Do you plan to extract figures about those gambit ? like winning % for white and black, draw %.
How does engine Elo impact the way those gambit are succesful or not ?
Thanks for the tournament Ed! Guenther has pointed this out in the Seer v2 thread, but I'll make a secondary post here. Seer v2.0.0 played many illegal moves in your tournament due to running out of time (getting so low on time that the bestmove from the previous position wouldn't be updated) as a consequence of buggy TM code which resulted in Seer sometimes going greater than 100ms over the allocated time budget. Many of these illegal moves were in completely winning positions as well.
Technically, the TM bug affects V2.0.0 at both cyclical and incremental TC, though is most problematic for cyclical TC where precise time usage is critical. In addition to resolving the buggy TM code, I've also tweaked my cyclical TM code to be less aggressive resulting in about 50 elo in my testing as apparently my old cyclical TM constants were quite terrible.
connor_mcmonigle wrote: ↑Sat May 01, 2021 9:28 pm
Thanks for the tournament Ed! Guenther has pointed this out in the Seer v2 thread, but I'll make a secondary post here. Seer v2.0.0 played many illegal moves in your tournament due to running out of time (getting so low on time that the bestmove from the previous position wouldn't be updated) as a consequence of buggy TM code which resulted in Seer sometimes going greater than 100ms over the allocated time budget. Many of these illegal moves were in completely winning positions as well.
Technically, the TM bug affects V2.0.0 at both cyclical and incremental TC, though is most problematic for cyclical TC where precise time usage is critical. In addition to resolving the buggy TM code, I've also tweaked my cyclical TM code to be less aggressive resulting in about 50 elo in my testing as apparently my old cyclical TM constants were quite terrible.
Seer 2.0.1 will be next
90% of coding is debugging, the other 10% is writing bugs.
xr_a_y wrote: ↑Sat May 01, 2021 6:23 pm
How does engine Elo impact the way those gambit are succesful or not ?
I don't understand the question, can you rephrase?
I was wondering if 2800 engines handle those positions very differently than 3200 engines for instance.
So maybe, the win rate for one gambit on a specific engine range is different from another range in a way that it is different from common knowledge (like draw rate increasing at a specific paste when Elo increase). Is it more understandable ?
xr_a_y wrote: ↑Sat May 01, 2021 6:23 pm
How does engine Elo impact the way those gambit are succesful or not ?
I don't understand the question, can you rephrase?
I was wondering if 2800 engines handle those positions very differently than 3200 engines for instance.
So maybe, the win rate for one gambit on a specific engine range is different from another range in a way that it is different from common knowledge (like draw rate increasing at a specific paste when Elo increase). Is it more understandable ?
I think that's a good question. I could run (as an experiment) a massive robin round between 2800-3200 engines. As an experiment because I am not sure if the games such a tournament produces are suited to include in the rating list. As a novice I leave this as an open question to the rating list experts.
90% of coding is debugging, the other 10% is writing bugs.