I just finished the first round of going through Karsten Müller's DVD 2 on rook endgames, and copying all of the starting positions into a database. That's about 70 positions in total, although some of them are more educational, in the sense that refuting the incorrect tries are much more important than the correct line. But quite a lot of them are also 'trainable' against Fritz as it is, meaning the meat of the positions lie in the correct mainline (along which engines probably step as sparring opponents). Which is exactly what I'm gonna do next.
So now I have all the positions in a chessbase database, even the 'untrainable' ones for completeness' sake. That way it's easy to load them up into Fritz (tools/options/training/endgame training), and use them instead of the pathetic default positions in the endgame training module.
On the first round I also went through the videos with some thought, and feel like I already gained a good familiarity with all the basic principles and techniques. Now the hard part begins, which is drilling the positions over and over, until the philidors, lucenas, vancuras, karstedts, saavedras etc. come as a second nature to me. I want to be able to blitz through all of it on autopilot if/when necessary, no matter how drunk or tired I might be. And at that point I'll attack the more discussional type of examples again, hopefully being able to spot the technical refutations and sidelines a mile away, as I should in a real game.
Quite a lot of the won lines end into a Q vs R endgame though, so maybe I should learn that first (it's on DVD 3, which I have but haven't gone through), in order to drill those positions until the bitter end? Sounds like a practical way to combine two drills, so maybe I'll just do that. it can't be that hard, can it? Although I did try it from the top of my head, and there's absolutely no way I could figure it out on my own. Seems more tricky than the KNB mate, which makes it pretty funny that most people probably resign rather than even think about trading the queen for a rook.
Showing posts with label fritz. Show all posts
Showing posts with label fritz. Show all posts
Thursday, July 10, 2008
Monday, May 14, 2007
Capture Blindness - A Horror Story
I've been thinking a lot about why I'm so terrible at blitz, when my slow games are going better all the time. What is it that I do so differently in blitz? Well one thing is extremely obvious: I miss a lot of simple 1-movers. Meaning I drop material in every game, sometimes even missing obvious mate-in-ones. I don't miss those almost ever in slow games anymore.
One thing I've come to suspect is my thought process in blitz. Do I really check if a move drops a piece? Not really. I just make a move that looks intuitively good. There's some checking of course, but it's more unconscious, and obviously not very efficient judging by my score.
So this got me thinking about how to exercise my blitz thought process. I could obviously do it on CTS, or even in real games. But somehow that gives me the nagging feeling that it's just my laziness trying to get way without dealing with any real, and inherently uncomfortable, change. "Just carry on dude, it'll all blow over in time." - Then I suddenly remembered someone mentioning some kind of a capture training mode in fritz, so I fired it up and took a peek at the thing.
It's called 'attack training', and the basic idea is to click every piece that can be taken, regardless of color or the result. If the capture is legal, click the target. You can set a time limit for a session, and fritz automatically selects new positions from the database, as soon as you've found all the captures correctly. When the time runs out, it displays your number of found captures in that session, the average score between sessions, and the highest score. "Sounds like child's play, this must be so easy it can't possibly be useful" I thought. "There's not even looking for the simplest tactics involved, just captures."
So I set the session timer for three minutes and started clicking. And after a few sessions, it turns out my average scores were horrible!!! There was always a simple capture or two escaping my eye, and finding those simply took a lot of time! My average score for 3 minutes was 10 captures found!?! That just must be incredibly weak. I also found it surprisingly hard to concentrate on finding my opponent's captures in an orderly fashion. Which indicates that I indeed have the bad habit of just ignoring my opponent's threats.
So it seems like there's yet another element of board vision I wasn't aware of. One which is possible to disregard even while doing a huge amount of tactics. And even though it can be argued that most of these captures are irrelevant, as they simply don't have any possibility to work out, I still think I should be able to find them fast or immediately. There's clearly a lot of work to be done on that area, and I think this weakness might actually be the very reason why I play so badly in blitz.
From now on, I'm going to do a couple of 3 minute sessions daily, also focusing on the thought process I use. Nothing complex, just conduct the search in orderly and complete fashion. First search for my opponent's captures, then my own.
I would also be extremely interested in how others do on the attack training mode. So please try it out with a batch of 3min sessions, and leave your result as a comment on this post, complete with your blitz rating. I'm hoping we'll see some kind of meaningful differences between blitzers of different level.
my stats:
3min sessions
average: 10
highest 17
ICC 5-minute 1100
update: I've just been told that the 'attack training' mode wasn't introduced until Fritz 9. I'd still love to hear about it from those of you who have it. Whether you do well or not doesn't matter, I'd just like to find out if seeing those captures faster is a property that good blitzers have. A bad blitzer doing well on it would also be very interesting, as it would pretty much prove me wrong and close the case as irrelevant to blitz training.
One thing I've come to suspect is my thought process in blitz. Do I really check if a move drops a piece? Not really. I just make a move that looks intuitively good. There's some checking of course, but it's more unconscious, and obviously not very efficient judging by my score.
So this got me thinking about how to exercise my blitz thought process. I could obviously do it on CTS, or even in real games. But somehow that gives me the nagging feeling that it's just my laziness trying to get way without dealing with any real, and inherently uncomfortable, change. "Just carry on dude, it'll all blow over in time." - Then I suddenly remembered someone mentioning some kind of a capture training mode in fritz, so I fired it up and took a peek at the thing.
It's called 'attack training', and the basic idea is to click every piece that can be taken, regardless of color or the result. If the capture is legal, click the target. You can set a time limit for a session, and fritz automatically selects new positions from the database, as soon as you've found all the captures correctly. When the time runs out, it displays your number of found captures in that session, the average score between sessions, and the highest score. "Sounds like child's play, this must be so easy it can't possibly be useful" I thought. "There's not even looking for the simplest tactics involved, just captures."
So I set the session timer for three minutes and started clicking. And after a few sessions, it turns out my average scores were horrible!!! There was always a simple capture or two escaping my eye, and finding those simply took a lot of time! My average score for 3 minutes was 10 captures found!?! That just must be incredibly weak. I also found it surprisingly hard to concentrate on finding my opponent's captures in an orderly fashion. Which indicates that I indeed have the bad habit of just ignoring my opponent's threats.
So it seems like there's yet another element of board vision I wasn't aware of. One which is possible to disregard even while doing a huge amount of tactics. And even though it can be argued that most of these captures are irrelevant, as they simply don't have any possibility to work out, I still think I should be able to find them fast or immediately. There's clearly a lot of work to be done on that area, and I think this weakness might actually be the very reason why I play so badly in blitz.
From now on, I'm going to do a couple of 3 minute sessions daily, also focusing on the thought process I use. Nothing complex, just conduct the search in orderly and complete fashion. First search for my opponent's captures, then my own.
I would also be extremely interested in how others do on the attack training mode. So please try it out with a batch of 3min sessions, and leave your result as a comment on this post, complete with your blitz rating. I'm hoping we'll see some kind of meaningful differences between blitzers of different level.
my stats:
3min sessions
average: 10
highest 17
ICC 5-minute 1100
update: I've just been told that the 'attack training' mode wasn't introduced until Fritz 9. I'd still love to hear about it from those of you who have it. Whether you do well or not doesn't matter, I'd just like to find out if seeing those captures faster is a property that good blitzers have. A bad blitzer doing well on it would also be very interesting, as it would pretty much prove me wrong and close the case as irrelevant to blitz training.
Subscribe to:
Posts (Atom)
