Chess Forum

No good news, all about books: I've started a new tournament with the latest Stockfish PB with a few books:

-Bloom
-Book (cubail's)
-Goi
-LesChats06072023
-OPTIMUS32

3 minutes, 3 threads per engine. On almost 150 games, not a single win...

I've started the thematic books project to try to gather some money because the bank will subtract the credit card fee at July 15 and I'm enough below 0... I've found out a program that actually cuts all the games at a certain move, it's pgn-extractor with a bash script. Sadly, I've discovered that the produced book hasn't the best moves marked as green... I imagined that the automation of the process would be precise because of my good choice of the initial database... Things didn't work like that... The book was on the A28 opening (fixed length: 15 moves), I included every player equal to and over 2500 Elo plus any engine game I have with A28... The resulting book has other initial moves than 1.c4 and it did not have the best moves marked green at #2 for White and Black.

What a delusional day...




This is the stub of the book I made on A28:

Tuned again the A28 book with engine games only today, got a big surprise: I've made it match 30 seconds + 0.2, 1 thread, no NNUE network against Goi and DON (December) and I got a surprising 9-7 for Goi-DON on 255 games!

I've manually tuned the book with the losses and started a new tournament, this time 600 games Run the gauntlet English - A28 vs Goi and DON.

Tuning the losses I have discovered a few errors on Goi too.

LesChats06072023 just won a 750 games tournament vs other books, my compliments!

1 thread, last Stockfish PB, no NNUE, CPU i7-4771, hashtable 64 MBs, interface Fritz 18.

books, Blitz 1.0min+0.5sec 2023

1 LesChats06072023 126.5/250
2 DON 126.0/250 -658.75
3 Bloom 126.0/250 -660.25
4 Goi 125.0/250 -758.75
5 Book 125.0/250 -761.25
6 OPTIMUS32 121.5/250

Games: