Howto analyze a game

Well since it's all very new to me I hope someone here can help me out.

 I'm playing an online game here on this site against Chess31.

This is my very first game against a human so I'll probably loose the game but that's a part of learning right ?

Now here is my question :

How do I enter this game into Fritz or Chessbase Light (or whatever) to analyze it after the game is played ?

And what would be the best tool to analyze the game ?

Do I just "replay" the game in Fritz and then have fritz analyze it or do I enter it in Chessbase or can I just download a PGN file after the game and load that into a tool ?

 So, I think for me to learn the most is to have that tool tell me which moves were bad moves (probably all of them ) and maybe why and offer a better option (I think I've read somewhere that it's possible to have these tools offer alternate moves).

Thanks

ray_fields> So how useful is going over games and learning from the computer recommendations? Do you really learn what to do by analyzing games like that?

You learn more if you write out your own analysis before running the engine or looking up relevant master games. [If you played at blitz time controls, this analysis may take awhile. If you played at slow time controls, it may simply be recording your in-game thoughts.] 

Eg, 8...Bc8? "8...Nc6! would be better because it also prevents Bxb7/Bxa8 but has the advantage of developing a piece (rather than undeveloping one)." or "White's threat is Qf7 followed by Qxf8. The only real defense is ...Qd7!"

When you run the engine, look at the best move(s) but also try to understand why the moves you selected were wrong, and why you missed better (or equally good) alternatives This is the key to improving--finding out why you made the mistakes so you can eliminate them. If you just look at the best move and go, "uh-huh" you're not learning much, especially if it happens after the opening (since it's unlikely you'll reach the same middlegame position twice--at least, not in this stage of your chess development).

In this way, yes, it will help you improve.

 Thanks for that explenation.

The only thing I kinde have a problem with is "write your own analasys"I mean, I could probably find bad moves and write that down but I guess for a beginner finding the difference between a Bad move and Very Bad move is not that easy. Also finding the differene between a good move and excelent move (not expect to make many of these any time soon though).

I'm now running Shredder in Analyse mode on an online game I played against a buddy but the output is kinda cryptic to me :

28.b4 Th8 -20.32/2 0s
 [28...Df1+ -M3/3 0s]

29.a4 Txh5 -14.40/1 0s
 [29...Df1+ M3/3 0s]

30.Ta3 Df1+ 24.36/0 0s
 [30...Th3+ M3/3 0s]

31.Te2 Th3+ 32.Kc4 Pd6+ 33. Kc5 b6#

Now I guess the stuff between the [ ]are alternate (better moves/options) but the "-20.32/2 0s" I don't understand

Oh, and I guess D = Queen, T = Rook, P = Knight (N) (The GUI is set to dutch so I guess it uses dutch annotation instead of the English ones).

 Yes, you're right, it will be difficult at first. But the point is that you will get better and better at doing it the more you practice. At first you should at least be able to look over your games and find some mistakes, both for you and for your opponent. You should be able to see some things like, "I could have taken his bishop here but I didn't notice". The computer analysis should help you by pointing out other mistakes. If you take the time to look through what the computer suggests, and understand why your move was wrong and the suggested move was better, over time your analysis skills will improve and your playing strength will improve too.

As for the cryptic computer output, I'm not 100% sure but I'll take an educated guess (and hopefully someone else can correct me if I'm wrong).

"28...Df1+ -M3/3 0s" means that if black had played queen to f1 check, then it would be mate in 3 for black (that's the 28...Df1+ -M3 part); "/3" means (I think) that shredder confirmed this by searching the position 3 moves deep (which is all it had to do, since it's a forced mate in 3) and "0s" means shredder took (almost) no time at all to see this.

"-20.32/2 0s" means that shredder thinks black is winning by the equivalent of 20 pawns (that's the "-20"); "32/2" means that shredder confirmed this by searching all 32 legal moves in this situation, 2 moves deep, and "0s" again means shredder did this in a negligible time.

Hope that helps. 

Here's what IM Jennifer Shahade writes on using Fritz:

"It's always a complex question for me when I'm analyzing a game or an opening: When to turn on Fritz? I'm tempted right away, but as soon as the engine purrs, I stop thinking and my eyes glaze over... mesmerized by the crunching of numbers on the bottom left corner of my screen. A frequent problem is that Fritz will often choose a weird move as the main variation while a normal-looking, but losing, move doesn't even show up on the screen. What this means for me is if I overuse Fritz in analyzing, I'll misunderstand the logic of a game. Much of using Fritz well is knowing when to turn it off. (or ignore it) "

 Well I need to re-read your reply a few times and compair it with some more output from Shredder I guess but in time I think I will learn to understand better what it means.

I'll also try to follow your advise and see if I can analyze the game myself and then see what the Engine tells me

 Is that the Fritz Engine ?

I've been reading somewhere else that using the stongest engine (Rybka) is "the best" thing to do if you want to learn by analyzing.

But maybe Rybka and most (if not all) other Engines are also not that logic as Fritz.

cloudywizzard> I've been reading somewhere else that using the stongest engine (Rybka) is "the best" thing to do if you want to learn by analyzing.

