Friday, November 25, 2005

Pruning Lines

(Creating a Master File) brought us to the foot of the mountain, now we start our ascent.
Here’s where you will have the opportunity to spend a tremendous amount of time pruning lines from your master file if you so desire.

While it’s tempting to try and be prepared for as much stuff as possible you have to draw the line somewhere. We are trying to get our books down to a size that we can use for review. You can always add a line to your book at a later date if you encounter something new from one of your opponents.

This is where reviewing your games is crucial. (A completely separate subject and one for a much later date.) Making mistakes and finding new ground are all part of the game. We just want to try and avoid making the same mistakes repeatedly.

If you have a good reference book(s) on openings you may want to keep it/them near by.
(MCO, ECO, Nunn’s, BCO, individual books on a particular opening, etc.)

For this exercise I’m going to use CB9’s “Reference Feature” to do the majority of my decision making on the lines to keep. If you don’t have CB9 you will have to search through your personal library and hope that you get all of main lines that your opponent can throw at you.

We are only going to cover the first set of Black’s possible responses after 1.e4 e5 2. Nf3
The possibilities were shown in the previous post as what to expect. Once you see how this is done you will see how easy it is to repeat the process for each of the branches.

Ok fire up the recently created Bookup file we called “E4-E5 Work”.
Crank up CB9 (Chessbase 9), open a new board.

Enter the following moves into CB9 1.e4 e5 2.Nf3 hit the “Reference tab” and wait a few moments while CB9 compiles a list of games. (It will take a fair amount of time in the early lines it gets noticeably faster the deeper into each line.)
This is basically a reprint of my earlier post no need to reinvent the wheel. :)

“My database contains 286,513 games

The report tells me that Black has played the following:
2...Nc6 (237,048) 83%*
2...Nf6 (31,153) 11%
2...d6 (14,147) 5%
2...f5 (1,904) .006%
2...d5 (938) .003%
2...Qe7 (533) .001
2...Bc5 (318) .001

* These are percentages that I have added to help throw some perspective on what to expect at this point. (Times occurred divided by total games found will give you the percentage) Example 2...Nc6 237,048 divided by 286,513 = .827 or 83% rounded up.

And even more moves than I have shown, but the number of times those moves have been seen in tournament play lessens significantly the farther we get from the top of the list. Does 2...Qg5?? Really need to be prepped?”

Ok now tab over to Bookup, it will probably be easier to resize the program windows so that you can see both the “CB9 board with reference tab info” and Bookup, unless you are really good at using Alt +Tab and remembering lots of info. But around my house it is nearly impossible. Those with children understand. :)

Now when you look at the starting position of the Bookup book/file “E4-E5 Work” we created you will see some lines for White that show possible transpositions.

Example: 1 d4 & 1 Nf3
You can delete everything but 1 e4. The same goes for everything found for Black’s replies except for 1...e5.
Now do the same with White’s second moves, delete everything but 2. Nf3 this way our
Opening book will match the move order of our current CB9 board.
Now switch your attention back to our CB9 board w/reference report.
As you can see this is a ton of information to digest and it is now time to start pruning heavily.
Obviously based on the frequency of occurrence we can expect to see the following
replies from Black.

Just because
Haven’t occurred in a lot of GM games doesn’t mean we shouldn’t have something prepped. The Latvian and Elephant Gambits do happen quite often at the club level
and since this is our present fishbowl we might as well keep them in the mix.

So now I arrange and trim my “E4-E5 Work” file to match the report generated by CB9,
keeping just those 5 Black responses. I use the (Alt + U) to arrange the candidates in the same order as they appear in the CB9 reports. Make sure you set CB9 to prioritize the moves by number of times played. Simply click on the “Games” sub tab to make this happen.

Now in the CB9 “Ref Tab” selecting the top candidate (based on highest frequency) I click on 2...Nc6 and wait while it generates a new report.
It now gives me a long list of moves played by White.

I’m only interested in 3.Bc4 [The Italian or Guioco Piano (Pianissimo)], so I click on that move in the “Ref Tab”. While I’m waiting for a new report to be generated I toggle over to BU (Bookup “E4-E5 Work”) click on the move 2...Nc6, then I proceed to delete all of White’s 3rd moves except 3.Bc4.

Once this is done I tab back over to CB9 and see what goodies Black has in store for us.
This is where it starts to get interesting. Black has a variety of very solid responses with a veritable who’s who of GM supporters for each one of the moves played.

