zaterdag 9 januari 2021

Opening selection TCEC 20 superfinal

Here are Jeroen Noomen's comments on the opening selection for the TCEC 20 superfinal:


"Here are the most important details regarding my selection of the openings for the TCEC 20 superfinal:

  • At www.twitch.tv/navratil25 two interesting matches between LC0 and Stockfish were played recently. The first match was bookless. It ended in 100 draws, with all games being predictable, boring and without any winning chances. The second match was played with more or less balanced lines. It ended in 2-2 with 96 draws. Again many predictable, boring games, in which the engines had nothing no play for. Thus the conclusion is clear: to keep computer chess interesting and avoid a huge number of uninteresting & boring draws, unbalanced openings with higher book exits are absolutely necessary.
  • The TCEC 20 superfinal book was mostly completed in June 2020, pre NNUE. Compared to all my other superfinal books, there has been a big change: the opening with the lowest book exit will be played first, followed by the next slightly higher book exit and so forth, until line 50, which has the highest book exit. This means it is possible that the same opening will be played two times in a row.
  • To determine the book exits I took the average evaluation of Stockfish 12 (5 minutes on my 8 core Ryzen) and LC0 v26.2 J92-100 (2 minutes on my notebook). Of course the book exits at TCEC will be different, because of much faster hardware and different Leela and Stockfish versions. But all in all we will see a superfinal in which the openings will become riskier as the match progresses.
  • As usual, all important openings will be played; just like in the previous two TCEC superfinals there will be a lot more risky opening lines compared to TCEC 16 and before. However, the TCEC 20 superfinal book will be less extreme than the TCEC 18 superfinal book, which was a bit too much in my opinion.
  • In the TCEC 20 superfinal we will see many openings (exactly the same, or slightly different) that have featured in previous superfinals. 
  • The chosen lines typically have a length between 1 move and 12 moves; there are two lines which are only 1 move (i.e. the first white and black move), there is one line with 14 moves of book and one line with 16 moves of book.
  • Average move length is 6.76 (which is 0.32 higher compared to the TCEC 19 superfinal).
  • There are sharp lines in the set, as well as positional linesmore complex lines, lines with opposite castling ("SALC", i.e. short and long castling), dubious lines, lines with an imbalance in material and gambits.
  • As I mentioned in my article about the TCEC 18 superfinal book, I now prefer 1-0 1-0 results over two straight forward draws. In effect the result is the same: 1-1. But two wins give more excitement. 
  • I have avoided positions that could lead to many exchanges and/or have a high draw rate in correspondence chess (I prefer draw rates of 50% or less), as well as positions with symmetrical pawn structures and positions that could lead to easy wins for the side with an advantage. The last aspect is becoming more and more difficult, with strong hardware, a long time control and engines being so strong. Of course risky lines means that the chance of 1-0 1-0 is increasing.
  • What do I expect? 20 wins is the minimum I'd like to see, but of course I hope for more wins. My personal favourite lines are #3, #12, #13, #27, #42 and  #47. Enjoy the superfinal!

ECO code distribution
ECO A: 12 lines
ECO B: 14 lines
ECO C: 11 lines
ECO D: 4 lines
ECO E: 9 lines

Move length distribution
1 move: 2 lines
2 moves: 2 lines
3 moves: 5 lines
4 moves: 8 lines
5 moves: 3 lines
6 moves: 5 lines
7 moves: 5 lines
8 moves: 4 lines
9 moves: 4 lines
10 moves: 6 lines
>10 moves: 6 lines

Geen opmerkingen:

Een reactie posten