Could data science help hockey coaches draft a dream team, or divine the right player to offer a signing bonus? Research by SFU computing science professors Greg Mori and Oliver Schulte could help tackle this multi-million-dollar question.

Working with Montreal-based hockey analytics company SportLOGiQ, Mori and Schulte are analyzing quantitative data to determine how it factors into predicting performance on the rink.

An expert in machine learning, Mori first applies computer vision algorithms to hockey game footage to automatically detect and track players on the rink. Every player action is then assigned a value, from hits and shots to blocks and penalties.

From this data, Schulte can gain a fine-grained analysis of player performance based on statistical patterns. Their predictions have paid off. Hockey right-winger Vladimir Tarasenko, for example, came out high in Schulte’s rankings – the next season his salary increased seven-fold.

Far from making games predictable and boring, however, both researchers agree that analytics could actually make sports more interesting by having the right players deployed where they can perform at their best. It could even help teams modify their training regimens to predict and prevent injury risk.