Trueskill model in pymc3

[junpenglao]

There are a few related posts you can check out:

Tournament Skill Estimator, some modelling challenges

Hi all, I’ve been trying to get some pymc3 models to work for my work & hobbies. Currently, I’m trying to make a skill estimator for a board game (fyi: Terraforming Mars), and I’m running into challenges regarding convergence (/“anchoring”) and modelling challenges. Unlike most games, we can not simply rely on the total points gathered by players, as the duration of the game varies a lot per game and is not necessarily related to players’ skill levels. The tournament will contain 3 rounds, w…

TrueSkill Model Setup

I wish to formulate the model specified in the TrueSkill paper by Tom Minka. Essentially, I want to figure out the posterior on skills and I have presented a simple case of the paper as below: Suppose I have 5 players with true skill levels denoted by s_i. During a game the player performance at level p_i\sim\mathcal{N}(s_i,\sigma^2). For simplicity sake, let’s assume that \sigma=1. The performance level is unobserved (latent variable), but we do see the rankings of each game that is played eg…