knight's multiple atacks

[Daniel Anulliero]

Hi all
Do you think it's useful to add a bonus for knights attacks (N atacks KQ, KR. ..etc) in the eval?
Or "a good engine must find such winning moves without that?
Anyone has tested?

[jdart]

Quiescence search will find this.

--Jon

[Daniel Anulliero]

Oups. . Yes , True ...
Thx John !

[hgm]

Quiescence search will find this.

--Jon

Well, not really. A standard QS would allow you to stand pat in the face of a fork on QR, not realizing the loss of the exchange is unavoidable.

[nionita]

Quiescence search will find this.

--Jon

Well, not really. A standard QS would allow you to stand pat in the face of a fork on QR, not realizing the loss of the exchange is unavoidable.

I introduced a few month ago en-prise terms in evaluation for Barbarossa (penalties only for side to move: hanging pieces, attacked, attacked by less valuable) and got +5 Elo after optimization. Based on this I optimized (with Clop!) the moving advantage (mid, end) and got further +13 Elo.

I'm sure those en-prise terms could be refined further, especially forks, but also other constellations. It seems to help in cases of stand pat.

[gladius]

Hi all
Do you think it's useful to add a bonus for knights attacks (N atacks KQ, KR. ..etc) in the eval?
Or "a good engine must find such winning moves without that?
Anyone has tested?

Stockfish has a fairly well tuned threat evaluation term. It does seem to help quite a bit, more than 10 ELO.