Tuesday, February 10, 2009

Chess exam

  • Overall rating 1768
  • Opening 1920
  • Attack 1902
  • Strategy 1839
  • Defense 1826
  • Middlegame 1814
  • Tactics 1759
  • Sacrifice 1710
  • Endgame 1691
  • Standard positions 1684
  • Recognizing threats 1575
  • Counterattack 1534
  • Calculation 1319
Well, that is about the opposite of what I expected!
Painful, of course, as it always is to lose an illusion.
But quite interesting too!

So all the circles and tactical training didn't effect my calculation skills. At all.

At least it explains where my latest boost in rating originates (see ratinggraph)

After long plateuing I finally found the way up again. That is the immediate result of changing my opening repertoire from gambits to positional openings. From heavy duty-calculation to less calculation intensive. The lack of calculation skills explained why I always got in time trouble with gambits. By playing simpler chess I get not in time trouble anymore hence I win from lesser rated players more often hence my rating improves lately. I wrote about simple chess here.

In the mean time it becomes clear that I cannot expect all that much from my positional approach. At Corus I noticed that there is still room for improvement by following the positional road, maybe another 100 points or so. But to make any serious improvement I will have to adress my calculation weakness.

Sigh. . . well at least I now can avoid a whole bunch of methods that don't effect calculation skills.

First I must know what Khmelnitsky means with calculation. I will study the questions again that are related to calculation and that I missed. In order to find out what they have in common and what my problem really is.


  1. Who would've guessed, from reading your blog, that you're an expert at openings and only so-so at calculation? I guess that's the value of external evaluation. Hopefully this will help you focus on the right things to improve.

  2. I wouldn't have guessed it. And yes, you need someone from outside to show you.

    Who would have guessed that the calculation muscle stuff in DLM's articles is nonsense?

  3. Well perhaps it isn't nonsense for everyone. Maybe you have more strategical potential than calculation potential. For somebody else, like his de la Mazaness himself it might be the other way around and then it could be more rewarding to try to develop this calculation potential than in your case.

  4. "During the first four circles... you will be improving your calculation ability. During the last three circles... you will be improving your pattern recognition ability."

    "[S]pend no more than five minutes trying to find the first move and no more than an additional five minutes working out all of the variations."

    Tempo, while you did solve a huge and noteworthy number of positions, it's worth pointing out that you didn't follow De La Maza's method too closely.

    First, you didn't "circle" through the same exact problems seven time. This likely affects how many patterns you internalized, but shouldn't hinder the acquisition of skill calculating.

    Second, you chose a "10s/problem" tool rather than the upto 5 to 10 min per problem De La Maza recommends. Intuitively, the easier the problem set the more it focuses on pattern recognition and the less calculation.

    Within that tool, you were in the "rating" rather than the "accuracy" camp. Sidestepping which is better, being 90% or 95% accurate requires more calculation than being 80% accurate.

    What I see is you spent 5-10s/position to solve 70,000 random tactical positions at 80% accuracy and that didn't improve your calculation.

  5. I'll add that Stoyko exercises where you spend many minutes per position are probably the most suited to calculation practice, and asking a coach to review the way you look at a few positions and suggest refinements to your calculation thought process may help.

  6. Papablanca,
    that is an interesting hypothesis. Let's stretch it a bit into absurdum to see what the consequences are. If my (lack of) calculation potential is innate, than it wouldn't make a difference which method of improvement I will use, since none can work.

    I don't need an opinion on this topic, since I'm going to find out. The problem is clear now.

    If I do find a method that will improve my calculation skills, I have proven that DLM's method is a lousy method to do the same.

    I can add to this "for me", but since this is my blog that speaks for itself.

    And then I have proven that it isn't necessarily an innate thing but that the right training can replace that to a certain degree.

    Further I can think of some kind reasoning how DLM could gain 600 points without improving his calculation skills.

    Everybody who is plateauting at 1500 can gain 250 points by intensive exposure to tactics alone. (I'm the proof of that). Everybody who did that can gain another 150 points by playing and studying much due to positional insight. (I'm the prove of that). That leaves another 200 points to explain by a rare yet not uncommon statistical aberration.

    What I'm trying to say is: to me all options are still open. I simply don't know.

    Maybe that is even the base of my calculation problem: I simply see always too many options. Leading to a short term memory overload.

  7. LF,
    let's be a bit more accurate:)

    I did the 7 circles exactly as DLM prescribed with 1359 problems of George Renko's CD intensive course tactics I. Which has about the same difficulty as CTart. With the exception that I didn't do all 1359 problems in one day, since chess study must be a joy. So I did it in two days.