We could spend hours at this point trying to decide what to keep and what to toss.
Fish or cut bait?
Keep all of the moves played above a certain ELO?
Pick a hypothetical number of lines?
Calculate the frequency played percentages?

All of this depends on you and how much time you are willing to invest.
I’m after main lines, since I’m using CB9 in this case as my primary guide. I pretty much have to look at each group of responses independently and truncate the lines once I establish a noticeable drop in frequency of occurrence. It’s either do it this way or spend a lot of time searching through other resources.

Like I said earlier you can always add lines at a later date if you face something new.
Our mission is to create an operational opening book that allows us to practice our new found knowledge. We aren’t trying to become a theoretical expert on any particular opening. We are striving for solid lines that allow us to direct matters into areas we understand.

For the sake of brevity I keep the top 4 responses for Black, and adjust my BU file accordingly.

I now click on the move 3...Bc5 in CB and allow it to proceed with a new report, while waiting I make that same selection on my BU board.

White’s 4th Move

Here is where knowing what you want to play saves you a lot of time.

Since I’m creating this book from White’s perspective I only want one move in my BU file in any of my openings. (Why add extra burden and additional study time, just having to worry about Black is enough work already.)

I know that I want to continue with 4.c3. So I click on that move in CB and while I’m waiting for the next report to finish I return to BU to trim away all moves except 4.c3 as White.

So far in this line we have played the following moves.

1 e4 e5
2 Nf3 Nc6
3 Bc4 Bc5
4 c3

Black’s 4th Moves

Now there is a considerable drop in the diversity of responses from Black.
I’m prepping for the following:

Continuing with my trend I update my BU file to match this move order, and select 4...Nf6 in CB to generate a new report.
In case you haven’t figured it out by now at some point we will have to return to
cover the other branches. By making these adjustments now I save time when I come back through here again.
It also makes it easier to take the highest move in the first pass, work through it until we hit a certain depth. Then back up one ply to solve each of the sub-variations.

White’s 5th Move

5. d4
Once again I enter this move into CB9 first and trim lines out of BU while I wait for a new report. (It doesn’t take long for a new report to be generated once you get past move 3.)

I think you begin to get the idea, nothing fancy just a lot of work.

The main trick is to remember where you stopped, usually once I complete a line I will mark it in Bookup with a red, green, and yellow color code that I can later go back and remove under the commands option.

The other trick is to decide how far to take these lines. It all depends on your opponents
and the particular opening you are researching. You can always go further with your knowledge, because these opening books are something you can keep and modify.

Hope this has been helpful!
If you have any tips or suggestions please don't hesitate to add them here.


Pawnsensei said...

Hey Sancho. I was working on my opening book over the weekend and I noticed that CB9 limits me to so many characters per line. Some of my lines are quite long before they hit a branch so they go beyond the character limit. Is there a way to increase this? It's not a big issue. If I can't do it then I'll just create a subkey to continue the line. Thanks.


Sancho Pawnza said...

Hmmm I'm not quite sure if I follow you on the characters per line limit?
Could you give me an example of where
you are inputting the information?
Is this under the "game data" entry?

Pawnsensei said...

Openings tab. If you add on a new key or subkey I usually type in the line like

1 e4 e5 2 Nf3 Nc6 3... etc.

CB9 puts a limit on how many characters you can type in each key.


Sancho Pawnza said...

I don't think there is a way to modify the number of character inputs allowed. But this shouldn't be a problem. Most of the key classifications are already broken down into subgroups.
Steve Lopez had an article on this a little while back.

But yes your idea of creating sub-keys is the way to go.
Since most lines don't really start until X amount of moves anyway.

Sancho Pawnza said...

Now that I think about it.
Your best resource for this very subject is in the Chessbase USA's
"Electronic T-Notes Archive"
To access these articles you just have to register with Chessbase USA, simply create a user profile.
He has some very good articles on covering just this subject. Creating Opening Key and Sub-Keys.

Pawnsensei said...

Ok, thanks. I'll go check out the archives.


Sancho Pawnza said...

Looking through those articles has given me a new idea. It will require some testing first.
But I'd like to see how involved it would be to create a series of mini opening books (i.e. *.ctg files) that I could force Fritz to use as its main opening book to practice against.
Off to the laboratory!

King of the Spill said...

Great post.