I've read in a book that one should play a line if (s)he thinks it is decisive and let the opponent proves that (s)he is wrong.

But what about playing a (risky, by risky, I mean that I am not sure) combination when I don't see the refutation?

The trouble I have with sacrificing when I am not sure is that if I am wrong, the opponent will win without effort, just by answering to my moves.

Here is an example of a game that I saw a combination (18. a5), but I didn't play it because I was not sure enough (and it turned out to be good):

[FEN "r4rk1/1b3pb1/pq2pn1p/1pn3p1/PQ1N4/2N1P1BP/1P2BPP1/R2R2K1 w - - 3 18"]

1. a5 Qa7 2. Ndxb5 axb5 3. Nxb5 Na6 ( 3... Qa6 4. Rd6 Nfe4 5. Rxa6 Nxa6 )
4. Qe7 Qc5 5. Qxb7

By the way, my ELO is around 1900.

  • 2
    Gary Lane advises having a long think before decisive moves. I can't tell you how many times I've spotted blunders in this fashion.
    – EMBLEM
    Commented Sep 18, 2016 at 6:46
  • Use your intuition.
    – limits
    Commented Sep 19, 2016 at 1:56
  • 3
    Well, the alternative is to only play combinations when you do see the refutation... Commented Sep 19, 2016 at 13:15
  • @RemcoGerlich I edited the title to avoid such a confusion :) .
    – antoyo
    Commented Sep 19, 2016 at 13:31

9 Answers 9


Often times, a combination is the most direct way to take advantage of superior coordination of your pieces. If you want to get better, you have to have more confidence in your analysis at the board and just go with what it tells you.

If there were multiple ways to win however, it can be simpler (and less error prone) to win by positional means and not have to rely on calculation to win the game.


It depends on how well you like your chances without the combination.

If, taking everything into account (material, position, opponent's rating, etc.), you are likely to win the game in the normal course of events, then play it safe.

On the other hand, if you have lost material, or are in a bad position, or your opponent is rated much higher than you, take a chance.


One one hand, a common piece of advice is to assume your opponent will always find the refutation.

However, I also seem to recall reading an story in "The Inner Game of Chess" by Soltis. During a game, a position may come up where a sacrifice is possible with say 10 to 12 possible continuations. Roughly 7 would win, 3 or 4 would lead to an even position and the remaining would lead to a worse, or even lost, position. One GM, having done all that calculation, would feel the time wasted, if he didn't go for the sacrifice, while another, upon seeing the losing lines, would not feel comfortable playing the sacrifice.

Simon Webb also talk about how to defeat a stronger player, in his book "Chess for Tigers", by getting into complicated positions, where they may lose their way just as easily as you could.

Basically, you shouldn't play "hope chess", where you play a move, and hope your opponent doesn't see the obvious refutation, but, depending on your opponent, you shouldn't necessarily be afraid to play a sacrifice or get into tactical positions.

  • 1
    I would say that you should not play "hope chess" against a weaker player that you can beat by conventional means. But that you should play "hope chess" against a stronger player that you can't beat otherwise. If you were playing poker against a world class player, you should go "all in" on a straight or flush draw against a pair, even though your chances of winning are less than 50-50, rather than lose for sure over time.
    – Tom Au
    Commented Sep 21, 2016 at 11:54

Without regard to the "refutation," you should evaluate the position after the combination.

At one level, you sacrificed a knight for two pawns. That's "bad."

On the other hand, you got two passed pawns for the knight. That's a "horse of a different color" (pun intended). A passed pawn is often worth two ordinary pawns. So your compensation for the knight may be 3-4 points rather than 2.

You could get your Queen to the seventh rank. Usually it's a rook, but in this case, the queen is just as good. That by itself is worth a pawn. It would have pinned a bishop, which was soon captured, leaving you two pawns to the good. Even if you can't see the whole sequence, your intuition should tell you that you are doing well.

So you got at least 3, and could have gotten as much as 5 pawns' compensation for the knight.


Whenever I do that, and if I survive, almost always I didn't have a good time, always worrying along the way. I learned that at our level (I'm FIDE 1996) just playing good solid moves consistently and waiting for the opponent to make a mistake or stretch more than the position allows is the easiest way to win games, even when the position seems boring and even. My openings is extremely solid but I win most of the games. I think I beat them through boredom.


