When is Enough Enough?

    When doing an Infinite Analysis with your engine, how do you know when it has analyzed the position long enough so that you can be reasonably sure it has discovered the very best line? 

    Engines look at millions of possible continuations and keep going deeper as time goes by. so. At firs it only yields a shallow analysis, but as it does deeper and deeper into the positions it uncovers tactical moves, clever defenses and long range strategic ideas. The engine might start out with one Centipawn number and after it runs a while that evaluation might change.Another factor is the computing power of your hardware.  Thus, the question, when is it safe to end the analysis? 

    Even if an engine does deep into a position, it it’s complicated enough the engine can still miss something. If you see the engine’s

 top choice keep chamging them you know it’s having difficulty determining exactly what the best line is. If the evaluation of the main line stays stable then the engine has probably found the best line. But what of ot needs to go a little deeper and you end the analysis a bit too soon? 

    Another thing that comes into consideration is if the engine is running in MultiPV mde (showing more than one move) and the evaluation of two or more moves are very close then the engine is not certain as to which ois the absolute best. Yet another factor is the hash table size which the engine uses to store and reuse analyzed positions. If this becomes overloaded, the engine may start losing information and its analysis will be less reliable. 

    All of these factors explain why top level correspondence player ofyen use powerful dedicated computers, ket their engines run for days and tinker with different moves that the engine may not have fully considered. If you don’t have the time or patience for all that then the following guidelines should prove sufficient. 

     

* On most laptops an engine usually needs to run until it reaches a depth of 20 to 24 plies (10-12 moves) which usually takes from a few seconds to 30 seconds depending on the complexity of the position. 

 

* For a quick tactical check 5seconds reaching 8-14 plies should be sufficient to spot obvious hanging pieces, gross blunders or simple tactics. 

 

* Generally though 15–20 plies (about 15 seconds per move) will yield a better result when it comes to analyzing most middlegame positions and identifying major turning points.

 

* Sharp, complicated middlegame positions require about 25, or more, plies amd shpid take about 30 seconds per move. 

 

* When doing an Infinite Analysis it is likely to be a waste of time to let the engine run for more that a few minutes and once the engine’s top choice appear to have satbilized the best move has likely been found.