    I followed the same approach (7 circles) with all 500 problems of the steps-method (step 3-5). Those problems are somewhat easier.

    I did all other CD's of Renko only once, often using 1-10 hours per problem.

    I did 74000 lowlevel problems at CTS (about 7 times the same dataset) with an accuracy of 80% and I did another 3000 or so as Slowtempo with an accuracy of 97%.

    Further I followed every good advice under the sun.


    I could add "for me" to it, but then again, this is my blog.

  8. "I did the 7 circles exactly as DLM prescribed with 1359 problems of George Renko's CD intensive course tactics I. Which has about the same difficulty as CTart. With the exception that I didn't do all 1359 problems in one day, since chess study must be a joy. So I did it in two days."

    I was referring to your 70,000 on CTS only (I met you while you were working on those) and wasn't aware of some of the work you had done before:


    So solving tactics (easy, medium, or hard) with or without repetition has not made you strong at calculation. You thought it did but now the exam saying otherwise has got you thinking.

    You have lots of practice. There must be a flaw in your technique. If you figure out a solution please share--my coach showed me some trick to improve my calculation skills but I still have a long way to go in that area.

  9. So solving tactics (easy, medium, or hard) with or without repetition has not made you strong at calculation. You thought it did but now the exam saying otherwise has got you thinking.

    I always was aware that the circles didn't work. I made quite some effort to proof that they didn't, in order to exclude every misunderstanding about it for posterity. The circles have made me strong in attack, though. From there I became better in defense and strategy.

    The first task now is to define what exactly calculation is. That's what my next post(s) will be about.

  10. Aagaard's book on calculation has great exercises for Styoko.

  11. I used to think your calculation probably wasn't on the level you thought, but for all the wrong reasons. because I simply forgot how much other tactics you had done in addition to CTS. it seems very likely now that you must be at least close to truth with your conclusion about DLM & calculation.

    very interesting this new development. and it makes me think maybe my main (or at least very big) problem is calculation as well, as I'm pretty sure I'm far worse at it than you. certainly it must be a bigger problem than I had thought. -it just doesn't become an issue in CC because of the analysis board.

    man, there are just so many huge problems to fix in my chess. and none of them can be fixed easily or fast... my worst fear about this calculation problem, is that maybe it turns out the kotov method actually IS the way to go ahead! and I don't know whether I can deal with that option. :)

    one thing about DLM & calculation: he played a huge number of slow games during his improvement phase. my understanding is that your tournament load has been significantly smaller? so maybe that's what complemented his lack of calculation training from his tactics diet? maybe he experienced exactly the same effect as you, getting much better at attacking because of tactical drills, but in reality it was the mass of slow games which gave him the missing piece of calculation for the puzzle?

    this makes me think about dropping radio commentary off from watching tournament games online, as it tends to steer me off from calculation. because I'm quite sure I won't start playing slow games myself, at least no in a sufficient quantity, but the top games do interest me enough to stay on it. also the usual 'study master games' method doesn't have the forced snail pace of real time games, and I invariably end up close to blitzing through them. yeah, I think I'll try this for a while, starting from the upcoming topalov-kamsky match.

  12. I took that same exam a couple years back: http://blunderprone.blogspot.com/2006/11/had-my-head-examined.html

    I think it's time I revisit this. It showed me areas of weakness I would never have guessed as well.

  13. I should elaborate a little as well. After completing the circles, I also took the exam and much to my dismay, I was low in Tactics, Calculations, Threats and Openings ( which the later didn’t bug me as much).

    First, to credit MDLMadness in its fundamental concept of rapid immersion of pattern recognition of tactical ideas. I think its good exposure for a patzer to go through some level of exposure of a great number of tactical patterns to get an idea of what is beyond the comfort zone of current knowledge.

    But a patzer shouldn’t live on MDLM alone. Rather, the exposure above should be used more like compass as to what to focus on for tactical training in your own games. Then, the most bang for the buck is to study tactical positions that most likely come about in your games… a topic in itself of a different horse.

    Now, to critique the MDLMadness, the method does not develop calculations nor does it help you synthesize tactics when they don’t exist. Thus the low score I got in Calculations, Threats and Tactics, if I recall, the problems were not all that clear and required more synthesis rather than pattern recognition. I think this is the fundamental key. Static recognition of patterns can only go so far, but in real games, if you lack the creative ability and calculating power to turn a rather benign position into a tactical marvell then all the thousands of pattern recognition drills will never amount to anything.

