Engine selection - heavyweight

Updated: May 8

Nowadays, it is hard to imagine effective chess training, game analysis and sparring without the involvement of a computer opponent. It is the possibility of using modern tools that makes a simple, even old and slow computer sufficient – to take advantage of the benefits of computer support.

For training, analysis or checking errors in games, all you need is dedicated software and one of the proven and recognized only one chess engine playing at least at grandmaster level. Great for such activities are commercial engines such as: Fritz, Hiarcs or TheKing (Chessmaster). Of the free equivalents, Komodo (one of the older versions) or Rodent (versions II or III) engines are worth mentioning, as they are very customizable (parameters, personalities). From commercial, but not expensive alternatives, I suggest you get interested in ChessBase software, released a few years ago and available e.g. on popular auction sites. I mean, e.g. older versions of Fritz, Junior, Shredder - containing engines playing at 2800-3000 Elo - sufficient for the above mentioned chess tasks.


But... what to choose when you are a

  • heavy user, and you analyze chess games and positions a lot and very often

  • you are looking for novelties and ideas

  • you play many correspondence games at once

  • you need to catch the smallest inaccuracies, or

You play against another very determined and strong opponent who, like you, uses a computer with powerful chess software?


In this article, I will try to answer this question subjectively and with a pinch of salt.


Let's look at the ranking list of the ten ( 10 ) the strongest chess engines.

(The list is current as of the writing of this article)
Source: CCRL

Oh, in first place is Stockfish. So if I choose Stockfish - the strongest engine in this ranking list, will my analysis, my news and ideas -> finally my results be the best ?


No :-)


Of course not. And I write this as a correspondence chess player who has been playing for several decades and who has spent dozens of hours on analysis every week.


It is a mistake to choose only one engine for such a heavy task, it is a mistake to choose the same engine which is most often chosen by rivals and opponents.


This is why in hmm some well-known organizations of correspondence chess players 99% of chess games end in draws in games lasting many months and sometimes even several years.

And that's why the best correspondence chess players don't play in those organizations anymore; it's a waste of time and lack of satisfaction from wins after breaking dozens of opponents using the same chess engine.

(moving a bit away from the substance, where to play correspondence chess then - we will check out various cool places on the web in the next article).

So what to choose? How can an ambitious chess player work at such a high level ?


I suggest using not one ( 1 ) but four ( 4 ) chess engines.

The first engine.

The strongest engine, that will always have an advantage over others is your Brain.


Supported by knowledge, developed through further matches and games, developed through chess literature and movies. Add to this year's of experience and the right set of chess software and you get a level of play well above 4000 Elo (in the correspondence game). This is not a joke - in the text below some examples :-)

Just to clarify, obviously an ICCF correspondence player ranking of 2500 is not equivalent to a 2500 Elo ranking of the FIDE federation. Games played by correspondence are 1000-2000 "standard" Elo points higher than ordinary games played e.g. in FIDE at rapid or tournament pace.


Knowledge of chess, thanks to the work of chess players and new technologies such as chess engines, is moving forward, not standing still. It is worth deepening your knowledge and gaining experience, if only by playing regularly online with people or computer opponents.


Second and third engines.

They are "computer" friends. It is worth getting to know them well, it is worth getting to know their style of play, understanding how a possible change of parameters will affect their game and understanding the positions in which these engines achieve the best results.


The style of play of these two engines is very important to me, because I always choose two engines that play in a similar style to mine. If, for example, I like to play closed positions full of many strategic possibilities, then I want the two engines to aim for that kind of position as well.


The second engine I use is Komodo Dragon.

The third engine I use is Berserk.

I use both of these engines mainly in the early and mid game.


The third engine I use is Stockfish.

I use Stockfish for end game position analysis and for quick tactical analysis and checking/finding errors and inaccuracies.


Why I chose these engines - I will show you some examples.

All of the following positions were chess engine tested on a MacBook Pro with an Apple M1 processor (7 threads) using the Hiarcs Chess Explorer GUI.


 

First, a chess position from the famous 1998 game between Topalov Vesselin and Shirov Alexei.

Shirov played a brilliant 47...Bh3 !! (from a human point of view).

For years chess engines were unable to find this move, or arrived at it after a very long time of "thinking"


A modern engine finds Bishop's move to the h3 field in... seconds.

Stockfish managed within 4 seconds and...

