Comments on https://adjva4.dpdns.org/@/lkama/blog/finding-opening-gold-a-statistical-hunt-for-wicked-lines/c0x9TFV7
Could you give the results for rapid and classical games (only) at 2000-2200 level ? (or hint me how to do it).
I would love to see the results for games at 1900-2400 elo for 1. Nf3 d5 for White.
Hello,
I'm working on an update so that it's easier to get the big picture for people that are not comfortable running scripts by hand. In the meantime, you can find some of the "hunts" performed here:
https://github.com/RemiFabre/WickedLines/blob/main/HUNT_INDEX.md
The instructions on how to run the tool can be found here:
https://github.com/RemiFabre/WickedLines/tree/main
Example, this will "hunt" for rapid only, 2000-2200 Elo:
python wickedlines.py --speeds rapid --ratings 2000 hunt
Best!
@LKama said in #1:
> Comments on adjva4.dpdns.org/@/lkama/blog/finding-opening-gold-a-statistical-hunt-for-wicked-lines/c0x9TFV7
2 existing alternatives based on statistics
https://chesspath.pro/
https://github.com/rooklift/nibbler
So seems many people are working on very similar projects as often is the case.
@mvhk said in #5:
> 2 existing alternatives based on statistics
> chesspath.pro/
> github.com/rooklift/nibbler
>
> So seems many people are working on very similar projects as often is the case.
Hello,
Thank you for the inputs. It's very welcome because I didn't properly search if the stuff had already been done...
However the first one seems to be a closed source project and the second one has a different use case (live engine evaluations). My approach is a modest statistical analysis at best.
I really like your idea but this is how I feel about installing dependencies on Python (as somebody who wasted 3 days trying to get Wan2.1 to work):
https://www.reddit.com/r/fuckpython/comments/1g9fqho/fuck_python_with_passion/
@identity-crisis said in #7:
> I really like your idea but this is how I feel about installing dependencies on Python (as somebody who wasted 3 days trying to get Wan2.1 to work):
>
> www.reddit.com/r/fuckpython/comments/1g9fqho/fuck_python_with_passion/
I feel you. I do that as a living and I agree :D
Maybe I'll try to build a website for this
@LKama said in #8:
> I feel you. I do that as a living and I agree :D
>
> Maybe I'll try to build a website for this
I'm a C++ programmer and everytime I see Python, I wanna cry xD
However, I got your program to run yesterday - fortunately it wasn't that diffuclt.
"Reachability ("If Wants %"): This calculates the probability of reaching a position assuming one player actively tries to get there."
* This means move 1 deviations have the highest reachability, like 1 b3, 1 g3, 1 e4 Nf6, 1 e4 Nc6, 1 e4 a6, 1 e4 b6, 1 e4 g6
"lines that create an asymmetric preparation battle are incredibly effective"
* Also a line that you play with the other color can be effective, i.e. that what you play against yourself.
"This discrepancy gives the Caro-Kann player a significant theoretical and practical advantage"
* Then 1 e4 b6, 1 e4 g6, 1 e4 Nf6, 1 e4 Nc6 should be even more effective, as white will encounter those less than 1 e4 c6 and are less likely to play these themselves as black.
"The King's Gambit (1. e4 e5 2. f4) is another excellent example"
* Then 1 b3 or 1 g3 should be more effective: more reachability.
"The analysis immediately flagged 3... f5, the Schliemann Defense, as the top opportunity for Black."
* Why more than 3...Nd4, 3...Bc5, 3...Nge7, 3...d6, 3...g6 with the same reachability
"By playing the aggressive Schliemann" * Here is a point: aggression. However, that contradicts the Caro-Kann example.
"What lines are you curious about?" 1 e4 e5 2 Nf3 Nc6 3 Bb5 vs. 3 Bc4 vs. 3 d4 vs. 3 Nc3