Continue your opening line
Paste the PGN of a line you are learning. The mainline is replayed, and the model replies from the position where your preparation ends.
Open PGN setupChoose either approximate playing-strength scale: Lichess from 600 to 3000 or FIDE from 1500 to 2600. Both control the same human-like opponent and are playable now.
Play a complete game from the initial position, or load a FEN or PGN and repeat the kind of position you want to practice. The selected rating target controls the opponent; the board and clocks follow the same rules in every mode.


Build the setup beside the board. The side-to-move clock starts only after Start game.
Five choices, one clear starting state.
Approximate Lichess playing-strength scale, mapped to the current supported bot target.
The first number is each side's starting minutes. The second is the increment added after every completed legal move.
Paste a PGN and the model takes over after its last mainline move, or paste a FEN and it plays from that exact position with its side to move. Leave it on Standard for the normal initial board.
Four things decide every game on this page: the player pool the model was trained on, the rating target you set, the clock you both play, and the position the game starts from.
The opponent is trained on games played by people, not on engine analysis. The scale selector changes how its supported strength target is presented.
Choose in 100-point steps on either approximate scale: 600 to 3000 Lichess or 1500 to 2600 FIDE.
The reply delay adjusts to the chosen time control and the bot's remaining time. Your clock follows the same base-time and increment rules.
Every reply is a legal move in the current position, played from the board you chose: the standard start, a pasted FEN, or the end of a pasted PGN line.
| Scale | Reference | Rating target | Time controls | Status |
|---|---|---|---|---|
| Lichess | Approximate online playing strength | 600 to 3000, 100-point steps | Bullet, Blitz, Rapid, Classical | Playable |
| FIDE | Approximate over-the-board playing strength | 1500 to 2600, 100-point steps | Blitz, Rapid, Classical | Playable |
What this bot is not. It is not Stockfish and it does not show an engine evaluation. For engine-precision opposition, use the Against the computer page. Rating targets sit on the model's own scale, so start near your usual rapid rating rounded to the nearest 100.
You do not have to start from the initial board. Paste a PGN and the model takes over after the last mainline move, or paste a FEN and it plays from that exact position with its side to move. Both are set in step 5 of the game setup.
Paste the PGN of a line you are learning. The mainline is replayed, and the model replies from the position where your preparation ends.
Open PGN setupPaste a FEN to load one exact board with its side to move. Play the same structure again until the plan holds up against human-like resistance.
Open FEN setupLeave the starting position on Standard and the game begins from the normal initial board with the model, target and clock you chose.
Keep the standard boardPractical answers about the models, rating targets, clocks, imported positions and mobile play.
The ChessDoctrine opponent is trained on millions of games played by people. It uses the current position and the selected rating target to produce legal moves shaped by human play at that level, and it manages its clock the way a person does at that time control.
No. This page uses the ChessDoctrine human-like model. Stockfish is the separate opponent on the Against the computer page.
They are two approximate ways to set the same human-like opponent. Lichess runs from 600 to 3000; FIDE runs from 1500 to 2600. Both map to the bot's current supported strength range.
No. They are approximate playing-strength targets for this opponent, not rating conversions. Start near your usual rapid rating rounded to the nearest 100, then adjust one step at a time.
Use your usual rapid rating rounded to the nearest 100. Choose one 100-point step lower for a warm-up and keep the clock stable while judging the fit.
Each side starts with ten minutes. Five seconds are added to that side's clock after every completed legal move. The clock starts after Start game.
Yes. Choose your color before starting. Random assigns your color when the game starts.
Yes. A complete FEN loads that exact position and side to move. A PGN is replayed through its main line and loads the position after the last move. The model then plays against you from there.
New game reuses the current setup. Change setup stops the clock and unlocks the model, rating target, time control, color and position. Reset settings ends the current game after confirmation and restores this page's defaults.
Undo is unavailable during an active human-like game. Review game becomes available after a finished game.
Yes. The board stays large, while a compact dock shows the current settings and Start. Setup opens in an in-screen sheet, so you do not have to scroll away from the board. Tap-to-move works on touch screens, and no account or download is required.
We use cookies to collect data for analytics and marketing (Google Analytics). By using chessdoctrine.com, you agree to this. Privacy policy