    There is a link, I think, with strategic positional planning and synthesis of tactics. You have strong positional sense ( I scored well in that area as well). I am learning to draw from that strength by studying a little deeper in analysis ( proper positional evaluation for instance) in order to recognize placement of the pieces and when to recognize “pre-tactical” opportunities.

  14. Soltis defines calcuation as the ability to consider various possibilities a few moves ahead, see the resulting position in your head, and evaluate it relative to the current position.

    In my mind, this is a proxy for your thought process, something you do every move and encompasses everything in chess.

    We can break calculation down into major categories like Candidate move generation, the actual process of analysis, and positional evalutation.

    I still think the circles is a useful program, but it only aids in certain aspects of calculation. Building tactical pattern recognition helps you know WHEN to calculate, and WHAT to calculate, and gives you practice at HOW to calculate. All valuable, but not the whole enchilada.

    Other components of calucation such as positional candidate moves, or how to evaluate positions are not learned from tactical drills.

    Coach B

  15. High speed tactic repetitions with 80% dont improve the calculation ability. That sounds reasonable, you dont train calculating then. I would guess: low speed tacic training with "high" % should do. Whats about some visualisation training like CVT: http://chesseye.alexander-fleischer.de/o/ ?
    And by the way:Is the Chess Exam of Igor Khmelnitsky veryfied by anyone? All these suggestions of how to improve i find in the net are not checked by someone, just guesses by some individuals. The only? good example of veryfying is here : http://www.convekta.com/softscho/l1/lesson_3.html

  16. There is not much calculation going on below ELO 2000.
    That makes it so strange that Khmelnitsky assigns an ELO 1300
    to your calculation abilities. As if players with a similar
    strength to you do have significant better calculation abilities.
    Not so.

    Suppose someone with ELO 1800 can calculate very well
    how is he supposed to evaluate the position at the end of
    a calculated variation? Or how is he supposed to compare
    different long variations starting from the same position?
    If he could then his rating would be much higher.

    To calculate well one must also know how to prune.

    The Stoyko exercises, as Dan Heisman calls them, capture the idea well.
    Aagaard's book is for people seeking to improve calculation, but it is
    an advanced book. Imagination In Chess is probably also worthwile.
    And look at the paragraph headed "Chess Improvement" by Rune Vik-Hansen
    at http://www.chessbase.com/newsdetail.asp?newsid=5055 .
    Silman HTRYC3 also explains well how to train calculation and
    he rightly warns it is hard.

    Calculation is about finding variations deep and wide
    in positions which offer no directions. Where a lot has
    to be discovered primarily by calculation alone.
    No directions means no trainable patterns apply.

  17. Blue,
    before I embrace another improvement method I want to dive first in what calculation exactly is.

    I like the way Aagaard think. It is a good tip.

  18. Some interesting reading. I'm gonna give "the Stoyko exercises" a go. Besides the Aagaard book, any other material that's suitable for these exercises?

  19. WW,
    good to see that you grasp the main point immediately (DLM does't help your calculation:)

    Luckily calculation seems to be my only or main problem. The rest seems to be fairly balanced.

    CC - I wil talk about that later.

    My average workload is about 80 slow games a year, 100 CC games a 3days/move ("blitz" in CC terms:) during 10 years.
    We have been a year or two member of two clubs, which doubled the workload. Allthough not so compacted as DLM, but not too shabby either.

  20. BP,
    funny that our experiences have a few things in common. Thinking that we are good where we are bad and vice versa, the failing of the circles to adress calculation.

    When you write about the CK for instance it is clear to all of us that your strategical insight is pretty well developed.

  21. Coach B,
    Somehow the short term memory is engaged in too many tasks during calculation. Some tasks has to become automated to prevent the mind from overloading. As I experience it.

  22. Kindling,
    the figures of Khmelnitsky have a statistical base. It compares me with other players. I trust that it gives a good idea of my relative strength in the respective area's.

  23. To all:
    thanks for the enormous amount of reactions.

    Don't panic, I'm going to do first what I always do: to think for myself. Please bear with me and give me a few days.

  24. Fritz11 does provide a training mode dedicated to improve calculation skills. It is well suited to record the thought process during calculation and it would mix well with the other suggestions for training in the comments. Have you tried this already?

  25. Anon,
    no (I don't use Fritz) but I will certainly have a look at it.