Rybka (3128) is much stronger than Fritz 11 (2948). On equal systems, Rybka scored a phenomenal +24 -1 =27 against Fritz 11. But on a practical level, you describe yourself as a "real beginner", so the difference between the evaluations of the two engines is probably not going to be a significant factor for awhile. And by then there will probably be a new King of the Hill anyway.

cloudywizzard> But maybe Rybka and most (if not all) other Engines are also not that logic as Fritz.

Yes, her logic would be the same with all modern and probably future engines.

This is a position I recently analyzed from the chess.com vs cheater_1 game. Look at it with your eyes and find a couple good moves for White.

One engine says the top moves are Rxa4, Ra2, and Bxg7. But you would be remiss in your analysis (and show a lack of understanding of the position) if you didn't seriously consider Kc6!? threatening to promote the d-pawn. The threat is so serious, in fact, it forces Black to sacrifice a piece to stay in the game.

What moves did you come up with? You don't have to tell me. Humans at every level of play often see things that the displayed lines from Fritz or Rybka or Hiarcs never show you.

Fritz will show you tactics that you (and your opponent) missed. It's great for that. In terms of actually improving your play... Improvement in chess comes at the price of hard work, there's no shortcuts that I know of... Fritz will help you answer questions-- is such and such a good move? -- but you've still got to put the mental effort into calculating variations, searching for good moves, puzzling over the best way to continue, etc.   The computer doesn't teach, you use the computer to teach yourself. There are many many creative ways to do this. For example, It's an awesome way to learn _technique_ ... set up a theoretically winning postion, and then see if you can actually _get_ the win, playing against the computer. If I don't understand a grandmaster resignation, I'll set up the pieces from the where a grandmaster resigned... I take the winning side and try to get the win... it doesn't always work out like it's supposed to ;) 

you can play thru a game, move by move, watching the computer give evaluations to each move and suggesting it's own preferred moves... this is an excellent way to review a game.  

Setting Fritz on "full analyze" with verbose comments on will give good feedback.  It'll generate alternate lines and comment on the game... but it's also the laziest method. You don't have to do any work to get the analysis... and as I said, honestly: no work no reward...  I strongly reccommend as much interactive study with the computer as possible.  

When I think of things that have helped me grow as a chess player, I definitely include computer chess on the list, along with: playing slow chess, solving difficult puzzles, studying master games.

That's my experience (and obviously I'm not talking about computer tutorials.)

For a beginner, crafty's "annotate" command [with 390-odd MB hash - it can't take more unless you alter source code (it is open-source) - and 80MB pawn-structure hash, 30 min/move on a pentium-4 -- dualcore is a con - efficient parallel programming is tough] could be a better option because : crafty gives the line after each move and its suggestion on that mov; stores it in a file; you have to give it a "margin" parameter - say if it is 0.2 then if its move evaluates to within 20 centipawns of your move it doesnt comment on it.

And you can let pgn file say x.pgn with lots of  of games in it and let it run --- it will create x.pgn.can with the annotated games. [if your machine is overheating put it on hibernate once in a while!]

Thanks all for the advise.

 I'll try to follow the sugestions and try to analyze manually before letting a chess engine look for other options.

I am an old-fashioned human player. My personal CPU is not as strong as computer's, but has some special qualities. A friend of mine, a women who train martial arts, say: "I like watch hevy-weight boxers, but I can't learn how to fight from them". I think it's an analogy. Hulks of 100 kgs move and strike different then a women (or men) of 60, as like as logics of human and computer programs are different. (They analyse - we also sintetise! They calculate - we also imagine! )

I publish a series of analyses of my games here in forums. One day, I will maybe use a program to improve my analysis.

F.e. look this possition:

An interesting queens endgame

I spent about two haurs to analyse the position. At the end, i decided that the best move, in a strugle for a draw, is unexpected at the first glance 23.... Qe6!? I made a draw, and I am Interesting to see what a program can say about it.

And the second diagram in the article too, when I decided not to take a pawn!

In the game Self-Proclaimed Best Computer Player? (still in progress) it's possible that black, cheater_1, using 3 different programs to chose a move, made a rough mistake in 35th move, which lead him from simple draw to a defeat. We wil see (we wait the 38th move of white).

I am glad to see people discuss about computers (i.e. programs) as simple tools we use! Not "masters" or "servants". I wrote two articles about it in a topicHow close are we to solving chess? , and re-published them at my blog: Philosophy about Chess, Computers and God

lydia,

I intend to analyse that position in your game, using programs as tools not masters :) -  in fact i noted it down as soon as you posted it.

The delay is because at present my computer is busy with vote-chess "positionals" vs. "tacticals" [i am in the tacticals: i convinced people to go for gruenfeld - white transposed into KID Pertosian system.] and a few games - i am playing correspondence chess after about 30 years!

In the team forum of that game I  have often quoted your statement about vote-chess in your last weekly blog. You can see it after the game is over. So much political campaigning is needed! I am mainly doing the political campaigning :) But i think i will not play vote chess after this.