Skip to content

Latest commit

 

History

History
35 lines (22 loc) · 1.33 KB

README.md

File metadata and controls

35 lines (22 loc) · 1.33 KB

Expected outcome of the TCEC 15 superfinal

Probability of winning the whole match after 97 games, 14 LC0 wins and 7 Stockfish wins:

LC0 Stockfish
100 % 0 %
Score Probability
53.5:46.5 54.0 %
54.0:46.0 25.0 %
53.0:47.0 14.7 %
54.5:45.5 4.4 %
52.5:47.5 1.6 %

Probability that LC0 is better: 91.6 %.


About

This script computes the expected result of the ongoing match between LC0 and Stockfish using Bayesian inference. It works like this:

  1. Assume a prior probability distribution for LC0's win/draw/loss probabilities. I chose the Dirichlet(3, 24, 3) prior. Dirichlet is basically a probability distribution over categorical probability distributions.

  2. Compute the posterior distribution based on the games played so far. This is simply Dirichlet(3 + wins, 24 + draws, 3 + losses).

  3. Perform millions of simulations of the rest of the match. Each simulation first draws specific win/draw/loss probabilities from the posterior Dirichlet distribution and then simulates the rest of the match.


Choosing the prior

The Dirichlet(3, 24, 3) prior provides a weakly held belief that the win/draw/loss probabilities are somewhere around 10%/80%/10%. It represents our state of knowledge when we start with Dirichlet(0, 0, 0) and observe 3 wins, 24 draws and 3 losses.