TalkChess.com

Posted: **Sat Nov 02, 2019 8:42 pm**

Chessbase has a 'Tactical Analysis' feature. While one would think about any engine would work as good as another as features seem to be preset, I am wondering how Komodo MTCS would fare vs say Komodo regular as concerns their analysis of an existing game.

I ask because when you set Tactical Analysis to go with regular Komodo or Stockfish, the GUI automatically displays 2 lines. I wonder if setting K-MCTS to 2 or 3 lines at about 4 core would actually produce 'better' results than a regular A/B engine?

Using something like the Fritz GUI which offers more analysis might even show better results for MCTS?

I am currently having MCTS analyse a blitz game I just played and so I have no data to look at myself and was wondering if anyone else had done this or had opinions on the merits of MCTS in such analysis - Tactical Analysis set to about 7 or 8 sec a move or Fritz GUI game auto-analysis.

Posted: **Sun Nov 03, 2019 5:46 pm**

I probably should not have tried this on a mere 3 min game because an engine will find so many 'errors' (not optimal move anyway) in them, but there is a definite difference in the output between Komodo and Komodo MCTS....find it hard to evaluate because of the game. I may try running each (in Tactical analysis...and the Fritz GUI) latter today if I have time. That would likely give a better understanding of which version performs best.

Posted: **Sun Nov 03, 2019 5:55 pm**

leavenfish wrote: ↑Sat Nov 02, 2019 8:42 pm Chessbase has a 'Tactical Analysis' feature. While one would think about any engine would work as good as another as features seem to be preset, I am wondering how Komodo MTCS would fare vs say Komodo regular as concerns their analysis of an existing game.

I ask because when you set Tactical Analysis to go with regular Komodo or Stockfish, the GUI automatically displays 2 lines. I wonder if setting K-MCTS to 2 or 3 lines at about 4 core would actually produce 'better' results than a regular A/B engine?

Using something like the Fritz GUI which offers more analysis might even show better results for MCTS?

I am currently having MCTS analyse a blitz game I just played and so I have no data to look at myself and was wondering if anyone else had done this or had opinions on the merits of MCTS in such analysis - Tactical Analysis set to about 7 or 8 sec a move or Fritz GUI game auto-analysis.

With best two lines displayed, Komodo MCTS and regular Komodo are of about equal strength; I would go with MCTS for reviewing human games because the move choices tend to be more "human". With both displaying 3 lines, MCTS is clearly stronger, and the gap widens with each extra line. When the number of lines exceeds 5, Komodo MCTS is stronger than Stockfish doing the same.

TalkChess.com

Auto/Tactical analysis with Komodo MTCS

Auto/Tactical analysis with Komodo MTCS

Re: Auto/Tactical analysis with Komodo MTCS

Re: Auto/Tactical analysis with Komodo MTCS