Saturday, November 04, 2006

Thrashing, Tweaking, Holes, Feedback

Note: I have been outed!


Not really, as I was never 'in', but there was a rumor going about that this blogger was the English GM Mark Hebden.

I wish. Well, at least the GM part...

Sorry to say, I am not him, but simply a USCF Class A player coming out of chess retirement using this blog as a vehicle to help other class players organize their thoughts on how to self-train in chess. I am probably the worst blitz player in the history of the game so take that for what it is worth. But, I am working on that as well...

I am making these various ideas available to the public in one place within an environment of free-exchanging ideas so we can bounce these ideas off one another to improve our methods of self-training and to be more productive with the time spent studying chess.

This is truly a chess-utopian environment!

Some of you might know what the term 'thrashing' means. Being a software engineer by trade, I certainly know what the act of thrashing is when doing any repeatable task and it is something you wish to avoid. 'Thrashing' in chess is doing "a little of this, some of that, more of this 'cuz it's fun, less of that 'cuz it's not fun...", and by adding structure to our training, we can avoid thrashing.

To improve, the class level player must

1) Play against stronger opposition consistently
2) identify flaws in your play by reviewing your losses with Better Players
3) fix your flaws so identified in #1
4) Don't Thrash

and since class players are generally weak in tactics (if you were strong in tactics but mediocre in other aspects of the game, you'd probably be rated somewhere between 2100-2200, which is tickling the Master-level), it makes sense that a daily regimen of tactical training will have the biggest bang for the buck. In fact, that is what you see in the 4-Day Generic Chess Training Schedule (GCTS) found elsewhere on this blog.

In the 4-day program you will find that tactical drills exist for 1 time unit on each day of training. Are tactics that important? Only you can answer that yourself. Look at your losses closely and examine why you lost the game. Rarely will you come away with a loss stating something like "Boy, that Knight on d5 really killed me." when in fact you overlooked a tactical threat that your opponent saw and won material and the game. So, tactics are the most important aspect of training we, AS CLASS PLAYERS, can do on a daily basis. A necessary evil. If you can confidently calculate tactical threats in a position, you will climb the rating scales slow and steadily, to a certain point.

Holes, Holes and More Holes
Studying tactics ALONE will leave huge holes in your chess game, however. Every class player needs to study tactics. They also need to address weak endgame play, wandering planless play in the middlegame, and superficial opening play as well. But we need to have a feedback system that tells us "Hey! Your openings are OK at the moment. Mayber concentrate on other aspects of your game and get those better?"

How do we do that? I have thought about this for a little bit lately and have come to the conclusion that this feedback system does not exist (within the chess self-training environment), so we will attempt to articulate it a little bit here.

Some necessary components need to be in place to assemble a feedback system and relate it to your training regimen:

1) Play on a Regular Basis ("PL" in the GCTS)
2) Closely examine ALL Losses (blunder?) and Draws (missed win?) with a better player. This entails annotation of your games by yourself objectively.
3) Keep simple records

Yuk. #3 scared me when I wrote it. I'm sure it gave you the eebie-geebies when you read it the first matter. Face what we fear and conquer it!

Tweaking the system to make it 'right for you' is a process that we can try to achieve over time. Over-tweaking is something to avoid. We do not want to alter the system to ther point where it no longer resembles that which it was.

Play on a Regular Basis
Play regularly, at least as regular as the basic 4-day schedule indicates. With online play (ICC, for example), and assuming you play G15's, you'll end up with at least 4 games a week. Play local tournaments/ quad events if possible, and play in the open section. If you are winning your games all the time, your competition is too weak - up it a full class. We want our games to test us over the board. I use the following formula for blitz games on ICC:

(rating > myrating + 200) & (rating <= myrating + 300) & established

This will give you games against players rated between 200 and 300 points above you, and they are established rated players. This formula moves as your rating moves, so it is self-correcting.

A word about ego
Some players will balk at this formula stating it is too difficult to play against players rated so much higher than you are, and it will hurt your rating. If you are one of those people, then you truly do not want to improve your chess game and your rating means too much to you. Did I tell you that you still look 21 years old? Or, do you want the truth that you look like a tired 30-something who smokes too much? your call.

