How to Use Computers for Post-Mortem Analysis

Submitted by sam_musil on Tue, 09/22/2009 at 11:05am.

Here is a series of notes from Whitesox7 that I think will be helpful to all of my students:

===============================================================

Original Message by WhiteSox7 on 9/19/2009 @ 10:49pm:

Hey Sam, i had just looked at one of your games that you annotated, may i ask..how did you get so good at analyzing positions so well? Is it from studying many GM games? or what?

-----------------------------------------------------------------------------------------

Original Message by sam_musil on 9/20/2009 @ 11:13am:

In short, the answer is practice! I have been playing tournament chess for 35 years and teaching chess for several years. My chess mentor Israel Kutasov taught me to follow the openings of strong GM's to develop my style of play, but then to analyze my own games in order to improve my middle-game play. Israel taught me the absolute importance of post-mortem analysis of my own games!  "Izzy" forced me to explain a reason for every move in my completed game that we were analyzing!

I started using Chessmaster, Fritz, and Sargon to help me study critical positions from my own completed games and became a "GM" in analytical ability just through learning the intricacies of my own openings and explaining the reasons to "Izzy." I love chess and teaching chess because it is truly a "search for the truth" about each position in a game.

If you keep reading my blog, your analytical ability will increase just by learning the critical aspects of certain opening positions. Opening books just scratch the surface. I demonstrate how to win with my openings!

Sincerely, Sam

=================================================================

Original Message by WhiteSox7 on 9/20/2009 @ 11:54am:

oh ok, thanks! but one more question if u dont mind, i actaully have fritz and chessmaster also, which i have used to help analyze my games, but how do u use the machines to their fullest so that i understand why they are suggesting such moves?  It recommends a move and some of the time i will understand the positional or tactical importance of the move, but other times i dont understand why it would suggest a move, is there any way RIGHT WAY that you should be using these machines?

----------------------------------------------------------------------

The right way to use these machines is to make them tell you the "top 3" moves in a key position.  I never look at just the recommended move as it is often a stupid "computer move" that is completely illogical!   If you disagree with the machine evaluation, you must make the machine show you its refutation to your line.  I often play ahead 20 to 25 moves to check out the two top alternative variations.  I only analyze a few positions from a losing game.  I look at moves that I was concerned about during the game referring to my game notes and at moves flagged by the Chess.com analysis

If the master DB was still "in play" at the position of the key move that I am analyzing, I definitely check out the "next best" master move in Fritz to see the computer evaluation of that move. If there are more than 20 games in the database, I put more trust in the DB than in any engine.  With less than 5 games in the DB, the engine is more trustworthy!

The only winning games that I analyze are games where I was clearly lost early and won due to an opponent blunder!

Typically, I find two or three innovations to a completed game and always play the earliest improvement in my next game!  If I am unhappy with my position in that game, I will try a later improvement next.

Sincerely,  Sam Musil

----------------------------------------------------------------------

» posted in sam_musil's Blog
 

Comments:

by Metal-Gerd - 54 days ago
Oldenburg Germany
Member Since: Sep 2009
Member Points: 32

A completely new way of using a computer for post-motem analysis  is Aquarium's IDE which stands for interactiv deep analysis. It takes quite a long to time to fully understand this feature but it it wort to put some effort into it. Most of the work has to be done by the player himself but the programm will give a very good assistance in any given position.

by Chessroshi - 54 days ago
Indianapolis United States
Member Since: Aug 2007
Member Points: 739

Am I seeing this correctly that the posted games are just DB cough-ups? I didn't really see any instructional value in what a DB finds as a higher percentage line. If this is what high-level chess is going to be on chess.com, then I want no part of it. What would truly help 'students' would be instruction on the reason(s) behind moves, other than the reason that the DB shows GM's like this move more. I guess the thing that rubs me the wrong way with the whole thing is the approach. A class player claiming TN's and 'perfect' games just seems a bit out of whack. If you honestly want to help people on this site, play some chess, make human mistakes like the rest of us, and summarize what you learned during the game. That to me would seem like a more humble coaching approach.

by RC_Woods - 55 days ago
Nijmegen Netherlands
Member Since: Jan 2008
Member Points: 537

@ philidor:

Just be patient. If anything increases your analytical ability it must be knowing what percentages are given to your moves by the Game Explorer. 'Understanding the position' is a vague term used by lesser chess players like Capablanca, Kasparov, Anand etc. Of course that is to be expected of players who rely on and write opening books that only scratch the surface.

It takes a Super-GM strenght analytical mind to understand that 20 chess players will practically always outplay Deep Rybka , no matter how many cores you run it on. The reason is, of course, that their moves will enter the big bad unbeatable database.

@ Prooz

You broke the main line, you should expect the reply to take three weeks. 

@ Sam

Your correspondence rating indicates you are a very good player. I do believe you spend a lot of time finding the right and interesting variations, and it is nice that you are willing to share knowledge. 

BUT

Your blog would be 10 times better if you dropped the bold italic arrogant overtone. Mostly you annotate games with just percentages. following your games would be 10 times more educational if you explained why moves are strong.

