Football modelling & statistics
We are a leading company in forecasting the outcome of football matches
Engineering at Football Radar
We know that happy programmers do the best work. We try to create a great work environment and a software process that will free our developers to be as good as they possibly can.
We value team spirit and have inherited some social democracy from our Danish founders: everyone brings ideas to the table, and the best idea wins; our company structure is simple with very little bureaucracy.
We have a variety of interesting and challenging projects. Our team works with complex regression models, time series models, high-dimensional data, text parsers and statistical classifiers, and the sheer amount of data we collect means that there are countless hypotheses to investigate and test.
Recent projects have included simulating billions of football matches; complex statistical modelling; realtime push over the web; and building a "myth buster" engine to test tacit football wisdom against hard data. Things move fast here, and we have new projects all the time. We’re always trying out the latest technologies and learning new things.
We're hiring!
Apply NowEngineering challenges
Stream processing
We compile data about football matches from many different sources, and must manipulate and store these streams with minimal overhead.
High-fidelity web messaging
Working with bleeding-edge web technologies, we relay huge bursts of data to powerful trading and modelling interfaces.
Parallel computing
With more and more data to process, and different iterations of our models to test, we are exploring ways to deliver a distributed, parallel infrastructure on demand.
High availability & redundancy
Covering football leagues in timezones around the world, it is important to deliver a zero-downtime experience.
Probability analysis
Combining sophisticated heuristics, our models deliver effective probability distributions to predict the outcome of football matches.
Time series modelling
We build complex regression models to analyse data trends and decay over time, and better understand historical data.
Cluster analysis
To better interpret the streams we work with, we apply various clustering algorithms to present a more detailed understanding about discrete and continuous data.