We need games against stronger players to expose the weakness in our own play and to punish us directly for that weak play. Any other play is a waste of time and is a game you should win anyway. Save your ego massaging against Fritz or Crafty at home at the lower levels. You will, without a doubt, thank me for pushing you to play stronger opponents in the long run.

Examining Losses/Annotating your own games
This can be a painful experience because, undoubtedly, a better player will often say something like 'Wow, you missed this simple tactic here.' or 'there is a win if you played this' and this can shake the foundations of your chess ego terribly. But, it is cathartic in nature to go through this process and accept the constructive criticism like a man (or woman).

Take notes at every critical position in the game. Get the better players' objective view on your play in the game if possible, at a high level ('You played the opening well, the middlgame so-so, but your endings and tactical play really sucked. Work on that.'). If a better player is not available, use Fritz or similar software, or post your game to one of the chess usenet groups and ask for some objective evaluation of the game. Any comments you get will be beneficial. The single most important aspect of this is to determine at what phase of the game you played the weakest.

Note: a Chess Teacher is the BEST option to review your losses with if you can afford it, or possibly an annotation service. Google it. They are out there...

Our goal is to adjust our training schedule every 20 games or so based on our feedback we get from our losses.

How to tweak
Ok, so we are 20 games into our program, ans we have played reasonably, but the competition has been stiff in the open section, and we have garnered a record of +5=5-10, or 5 wins, 5 draws, 10 losses. We also noted that only two of the draws were thrown wins by us, so in effect, we have 10 losses and 2 draws to contend with. Out of those 12 games, we discover that in 7 of them, we got outplayed during the transition from middlegame to endgame and during the endgame. In 3 of the other 5, we missed a tactic (it happens). The other two games were flat draws.

So it is painfully obvious to us that our endgame and tactical play is wanting (this is probably what is true for most class players, by the way). Let's try to devise a way to make one adjustment to our training that will expose us to more endgame study.

Overall, we have 16 units, broken down:
Opening - 4 units
Strategy/Middlegame - 3 units
Tactics - 4 units
Endgames - 3 units
Play - 2 units

First, we never take away from PL. Second, we never exceed 4 units for any one aspect of our game.

In this example, I would reduce opening study by one unit and increase endgame study by one unit, resulting in the following breakdown:
Opening - 3 units
Strategy/Middlegame - 3 units
Tactics - 4 units
Endgames - 4 units
Play - 2 units

Taking our original schedule, we could adjust to this:
Day1 - SO2, VT1, SG1
Day2 - VE2, PL1, VT1
Day3 - SG1, VG1, PL1, VT1
Day4 - SO1, VE2, VT1

We would then continue to play another set of 20 games and re-evaluate after that.

I hope this outlines a very simple feedback method to enhance the current trainig methodology we are experimenting with today. As usual, any comments are welcome!

Good Luck!


jambu said...

Hello! Thanks again for more excellent ideas and advice on applying the Generic Chess Training Schedule. I started a training diary dedicated to tracking my progress with GCTS. First it was a notebook but I decided to make it a blog. It is my Jambu Journal and here is the link:

Edwin 'dutchdefence' Meyer said...

Hi there! Just added you to my sidebar. Feel free to add me to yours. I'll be following your posts. Take care!

Patrick said...

The process you describe is just about exactly what i did, although i did not map it out so scientifically.

1. one rated tourn. game every week.
2. post-game analysis with a 2100 player and my 1700-rated buddies.
3. self-annotation of all my games.
4. replaying complete well-annotated GM games with a triple-weighted set-- either in my openings or Zurich'53
5. at least 100 tactics puzzles per week
6. a tight opening repertoire, going for depth not breadth

I became much stronger (400+ points) after 4 or 5 months of this. Maybe she's born with it; maybe it's maybelline.

My blog post about "the importance of failure" tries to say the same thing you said about "ego". Most players avoid failure as an ego-protection device, when failure is the catalyst for growth. (eg, weightlifting). I wrote a post (Self Diagnosis) that mirrors what you said about "feedback".

More info at my blog sitemap. To ME personally, the coolest revelation is that all these self-improvement tricks work in real life. Financial savvy, competence in romantic relationships, physical health, and musical aptitude can all be developed in a similar process. I've blogged about that too. Anyway, this is becoming self-hyping spam so i better shut up. GL in chess and life.

Anonymous said...

In the revised schedule, Mark suggested changes on Day4 from:

SO2, VE1 ...


