Critical Positions

“If you can't recognize critical positions or analyze them carefully and evaluate them correctly, then all that other knowledge is going to result in severely diminishing returns.”  NM Dan Heisman

A critical position is a point in the game where there is some sharp tactics or a key positional point that can determine the play in the remainder of the game.

It is exceedingly important to recognize these positions because in order to maintain your winning chances (or drawing chances) you have to find the best move.  Failure to do so will likely change the result to a result worse than you hoped for! These critical positions could be tactical, key endgame positions or a position where a specific positional idea must be followed.  Identifying the critical moments using a chess engine is relatively easy compared to OTB play and it is engine analysis I’m mainly concerned with here.

Basically any move that brings about a change in the position is a critical position.  In OTB games it usually happens when you run out of moves that you know in the opening.  In engine play it’s when somebody deviates from your database.  Transition from opening to middlegame and middlegame to the ending are also other critical points.

When analyzing with an engine and you run across a position where the engine evaluation jumps back and forth or you get significantly different evaluations from two engines (you DO use two engines, right?) you have a critical position and need to take special care.

After establishing the critical points, you can begin to analyze them. When using an engine, this will be positions where even they are having difficulty determining the best move and it is where human intervention is most often needed.  Unfortunately, in these positions most of us just aren’t strong enough to fathom the nuances of the position.  So, what can we do?

Giant databases can help unravel the mystery.  Perform a search by looking for pawn structure.  Of course, you will want to select the games from the highest rated players because if you don’t do that you will get thousands of useless positions from players like you and me!

The resulting search of positions will be a great value in helping you to understand the kind of structure you have reached in your game and you can play through the games the search discovered to see the plans used by strong GM’s.  This will enable you to get a better idea of which moves the engines are recommending are worth pursuing and which ones aren’t.

Very often when you reach a position where there are no tactics and positional play is called for engines sometimes have a habit of just shifting pieces back and forth without any real plan. The engine will point out the tactical moves but they are often unable to teach you anything about positional chess and strategies or explain to you why a move is bad.   This is what separates really strong CC players from the rest of us; they know junk when they see it. 

Another way to ferret out the best lines is to run shootouts from the given position with a couple of different engines.  For example, if the engines say that White has a slight advantage, but loses the majority of the games in a shootout from that position, then you will know the position is likely inferior but the defects aren’t going to show up for several moves.

I am playing a game now that is very interesting because critical points have abounded.  After much database searching and analysis with various engines I chose a plan of action involving a sacrifice that looked very promising.

Here is the position:

Should White sacrifice with 23.Nxh7

I played the sacrifice but my opponent apparently discovered an improvement that left him with what appeared to be a won game. 

Things were starting to look grim as the engines gradually showed evaluations swinging to my opponent’s favor.  Then he began relying entirely on engine recommendations in a materially unbalanced endgame position.  So far what’s been happening is the endgame database would show a win for him in about 30 moves or so but what the engines have doing is recommending the best moves according to the database for 3-4 moves then a less accurate move. The result is the database then shows the win being back to 30-32 moves away.  So a draw by the 50-move rule can’t be rule out because there is no guarantee the engine will pick the best move 30 moves in a row.  If this goes on long enough, I can draw.  Of course, there is also the possibility he will discover the win or I will make a mistake in my analysis.  Anyway, I hope to annotate this game as best as I can when it’s over because it’s one of those games where engines don’t seem to give very reliable results.  Maybe I can learn something.