Up until now, anybody that wanted to perform some analytics on BAR matches had to scrape data from the Replays website and clean it up themselves. Now, we start to publish documented, public datasets in easy to consume formats with information about public matches.

Check out github.com/beyond-all-reason/data-processing to see how to access the data.

The datasets are build twice a week from Teiserver and Replay databases. The processing pipeline is fully open source so if you would like to help extract more fields (currently it’s only very basic ones), have any bugs to report, or share more examples on how you use the data, your contributions are very welcome.

We hope this will help with development of improvements to the ranking and balancing but also help quickly answer questions like How often players pick Armada vs Cortex?