Answer Please: How to approach Chess Calculation

Strong player's insight needed: How can I approach calculation in chess? I mean, what is the correct method/sequence? Please, please answer!

Chess calculation starts with knowing when to calculate.

In many positions you don’t even need to calculate a lot of variations. You should mainly focus on calculation when there are forcing moves in the position, like Checks Captures and Threats.

Once you found your opponent’s threats, the second step should be to find all the candidate moves

Hi there,

For me board visualisation exercises have always helped with my calculation.

For example, trying to play as many moves from a game in your head as you can, or simple things like how does a knight get from d4 to e5 without looking at a chess board.

A lot of people think that Grandmasters spend the game calculating 20 moves ahead.

This is not true; trategy is equally important.

Hope I answered your question!

At least you should try to be original and quote your own ideas.

http://www.chess.com/forum/view/general/improving-evaluation

You're going to get a ton of different ideas here, but I'd break it down as follows.  

One type of "calculation" isn't really calculation, it is about pattern recognition, and being able to quickly see and think through short (2-3 ply) variations.  You can easily train these patterns by doing basic tactics regularly.  Knowing lots of those patterns helps with the more complicated process of real calculation which takes more time to do properly.

In addition to knowing when to calculate (ie, when there is a specific threat, there is some sort of tension on the board, you are considering a piece sac, etc), the basic process is this:

1. identify candidate moves

2. calculate each individual candidate to completion  

3. evaluate the final position 

4. compare your evaluations and pick the best (or least bad) move

That's it.  You can train this type of calculation by taking complicated positions and writing out all of your thoughts and analysis, and then comparing your work against the solution.  

You can train points 1 and 2 in the list above by solving (writing out) lots of mate in 2 positions.  This lets you focus on finding lots of candidate moves (step 1), and practicing the different techniques/shortcuts for line calcuation (step 2), while completely skipping the more difficult evaluation step (steps 3 and 4).  

Then you can move on to more complicated positions that involve something other than forced checkmate, and allow you to practice evaluating different kinds of positions (steps 3 and 4).

Thanks a ton! One thing I wanted to ask is HOW CAN I CALCULATE DEEPER, GIVEN ONE CANDIDATE MOVE?

Just go slow, and take time to visualize.  Also, alot depends on how complicated the specific position is.  If there are many branches, with many opportunities for both sides it can be almost impossible to do this in your head.  In Tal's book he wrote about such a position, where he calculated a knight sac for almost 40 minutes but couldn't determine if he was better in all lines.

If there are very few branches, and one clear mainline, you'd be suprised how deeply you can calculate now.  

The hard part about many branches/variations is keeping track of them all in your head.  That's why learning how to calculate by writing out your analysis helps.

I agree that in all positions you must search for all threats (make sure they are real threats, not "unreal" threats).  However, to clarify the point on "find all candidate moves", there are only five types of defensive ideas in response to any kind of threat.  Search for candidate moves that correspond to each of the five ideas.  No more, no less.

Good thread...tracking. Doing standard problems on ChessTempo and playing out the sideline variations of them has helped my board vision and calculation out tremendously. The key is doing standard problems, moving the pieces with your mind's eye, and taking as long as is necessary to find the solution.

I can give my ideas, but I'm mainly here to learn. This idea interests me a lot.

First of all, I started taking the "rule of thumb" to heart and it really helped me. I like to actually do it like this:

Check for checks

Check for captures

Check for threats.

90% of the time when I get a problem wrong, it's simply because there was a check, capture, or threat that I didn't see. (Check for them in that order).

I did this problem in 2:30 seconds. I actually got it wrong, I am deciding to show my thought process anyways to illustrate the important of checking all checks, captures, and threats. This is a 1715 rated puzzle on chesstempo.

Nice puzzle my first move lol I played by instinct was wrong HAHAHAH.

You explained why it was wrong very nice. I played Ra6 on my first move. It was my instinct to play this move because rooks should always be behind pass pawns. However, it not as great becuase of the resource Ba8. Which you explained in your comment post very nice.

After I got it wrong I tryed again lol I did find the move Rg8 afterward. Than I read your notes.

@xman720 -- first of all, kudos for sharing your analysis!  Very brave of you to do that on open chess.com forums!!  

