Friday, June 12, 2020

Building a training database


From my last post,  an intro to chess databases, a couple folks were eager to learn how I set up my training database.  I thought I’d share the steps I took and a little on how I use it.  A couple questions I asked my self when setting this up were:

 

  1. How well do I know my repertoire in as either Black or White?
  2. Which lines am I actually encountering in these openings I chose to play?

 

The second question is really important as training for ALL possible outcomes is like trying to boil the ocean. I suggest using the database to help whittle this down as the outliers will come but the intent will be to handle 80-90% of what you face and you can bolster your training with the sidelines as they come in post mortem analysis.

 

Finding the common lines played within your own games

Maybe in another post I will walk through some steps on how to build a repertoire data base. Chessbase has some tutorials. There are videos out there too. Last time I showed a simple step using the database’s statistical tools and coming up with a pareto chart of common openings based on the ECO labels. It provides a quick snap shot.  You can even filter your data base to games you played only with the Black or the White pieces.   Here’s that chart again for reference.  Note, B12 is a common line in my games with the black pieces.


 

You could also use the reporting functions in chessbase and Generate a Repertoire database of your own.  I’ll do a couple of short cuts here as I really want to spend this post on the training side.


  I generated a repertoire in both Black and White using this function in my games database and here’s a look at the Black repertoire:



Big surprise, B12 (Advanced Caro-Kann) is my common theme here.  How you get to the common lines played is up to you. Whether you use the statistical functions of the database or report generation, both are useful insights that can save you some time for preparing a more focused training database.

 

Finding Annotated games for the training database

If you made it this far, you now may  have an idea what lines are played most in your games and you are ready to start your collection.  The MegaBase … the 8million strong…. has some annotated games in it. You can filter the Megabase to only have annotated games and create a separate reference database of this collection as well.  I looked in this collection for ECO  B12 games and found a few but buyer beware. Some are in another language while some are “a couple of additional lines from someone’s chess engine analysis”. There are some “verbose” ones but it’s hard to understand if the audience is meant to impress the masters or appeal to us mortals.


I wanted something that appealed more to the simpleton that I am.

 

I’m a lazy man with moderate means

The everyman series of books have been my go-to resource for “patzer needs to learn an opening” because they are verbose and written not to impress higher rated friends but rather written so a drooling imbecile, such as myself,  can understand it.  If only there was a way to import these books into a chessbase.

Years ago, I got really into Mike Leahy’s Bookup database software (now called Chess Opening Wizard) where I would search pgn files for my repertoire and meticulously import them. I would manually pick my lines and enter the text from the books laboriously. That was a lot of work, I was …younger… but I needed a better system.

I found out that Everyman Chess has ebooks! Which means, not only can you get the book for an ebook reader, but for the cost of the book, you can get not only the PGN version, but the CBV version for chessbase too!  YAY! Take my money as I am done transcribing books into a training database.  For instance, here a couple books I picked up for the cost of a chess book.


 

The move by move series is wonderful. It sets up questions like “Why move the knight to d7 instead of c6?”  This is the level of stupidity I need and seek for building training.  Going back to the advanced Caro-Kann theme in my black repertoire, finding games from the ebook collection was relatively easy:


I could select these and copy/paste into the Black side training games.  When I do this, I am careful to edit the game data so I can read it like a training database and easily select the game to train on.  

In the example shown, I append the last name of the white player to Black player’s last name so it shows up as a hyphenated so-and-so vs what’s-his-face.  In the White last name field, I enter a label. Here I used C-K advanced 3…Bf5 4.Nc3.  Use a system that works for you.

Looking at my Black side training Games, I have a nice list I can see immediately and select where I might want to train.


 

From this list, I extracted and relabeled all my games from the various Everyman ebooks I (recently) picked up and I am very satisfied with the results.  Keeping this to  manageable short list of training games ( 30-50) means I can use this iteratively and go a little deeper each time.

 

Using the training sets, I can click on a selected game and use the Replay Training function in the game view and select which side I wish to train on.

So here’s the thing, because I play amateurs such as myself, knowing the line  to the 15th move order or more is a waste of time.  I might get to a known tabiya in about 30% of the games, but mostly it’s about understanding the first few moves enough so that I can avoid the crappy traps and the “anti-whatevers”  that other amateurs, thinking they are tricky, will throw my way. When you play at my level, it’s the wild wild west and anything goes. I like playing stronger players for the very reason I can get to a known tabiya but lose mostly to middle game stuff.  Rather, training into these lines no more than 10 moves deep and really understanding the verbose explanations within those first few moves is a good remedy to avoiding sucker punches.

After my games, I use this as a reference to see where I may have drifted off or my opponent played off the main variation. If the variation didn’t exist in my training, I will search back in the ebook first for any missed lines.  If that doesn’t exist, I will go back to my game and use the MegaBook I created and see if it’s in the book and what is the proper response.  I tend to make use of the Master reference database, Lower rated Amateur database and a Live Base offered by chessbase.

Well, I hope you found this useful or at least got a couple ideas of how to prepare better for your games. I will say, since most of the chess world has moved to online play, and that means mostly rapid games, my Blitz rating jumped 300 points since I started using this tool. Even though I am slow curmudgeon, I am learning to adapt to a quicker world.

 

Until next time,

-Blunderprone

 

 

 

 

 

1 comment:

LinuxGuy said...

Working out a repertoire 8 moves deep is quite decent, and I haven't really done so in a legit sort of systematic way, other than for in patch-and-play sort of way.

I'm on Linux, so I don't have Chessbase, and haven't used a database in years. Last time I used Chess Assistant, and the one that comes with the King or something, both databases would slow down my computer at the time, and eventually this proved too much for me. I've never tried Bookup.

I'm more interested in making my own sort of DB program some day, or maybe trying to use SCID again, but I rely more on an engine than the games of others. Still, databases are an interesting topic!