Most of this has already been said in response to earlier blogs. I think you should give critical replies more credit than you currently do.

This is the web, and people usually don't take much effort to deliver their feedback in a nice package. In fact, most remain silent. If I believed your blog had no potential to be really good I would join in with them. Seeing that I believe you really do have something to add, I just suggest for one last time to think about your blogging style. Annoying students is something no humble teacher should want. 

by PeterArt - 55 days ago
Luijk Belgium
Member Since: Feb 2009
Member Points: 569

haha now i get it..
So you discuss here when to use an engine or when to use a database, lol.
In a good chess program you can configure this, to use only good lines from a DB.
Unless the engine knows something better. You can do this with Arena, for example i once configured with a special customized DB i made to play Scandinavian opening only. Probably since you here are GM, you don't use Arena but other software, which most likely can do the same.
So in the end people use software (even a DB is software)

Still however using DB's or Engines.. doesn't improve your insights I'm afraid.
Your trusting others insights, and miss potential variations.

but oh wel last time i was reading GM's  buy good moves from specialists on particular openings so.. its a strange world.

 

Maybe nice to show, a position where people always seam to offer while computers most often retreat (look at the specific peaces in play only, the rest of the board my vary (as the situation often occurs). So if people don't play like it, whats the use of a computers in such analysis. Since it greatly impacts the later play.

by Prooz - 55 days ago
Netherlands
Member Since: Nov 2007
Member Points: 52

Sam, obviously you don't care about comments unless they're confirming your idea that you're a grandmaster. So I'm giving up on trying to give you advice.

I'll leave you with one last tip: stop calling yourself humble , 'cause if there is any adjective in the English language that does not apply to you it is that word.

by sam_musil - 56 days ago
Marysville, Kansas United States
Member Since: Jan 2008
Member Points: 268

DeepGreene ,

Of course, everyone can learn from analyzing their wins, but I wait to make some opponent break a main line from a variation before I try to change it!

It is a question of improving play and conserving time.  I spend up to three weeks on post-mortem analysis, if a main variation of mine is broken in a game. The time is spent checking both databases and engines!

I teach my students to use the master DB to choose their moves minimizing the white win rate with black and maximizing the white point rate (white win rate + 1/2 the white draw rate) with white !

I thought this note might help some of my students, but I will not blog suggestions like this again as many readers seem angry.

Your humble chess instructor,  Sam

by philidor_position - 56 days ago
United States
Member Since: May 2009
Member Points: 220

I have seen many under 1800 players analyze games much, much better than sam_musil on this site. It's quite interesting, since he is obviously a grandmaster strength player.

Oh wait, now I get it, they were probably his students ... right.

I can't wait till my "analytical abilities increase" by reading sam_musil's blog. It seems like a very slow process, but I know it will work. I only wonder if there were an easier way to get to 2600 on this site. Always wonder...

by rigamagician - 57 days ago
Toronto Canada
Member Since: Sep 2008
Member Points: 3805

Sargon is a computer program that was popular in the 1980's.  I believe you can download a MS-DOS version from abandonware-france .

Former world champion Mikhail Botvinnik used to recommend to Garry Kasparov that he analyze his wins with the same critical eye that we would his losses, looking for mistakes in his own play.

by PeterArt - 58 days ago
Luijk Belgium
Member Since: Feb 2009
Member Points: 569

..I never heard of Sargon, what software is this, do you have a link ?

by RC_Woods - 59 days ago
Nijmegen Netherlands
Member Since: Jan 2008
Member Points: 537

I have to second Prooz here. For me the overload of bold and italics isn't adding anything either. If you took a database of "GM rated" writers, you would see that they don't play bold and italics like that. It has a bad winning percentage.

by Prooz - 59 days ago
Netherlands
Member Since: Nov 2007
Member Points: 52

Sam, no offense, but your blogs usually show no sign at all of you using Chessmaster/Fritz/Sargon or any other engine in the post-mortem analysis. The only thing you seem to use in your analysis is the chess.com Master DB. Your blogs would be a lot more instructive if you also included the results from your computer analysis.

Just a tip: using bold and italic case in every sentence does not make the things you write more readable, quite the contrary. You would get the message across in plain text just fine.

Peter

by adamWheatley - 59 days ago
Maryland United States
Member Since: Apr 2008
Member Points: 57

How can you tell what the best move is based on a database? Is it the move most played? Or the move with the highest win percentage? Somewhere in between perhaps?

by DeepGreene - 59 days ago
Vancouver Canada
Member Since: Jan 2008
Member Points: 1227

Thanks, that's great advice.  I guess I'm pretty much following it, as far as method goes.  However, I'm not delving as deeply into the analysis as this advice would have me do. 

And sadly, I tend only to post-mort games I lost.  Of course, I'm not quite silly enough to believe that I don't have a lot to learn by analysing my wins too... it's more of a concession to the amount of time I have allotted for this sort of activity these days.  :-\

 

Add your comment:

Join Chess.com for free to add your comment! Already a member? Then login now to comment.