It is good to always see your checks, captures, and threats, but that should not be the only thing driving your candidate list.  Two other approaches help:

1. evaluate the position first.  You need to have a general idea of what is going on in the position before you start looking for how to play it.  Your evaluation will give you "ideas", and those ideas help you generate specific candidate moves.  Start out with simple material vs activity evaluations, and over time you'll be able to add to it (like applying specific endgame knowledge).  So my quick evaluation of this position is white is ahead a Rook and two pawns vs two bishops (2 points), his a-pawn is close to promoting, and his rook is active (restricting the black king, and supporting his passed pawn).  White is playing to win, and black is playing for a draw.  Try to think of what this means for your opponent as well.  If he is playing for a draw, how might he/she accomplish that?  You will get better and better at evaluation with practice.  

You saw the passed pawn only after you had started searching for candidate moves using your checks/captures/threats approach.  It is good that you saw that, but I'm sure you would have noticed the passed pawn much sooner if you had quickly evaluated the position first.

2. tactical visual clues.  Especially when you only have "line" pieces like bishops and rooks, you should always be aware of pieces in a line (rank, file, or diagonal).  So in this position I immediately saw the king and pawn on the 7th rank, but I also searched for double-attacks the bishops might have against white along the diagonals (and found none).  

Those two simple steps helped me come up with several ideas about the position that I then used to look for specific candidate moves.  So I knew (1) white should be playing to win (so not looking for stalemate or forcing a drawn endgame), (2) that the a-pawn is tactically defended (...Bxa7 is an "unreal" threat), and (3) I knew the passed pawn is important.  

Then I looked at a few candidate moves based on those ideas.  I did not ever consider the "threats" Rg4/Re6/Rc6 since they do nothing to support my ideas (only 1. Rc6 looks interesting because it blocks the bishop from defending the promotion square, but it fails to Bxc6...of course Kxc6 would be bad for black).  However, if you do add moves like that to your candidate list, you can very quickly eliminate them once you see they are easily refuted.

Your two mistakes with 1. Ra6 were that you only found one move that accomplishes the idea, and then you did not calculate the defenses (but I'm not sure why).  You should have added Ra6 to your list, and then kept searching for ways to accomplish the same idea.  

Also, you say in your analysis that did not calculate all checks, threats, or captures by your opponent after 1. Ra6.  But I'd ask if you even looked for your opponent's reply at all?  Did you seriously ask yourself how your opponent might defend against your threat?  If so, kudos for looking!!  If not, you must do that!!

Now, if you really did look for your opponent's reply, keep in mind that you do NOT need to search for all checks, captures, and threats.  You only need to search for moves that defend against the pawn promotion, and in this case the move ...Ba8 does that very nicely.  You would not have found that move if you were only looking for checks/captures/threats.  You need to search for ideas specific to the actual threat on the board (promotion in this case).

So I would have seen 1. Ra6 Ba8, and black's threat is now to move his king over and capture the a pawn (black's plan is to draw by capturing your pawns).

Finally, I'd also ask for your evaluation of the final position?  What are white's winning chances, and what does he need to do to win?  What are black's drawing chances, and what does he need to do to draw?

Doirse, your idea of selecting candidate moves based upon the ideas gained by the board is very suggestive. Therefore, to come up with tactical moves, we must first analyse a position which only comes up with positional knowledge and experience. 

I was on a quest to find a brute-force approach to finding the best chess move, but I now know that there is a line betwixt us and the computers. We can only evaluate on our understanding, which comes up with practice.

By "come up", do you mean "improve"?

^I mean "play" by "come up".

As xman720 clearly said, his main fault in the first try was that he did not calculate the opponent's move. My question is, do we have to follow the same exhaustive method to find the opponents candidate moves to our candidate moves, and then our candidate moves to the opponent's candidate moves? That is why I was searching for a brute force and easy approach. 

I'm not sure I understand your meaning.  Evaluation does not involve any calculation -- you're simply looking at a static position (without looking at specific moves), and asking questions.  You do not need any specific positional understanding to evaluate a position.  

Valeri Beim wrote an excellent book on calculation, and in the opening discusses the difference between tactics and calculation.  On solving tactics he suggests:

"Train tactics by doing exercises daily to learn patterns, and by using logical analysis to help understand the position:

1. Count material

2. Find all weaknesses and strengths in pawn structure

3. Assess coordination and mobility of your pieces"

You will be slow and make lots of mistakes at first, but eventually you will improve, and you will be better at evaluation and, since evaluation is an important skill in chess, you will therefore better at chess.

Great question, and NO!!  There are two keys to calculating forced variations -- first, understand the threat (by evaluating the position), and second, ONLY calculate defensive replies based on the specific threat.  

There are only five defensive ideas, and in most positions, you cannot implement all five.  Usually you will only have two or three defensive candidate moves in the position, and you can often refute those candidates ("trim") and quickly find your mainline.

Thanks Doirse!