I've had the same problem, and I think it affected my chess. I used to be a bit more carefree. About all you can do is check off on things like, is any piece loose? Did I miss an obvious capture? Did I miss any double attacks? Also, does his position have any clear signs of weaknesses, and does my combination capitalize on that?

The problem with cutting down risk is that you may stop asking interesting questions about the game, period. And even at 1900/2000 you (meaning me, too) will still make mistakes well worth learning from.

But on the other hand, as you said, you don't want to just lash out and make a combination that fizzles. I know I needed to track what sorts of combinations I couldn't pull the trigger on, to notice patterns of when and where I needed to look a move deeper. In your game's case it would be interesting to know why you didn't play things. You wind up winning material pretty safely after a few moves. Did you not see the material win, or were you worried Black had a resource?

My guy feeling is that your opponent is as scared as you are about tricky positions and moves. I think we all are. So I think if you are worried about potential simple refutations, have a list of things that you must check straight out so that, if your combination is refuted, the refutation will be interesting.

However, if you're clearly winning anyway, you don't want to do anything fancy.


Fear of losing is no way to play chess. Also, the sequence of moves is forced so it becomes evident that your ability to see the position after a series of forced moves to point A needs some work, but that is almost always the case for sub-master chess players. Since the series of moves is forced, you can safely play the first several moves to point B without any great concessions, reaching an intermediate position where a longer think with a clearer positional vision is possible. So:

[point A] After 1. a5, Qa7 is forced or black drops the Nc5. After 2. Nxb5 axb5 is forced. After 3. Nxb5 Na6 looks forced again. At this juncture you have 2 connected, passed pawns for your sacrificed knight. That is marginal and you could have played to this position safely. [point B]

At this point, the finish of the combination is trivial.

That is how my thought process goes with positions that have a forced continuation.

Good luck and keep improving!

p.s.: I'm sure masters don't see everything. They do the same thing in the context of a real, OTB game more than you might want to believe.


The question implies that you can't see a refutation, but it doesn't presume there is one to see in the first place.

So, the question really becomes: If I can't see a refutation, what are the chances my opponent will find one?

Well, clearly, if there isn't one, he won't find it. But there are several ways he could find one if it's there.

  1. He could see it as soon as you play your move.
  2. He could find it as the position evolves, and you play out your combination.

Of course, #2 is much easier, since the pieces are moving into position, and if there is a refutation, it may become much more obvious as it approaches.

So, #2 represents the real risk. If you can visualize the position after you make what you believe is the last move of the combination (mating, or winning material), then sit still and figure out all of the moves he could make that might affect you in any way. For each one, you need to know that you have an adequate reply.

If you can't visualize that position clearly enough to identify those moves, then you can't determine whether the move is safe to play. In that case, I would advise that you not play it unless your position is desperate.

By all means try to confound a superior opponent; for one thing, it will burn time on his clock.

If you are a creative and resourceful player, it may be worthwhile to play a sacrifice that offers you some positional advantages, even if it may have a hidden refutation. Whether you should do this or not depends on how well you generally convert such advantages into wins.

Mikhail Tal was famous for his tendency to attempt sacrificial combinations that were too complex to calculate fully; he depended on his ability to improvise as the position evolved, and find new resources.

Karpov, on the other hand, while a very competent tactician, would never take such a risk. He would only play the combination if he could visualize the final position and determine that the opponent had no meaningful threats.

So, it's a matter of style, to a large degree. Build your tactical vision by playing through games, 4-5 moves per side at a time, without moving the pieces. Visualize the board, and try to predict the next move. Then, bring the position on the board up to the current move, and check how well you visualized the position. Finally, see if you predicted the next move correctly.

If you can't visualize the board reliably 90% of the time, reduce the number of moves and try again. If you can, try stretching yourself to increase it. If you can handle 6-7 moves per side, you're doing pretty well.


While this is a not a direct answer, when not sacrificing, there's only small risk and a good chance of reward as your opponent might not see the refutation either. When sacrificing, it's often better the play the direct approach of superior force.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.