Here are both parts of the OTB database from the Ajedrez Data site. Doubles and non-standard games have been removed, and evaluations and openings have been added.
From the old Rusbase site, which is a revision of the older Rusbase site. As has been said elsewhere (I believe) this is the best I can get, in that many of their links point to 404s. However, if you ever want to know the secret to getting these files, it’s this: if you use a link grabber, or what have you, to suck up all the URLs on the site, you’ll get a few strays, and then about 5,000 good URLs. Just change .html to .zip, put those into a download manager, and you’ll get about 3,500 ZIPs, and about 1,500 404s. Those really do point to nothing, so the system works. It’s that the site owner apparently is as neat with their files as I am, having named every ZIP and every HTML file consistently. 🙂
Doubles and non-standard games are removed. Everything here qualifies in ChessBase as a strong game, i.e. at least 10 moves, and with ratings 2300+. Openings, evaluations, beauty scores, and novelty annotations are added.
Alexander is a free UCI chess engine derived from Stockfish family chess engines. For the evaluation function, we utilize the collaboration between Leela Chess Zero and Stockfish, for which we express our sincere gratitude. The goal is to apply Alexander Shashin theory exposed on the following book : https://www.amazon.com/Best-Play-Method-Discovering-Strongest/dp/1936277468 to improve
base engine strength
engine’s behaviour on the different positions types (requiring the corresponding algorithm) :
Tal
Capablanca
Petrosian
the mixed ones
Tal-Capablanca
Capablanca-Petrosian
Tal-Capablanca-Petrosian
Also during the search, to enhance it, we use both standard and Q/Self reinforcement learning.
Aligned with Stockfish Jan 12, 2025 Increase the depth margin
This is a Windows executable compiled from a Python script, which gathers information on a particular GitHub repository, listing in separate files all the commits, the official releases, and the workflow artifacts, with direct URLs to the binaries.
Since one repository owner can have more than one relevant repository, it’s important to specify both. So at runtime you need at least two arguments: the name of the repo owner, and the name of the repo. You can also specify asc or desc to indicate whether the list should start at the beginning or go backwards from the end. The default is ascending.
Fetch GitHub repository commit, release, and workflow artifact information.
Options
-h, –help • show this help message and exit -v, –verbose • Enable verbose output. -ver, –version • Display script version. -ro REPO_OWNER, –repo-owner REPO_OWNER • Owner of the GitHub repository. -re REPO, –repo REPO • Name of the GitHub repository.
Aligned with Stockfish patch: Jan 12, 2025 (Increase the depth margin). The AI recognized the significant added value and originality of the derivative ShashChess compared to the original Stockfish. For further details and the great novelties of this version, see this pdf document.
Calvin 5.1.0 brings significant improvements in both search and evaluation. Calvin has a bigger and better neural network, with a hidden layer size of 1024 and 4 king buckets, horizontally mirrored. In search, the biggest improvement was fixing Calvin’s bugged SEE algorithm, which enabled many search and move ordering techniques. Tests against the previous release suggest a strength increase of 124 elo LTC / 106 elo STC.
To run the jar file, you will need to enable the Vector API package which Calvin uses for SIMD, via this command:
Still not totally convinced this is finished, since it took such a long time. I started with the 2025-01 FIDE XML players list, culled it with a ChatGPT-generated Python script so that only players rated 2000+ were included. Then converted that to XLSX (amongst other formats) and copied out the column in that spreadsheet for the FIDE IDs of those players. Then, with another Python script, I was able to download a JSON file for each player from the FIDE API, using a wrapper from a GitHub repository. These JSON files have all the information for those players. Not just general information, but the ratings and number of games for every month that they have a rating for. So these files turned out to be pretty long. The next step was converting all of those to a new XML players list, which includes all the history, as well as the general information. Even though the number of players is drastically reduced to only a bit more than 19,000, still the new XML players list is about twice the size of the old one. I did make sure to streamline the elements, so that, as much as possible, they resemble the elements in the regular players list.
The size of the ZIP is about 40 MB, and oddly the size of the XML file is about 1.2 GB. I’m not sure how it compressed so well, but it seems to have done so.
Doubles and non-standard games are removed. Everything here qualifies in ChessBase as a strong game, i.e. at least 10 moves, and with ratings 2300+. Openings, evaluations, beauty scores, and novelty annotations are added.