Analysing own games - what is needed?

Hi there!

I want to improve on analyzing my own games. Right now I have Arena and SCID with Houdini 1.5.

The way I'm analyzing my games now, is that I'm looking at positions where I was in doubt about the best move. Then I see what the engine would have done, but it doesn't really make me aware, if my actual move was any good. Usually I only see the engine's four best moves, and maybe my own move would have come fifth or sixth on the engine's list over best moves.

Maybe I don't need to know, if my actual move was fifth or sixth on the list - but at my current level (beginner) I think it's nice to know. If I shouldn't be concerned with the strength of my actual move, please tell me so.

I'd like to be able to do the following things (which I might already be able to in Arena/SCID without knowing how):

1. To be able to search for blunders in a played game
2. To be able to click on a specific move, and be told the strength of the move or if it's a blunder
3. To let the engine analyze the entire game, and afterwards see a list of the strength of the actual moves

Can I do this with Arena/SCID?

I consider buying Chessbase - since I'd like to study master games - but would I be able to do these analyzing tasks in Chessbase (I consider it more as a database where I can compare with master games)? Can Chessbase perform the same analyzing tasks as Fritz (the application)? Basically, do I need Fritz for analyzing, if I buy Chessbase?

I can address some of your points, I believe:

a) "If I shouldn't be concerned with the strength of my actual move, please tell me so."

Sure, you should be concerned with the strength of your actual move ... and you can discern this by comparing the evaluation by Houdini for your move versus the one it suggests is best ... if there's a large discrepancy (> 0.4 pawns), then you're move is surely not optimal, maybe not even good.

b) "I consider buying Chessbase - since I'd like to study master games - but would I be able to do these analyzing tasks in Chessbase (I consider it more as a database where I can compare with master games)? Can Chessbase perform the same analyzing tasks as Fritz (the application)? Basically, do I need Fritz for analyzing, if I buy Chessbase?"

ChessBase can use your Houdini 1.5 engine, but also comes with some version of a Fritz engine (probably not the very latest though). ChessBase can do a lot of analysis with either of these or other engines ... but it typically does not have all the analysis modes of Fritz the program. What I purchased was ChessBase and also the full Fritz program as well. ChessBase has a feature in it where you say "Go to Fritz" and it will move the game to analysis by Fritz with all its heuristics and commentary. This is the best of both worlds but it ... requires two purchases. It's been a while since I procured these ... maybe my data is old by now.

Best wishes.

@ DrawMaster

Thanks you very much for your thorough reply! Your info is very helpful.

If someone sees this post and knows that the new ChessBase can do some of the stuff that DrawMaster refers to (the Fritz stuff), please post in this thread. :)

@ riswang

I think it's best if you start a new thread and/or go watch this: http://www.chess.com/video/player/how-to-analyze-your-own-games-beginner

@ riswang

I can't really help you with that, since I'm a beginner, and I'm trying to improve the way I analyze my own games, but look at the video I linked to, and also these: http://www.chess.com/topic/how-to-analyze-your-own-games

Personally, I use chess engines to review my games for basic tactical oversights and because I don't have anything better to review my games with. That said, reviewing your games with an engine is no substitute for having a stronger, more knowledgeable player review it. While engines can be helpful in many ways, their usefulness is frankly limited (especially to newer players) and in some ways, their feedback can even be harmful.

In particular, an engine can indicate that you made an inferior move (assuming it even was inferior), but it cannot tell you why it was inferior. More often than not, the answer to the 'why' question is more important than knowing what the "correct" move was anyway.
Also, engines are notorious for horribly mis-evaluating certain positions. In particular, many do not understand "fortress"-type endgame positions where the stronger side cannot make progress. As an example, try plugging in a position where white has a king + 8 dark square bishops vs. black's lone king. Turn on the kibitzer and if your engine is like mine, the evaluation will likely be something horribly large, like +8 or more or something, yet the position is in fact dead drawn. Why? Because it's impossible for white to force a checkmate without an additional piece that can control the light squares.

In other words, engines are far from perfect advisors and they're only good at pointing out mistakes on a superficial level. They'll never be able to explain the reasons why you played the poor move when did...what fundamental misunderstanding in the game that made you think your move was good when, in fact, it wasn't.

Scid has an automatic annotating feature, in the analysis window at the bottom there is a notebook icon with a pencil called "Annotate". Click that button and a new window will appear for further configuration of analysis.

If you want to analyse your games yourself you need to have some understanding with which to judge your decisions.  This is necessarily hard or impossible for a newer player to do, so it's always beneficial to get the critique of stronger players, ideally master level (you can post games in the analysis forum).

The computer will give you alternate moves, but it can't give you the ideas behind the move (it doesn't even know the ideas itself).  If you want to analyse on your own, the first step is reading a middlegame strategy book like Pachman's modern chess strategy, and also useful would be an endgame book.  Then you have the knowledge to judge moves and come up with reasonable alternatives.

For example a computer won't differentiate between two moves that both yield 0.25 evaluations even if one move is a tactical mess and would require the player to find a few (or many) very difficult moves while all other moves lose.  The 2nd move may be a safe and logical plan where nearly all the reasonable moves maintain the .025 evaluation for you.

Also moves aren't the only thing that are useful to judge.  Sometimes a mindset or plan is incorrect and so fixing one product of bad thinking by having the computer tell you g5 is better than f6 is merely cosmetic because the next game you'll make the same type of error.

Computers will help you catch tactical errors though.

Recently , a vry strog titled player told me to toss any computer engines and do it the old fasioned way.

This is certainly good advice, as chess is very much about learning how to make decisions by yourself. Unfortunately, many young players are fascinated by the rough strength of computer engines and can't appreciate that they bring confusion to their developing understanding of the game...