SO1, VE2 ...

This sounds plausible, but it either assumes that the player has had _theory_ in endgames, and is just rusty; that looking at endgame problems is an effective teaching tool; or that endgame theory isn't very complicated and doesn't need to be studied.

I'll dismiss the 3rd possibility out of hand; take a look at any analyzed game that involves a GM making an endgame error (that would be most of those played by GM's... ;-), and you will, too.

I'll deal with the 2nd next: Take a look at _any aspect_ of endgame theory, and you'll realize it's way too complicated to absorb from doing a few problems. In fact, some moves in the solutions will appear counterintuitive unless you know the theory.

So, back to first principles:

In general, every teaching rubric works on the principle that _most_ people learn fastest and mostest with the sequence:

"Theory, Example, Practice"

[This doesn't address the needs of those people who can only learn from trial and error, but you guys can go back to the schedule suggested by Mark, and do more problems. I'd plan on doing _a lot_ of problems...]

Back to the rest of us:

The SG ("study games") part can help with "Example" part of the rubric (when they include endgames). But where's the _theory_? Uh, it's not there...

So, I'd like to suggest including an endgame course in the study program. You can practically pick any one, from Pandolfini to Averbakh or Dvoretsky, because they'll all help. That means Day 4 becomes:

SO1, SE1, VE1, VT1.

Now the SE part is teaching _theory_, while the VE part is _practice_.

The other way to use the VE is to isolate the types of problem you are having the most trouble with (usually this means by types and amounts of material, though the distributino of pawns is also important). For example, if you do poorly at rook vs bishop and knight endgames, you can study those more.

One more thing: Different types of endgames occur with _very_ different frequency. So if you've never studied any theory, it makes sense initially to go through the theory for the key types of endgames once.

Here's how often they turn up (based on a DB of 600K games):

Type % Cumul. %
R + P 21% 21%
R + B v R + N 13% 34%
B + P v N + P 9% 43%
R + B v R + B= 7% 50%
Q + (B/N/R) v Q + (other pieces) 7% 57%
2R v 2R 6% 63%
R + N v R + N 6% 68%
N + P v N + P 4% 73%
P v P 4% 77%
Q + P v Q + P 4% 81%
R + B v R + B! 4% 85%
B + P v B= + P 4% 88%

"B=" means bishop of same color
"B!" means " " opposite color

"Wow!" I can hear you saying, "If I only study 4 kinds of endgame, I'll know the theory for half of the endgames I ever see!"

And three of them involve rooks, while three of them involve bishops... Now you also know which two pieces you need to learn the endgame intricacies of the most.

Notice that Q + (piece) v Q + (different piece) is also pretty common, but there are a lot of possible combinations, making this a long road to study completely. Give it a pass until you've finished Q + P v Q + P, when you'll understand the material better and save some time.

Hope that's helpful!


Mark said...

"SE" and "VE":

I think this is a great idea and a very useful tweak to the system to include and separate "SE" from "VE". I could be accused myself of making assumptions here on the way one trains endgames, as I did assume that students would use some sort of well-written guide like Dvoretsky's or Pandolfini. I personally use Dvoretsky, Mueller/Lamprecht or the Hooper/Euve books to train endgames with. All of these have a certain "theory" and "practice" paradigm to them. But, if you would like to make this distinction, then Jaxter's suggestion is very valid here, and I encourage this modification without question.

I assumed when you study openings you would study ones you are going to play. The same can be said for endgames - study those first that occur mostly in practice. Therefore, R+P, R+Minor Piece, and Minor Piece endings are the most frequent. You should ALWAYS include K+P endgames as well. My fault for not articulating that further, and thanks to Jaxter for noticing this ommission.

It should also be noted that endgame theory (opinion to follow), in my mind, can be broken down easier than, say, opening theory. Endgame theory is a means to an end: Opening theory is really not. Knowing "Botvinnik's Rule" on Knight endings makes one think first before exchanging Knights down to a K+P ending. Philidor's and Lucena's position shines a light to how to draw or win Rook Endings in their most basic form. One can become much better at endings by knowing a collection of some of these simple theoretical positions and rules, and avoid losing positions somewhat.

Naturally, knowing and practicing are two different things, and we hope that this very slight tweak to the system will help people organizationally and also result in some better-played endings by everyone using it.

Great Observations and thanks!