About ChessDoctrine.com

Chess Doctrine is a free chess training site. Anton Shuravin writes every page and builds the tools, and critical lines are checked with an engine and master databases before they are published.

Anton Shuravin, founder and sole author of Chess Doctrine The author

Anton Shuravin

Founder & Author

Anton Shuravin runs ChessDoctrine and is its sole author. He has played chess for more than 14 years, is rated 1900+ on Lichess in bullet and blitz, and has recorded 88 lessons for the ChessDoctrine YouTube channel.

  • Over 14 years of play, online and over the board
  • 1900+ on Lichess in bullet and blitz
  • 88 chess lessons recorded for the ChessDoctrine YouTube channel
  • Writes the opening guides, strategy and endgame articles, and annotated games
  • Built the puzzle trainer, the training bots and the free game diagnosis

Profiles

Longer author bio:

What you will find here

How the site grew

Chess Doctrine started as a set of opening guides that explain the ideas behind the moves in language a club player can use. A glossary of chess terms and a blog followed.

Then came the training part: human-like bots, the chess legends, the puzzle library and the opening trainers. The newest addition is the free game diagnosis, which looks at your own games and points you to the guides and puzzles that cover your weak spots.

How the bots and the diagnosis work

Human-like bots

The bots use one model trained on human games across the whole rating range. You choose a target rating, and the model picks the moves a player of that strength tends to find, and misses the ones they tend to miss. A strong engine that blunders at random plays very differently. A bot set to 1200 plays like a 1200 over many games; a single game can go better or worse, as it does for a person.

Chess legends

A legend bot learns from that player's recorded games, so it plays their openings and follows their habits. It is not the player, and it will not replay a specific historical game.

Rating estimates

The diagnosis estimates your strength from the moves in the games you give it. The estimate is calibrated against players with known ratings, so a result of 1650 means your moves look like those of a 1650 player.

Puzzles

Every puzzle comes from a real game and goes live with one checked solution and the rating band it belongs to.

The models and the analysis code are not published. Every number in a report is computed from your own moves.

How the guides are written and checked

  1. Written by a person. Anton drafts each guide from more than 14 years of tournament and online play and from his own games in that opening. Nothing is published automatically.
  2. Checked with an engine. Main lines, sidelines and traps are replayed with a modern engine at depth. If the engine refutes a line, the line is removed or the text is rewritten.
  3. Compared with master games. The main line is the one masters actually play, according to the Lichess Opening Explorer and master-game databases.
  4. Consistent terms. A term used in a guide matches its entry in the glossary, and when a definition changes, both pages change.
  5. Diagrams from the position. Boards are drawn from the actual position, so a diagram always matches the moves printed next to it.
  6. Signed and dated. Every guide shows the author and the last-updated date, and it is revised when theory changes.

Corrections

If you find a mistake, send the page address and the move or sentence to media@chessdoctrine.com. The line is checked again with the engine and the master database, and if you are right, the page is corrected.

Typos and broken links are fixed without notice. A change to chess content, such as an evaluation, a move order or a recommendation, also updates the date on the page. You get a reply either way, and if the line stays as it is, you get the reason. Pages are never deleted to hide a mistake, and a guide that changes its advice says so.

Questions about the site itself are answered in the FAQ.