At times it is almost confusing watching the engine play with the new evaluation, it does not play like it used to and some moves just seem weird, but then in the end it comes out ahead usually.
I have high hopes for this new version and it will be very interesting to see how it performs on Le Fou Numérique etc.
Things I will leave for the next version (0.31 or so) are
- Bishop/knight outposts
- King tropism. Awarding pieces near the opponent king even if they do not directly attack the squares around it
- Only rewarding rook on seventh rank if there are still pawns there or opponent king is on the 8th rank
- Implement x-ray attacks in the general attack table, a queen behind a bishop that attacks the king should be taken account of
- Win chance evaluation. If near the 50 move rule the score should get closer to 0, and opposite color bishops ending etc.
- Some more endgame knowledge to avoid going into a losing endgame
- Pawn and evaluation hashtables to not repeat evaluations of already evaluated positions