Fen: 8/8/4kpp1/3p1b2/p6P/2B5/6P1/6K1 b - - 0 47

...Komodo Dragon found the best move in just 1 second !

Fen: 8/8/4kpp1/3p1b2/p6P/2B5/6P1/6K1 b - - 0 47

So much for difficult test items for engines ;-)


There are many such well-known test positions from various tests that once seemed "unsolvable" for engines, and today modern engines take a few seconds at most.


But what wonderful progress has been made in the domain of chess engines, hasn't it ?

The next entry is from a correspondence game played with white by the author of this article in 2014.

At the time, neither engine "saw" a winning line for white based on white's King maneuver from the g1 field to the e1 field.

Today, in March 2022, some engines see white's positional advantage but only a few engines are able to give a winning maneuver within minutes - Berserk is one such engine (44 seconds).

Fen: r6k/4qp1n/3r1np1/p1pPp1B1/PpP1P2P/1P3R2/5RQ1/3B2K1 w - - 0 49

An position from another correspondence game of mine. It takes a good computer and a strong chess engine to find a winning move 25.Rxg5+ in a reasonable time.

Komodo Dragon finds the right move in 18 seconds using the MTCS algorithm.

Fen: 5rk1/8/pqPp1r1p/1p1Pp1bR/4B3/5PP1/PP2Q1K1/R7 w - - 0 25

Below is an chess position invented by Mr. Lyudmil Tsetkov. Very difficult to solve by contemporary modern engines. The Berserk engine, which has a wonderful ability to "look beyond the horizon" found a winning and impressive move and continuation lines almost immediately !

Fen: 1r3r2/4bpkp/1qb1p1p1/3pP1P1/p1pP1Q2/PpP2N1R/1Pn1B2P/3RB2K w - - 0 1

Another very difficult position for chess engines but relatively easy for the human chess player. And in this case the Berserk engine gives the right move after a while.

Fen: 4n3/4k3/p4n2/Pp1p1p1p/QPpPpPpP/2P1P1P1/1R3BK1/7R w - - 0 1

The next position comes from a set called Nolot, which is difficult for engines to solve. For the Stockfish chess engine, finding the correct tactical hit is not a problem. After 4 seconds we get the correct solution.

Fen: r3rbk1/ppq2ppp/2b1pB2/8/6Q1/1P1B3P/P1P2PP1/R2R2K1 w - - 0 1

Below is a demonstration of the incredible tactical power of the Stockfish engine - only 15 seconds and a correct e5 indication.

Fen: r2qrb1k/1p1b2p1/p2ppn1p/8/3NP3/1BN5/PPP3QP/1K3RR1 w - - 0 1

And another position from my correspondence chess game played in 2002.

Indisputably 46.Qh3 leads to a draw. The Komodo Dragon engine using the MCTS algorithm finds a draw move within 4 seconds.

Fen: 6k1/3q2b1/8/7Q/2n5/P7/3p1PP1/3R2K1 w - - 0 46

Below is a very complicated position. Komodo Dragon once again finds the strongest move in a short period of time.

Fen: r2q3r/1p1bbQ2/4p1Bk/3pP3/1n1P1P1p/pP6/Pn4PP/R1B1R1K1 w - - 0 1

And the last of the positions where Stockfish can find a spectacular tactical strike in just a few seconds !

Fen: r1qr1bk1/2p2pp1/ppn1p2p/8/1PPPN1nP/P4NP1/2Q2PK1/2BRR3 w - - 0 24

As I wrote earlier, Komodo Dragon is the main analysis engine for me in the early and middle stages of a chess game. This engine plays in a style that suits me, has great chess knowledge implemented by outstanding programmers and Mr. Larry Kaufman, who is the Grandmaster and one of the creators of Komodo Dragon.

Extremely useful is the ability to change the parameters of this engine and to choose the MCTS algorithm, which in some positions handles extremely effectively.


Berserk, on the other hand, is a free engine that plays at a very high level distinguished by its "human" approach and ability to find amazing moves, ideas and novelties.


Stockfish, on the other hand, is an engine - a tactical monster and probably the most versatile in every phase of a chess game - but for me and in the positions I play - not at all the most effective of engines.


In my opinion, the key to success in the field of chess I'm describing, besides your own knowledge, is the ability to choose the right tools and use them, e.g. by changing engine parameters.

Of course, the more powerful the computer on which the engine runs - the better usually, because we will get to the desired results faster.

Although it is worth bearing in mind that a very powerful computer can simply ... can be