SPCC: Huge experimental test of Dragon 2.5 playing-styles

[pohl4711]

Huge experimental RoundRobin tournament (13200 games) with 6 different playing-styles of KomodoDragon 2.5 (Default, Defensive, Positional, Human, Active, Aggressive), each style combined with MCTS on and off = 12 engine-settings. See the result and download the games in the "Experiments"- section finished.

https://www.sp-cc.de

(Perhaps you have to clear your browsercache or reload the website)

[mehmet123]

There are huge differences between default engine and different style versions of Dragon.
Elo diffence between Dragon 2.5 Default and Dragon 2.5 Aggressive is 465 elo. Even if Dragon 2.5 Aggressive played without using a net there wouldn't be such a big difference. It would be more correct to call this style Carlsen style. Because with agressive style, the engine plays almost at the level of Carlsen. The situation is no different from other styles.

[carldaman]

My understanding is that these personalities all use the hand crafted evaluation (HCE) and therefore it's just different stylings of the old HCE Komodo, and not the newer Dragon which uses NNs.

[Rubinus]

There are huge differences between default engine and different style versions of Dragon.
Elo diffence between Dragon 2.5 Default and Dragon 2.5 Aggressive is 465 elo. Even if Dragon 2.5 Aggressive played without using a net there wouldn't be such a big difference. It would be more correct to call this style Carlsen style. Because with agressive style, the engine plays almost at the level of Carlsen. The situation is no different from other styles.

I don't think it would make that much of a difference for the longer term.

[Rubinus]

There are huge differences between default engine and different style versions of Dragon.
Elo diffence between Dragon 2.5 Default and Dragon 2.5 Aggressive is 465 elo. Even if Dragon 2.5 Aggressive played without using a net there wouldn't be such a big difference. It would be more correct to call this style Carlsen style. Because with agressive style, the engine plays almost at the level of Carlsen. The situation is no different from other styles.

I don't think it would make that much of a difference for the longer term.

Looks like I may have made a mistake. Even at 15+10 pace, the difference is significant. Caissa 0.4, tip according to my tests around 3250 CCRL - Dragon 2.5.1 positional sofar 0:1=7! --> cca positional is weaker by 220 ELO (15+10) It may be a biased opening, all Kings Gambit, I'm testing my library rather than the engines.