A machine that learns to predict football scores
Clone or download
Latest commit 711b9f7 Feb 5, 2017
Type Name Latest commit message Commit time
Failed to load latest commit information.
data Merge branch 'master' of github.com:Chimney42/positronic-brain Apr 8, 2016
src Do not add empty market values Nov 11, 2016
.gitignore ignore config file in repo Apr 8, 2016
README.md Update README.md Feb 5, 2017
package.json implement neural network Mar 13, 2016



This is an application to predict soccer results. More specifically, you can use the positronic brain to get a probability distribution for three classes of possible soccer results (home team wins, draw, away team wins). As a basis for this prediction, historic results along with the market values of transfermarkt.de are used. This project also provides a crawler to store data locally in a CouchDb.

Positronic brain was originally conceived as a way to apply learnings from this 12 week machine learning course

In the current state, the following data is gathered:

  • Market values at the time of the game
  • Current position in the league before each game
  • Matchday



Start a couchdb on localhost port 5984, e.g. with docker

docker run -d -p 5984:5984 -v $(pwd)/data:/usr/local/var/lib/couchdb --name positronic-brain-couch couchdb

Install npm packages

npm install


node src/main.js