6, white plays h2-h4. I cant see it all, and I would not see how all that continues.
And to be honest, I would not have seen g3-g4, but I can see "afterwards", that the answer f5 loses the e5 pawn, and that g3-g4 is hence attacking the kingswing, and might help the f3-knight to become more active, or to create weaknesses in blacks kingswing in one or the other way. The move I would have chosen had been: doubling the rooks on the c-file.
But I am a sucker and silly. What do I know?!