Player ratings are derived via OpenSkill’s implementation of the Weng-Lin Bayesian inference framework, utilizing conjugate priors and analytic Gaussian posterior approximations. Match volatility is dynamically modulated using a certainty-weighted β scheduler, encoding outcome ambiguity. Time-evolved μ drift is governed by bounded directional correction toward equilibrium, with rating state transitions committed via atomic transactional replacement over a chronologically ingested event stream. All upstream data is preconditioned via rigid structural constraints and sourced from a non-fungible, externally governed data issuance vector.

This rating system uses OpenSkill, based on the Weng-Lin Bayesian approximation method, as an alternative to TrueSkill.