How do xiangqi ratings work?
These answers restate the rules implemented by our open-source engine, which follows the WXF laws and is regression-tested against the official 110-diagram repetition casebook. Where an answer refers to a specific position, that position lives on the linked reference page, where the build verifies it against the engine before publishing.
Direct answer
Almost all of them are Elo-style: your rating predicts your result against an opponent, and the difference between the prediction and the outcome moves it. Beat someone stronger and you gain a lot; lose to someone weaker and you lose a lot. Ratings from different sites are not comparable.
Why cross-site comparison fails
A rating measures performance against that pool, calibrated to whatever starting value the site chose. A 1700 on a small site with strong players may be stronger than a 2000 on a large casual one. The only figure meaningful in the abstract is a national federation rating, and only within that federation.
What moves it
Not opening study. Tactics and basic endgames decide almost every game below expert level, and reviewing your own losses is the highest-return habit. Rating follows from those and does not respond to reading.
Related questions
What is a good rating?
Roughly: 1000–1200 new, 1400–1600 no longer hanging pieces, 1800–2000 strong club, 2200+ approaching master.
Should I play rated or casual?
Both — casual for experiments, rated for feedback. Avoid rated games when tired.
.