How The Sportsbet Data Science Team Are Having An Impact With Standout Projects

We live to innovate at Sportsbet and that is especially true in our Data Science team! We use rich insights and feedback to build new and exciting customer experiences, and there’s always something fresh and cutting-edge on the table. The team have contributed to a number of impactful and standout projects.

Being part of Sportsbet’s Data Science team is incredibly exciting. We’re constantly growing, and we lead the market with game-changing innovations and customer experiences – powered by the insights and modelling we’re doing with data.

With access to the best tech stack in the data science, machine learning and data engineering space, we're always coming up with something new and exciting – it’s all in a day’s work here!

Here’s what members of the Data Science team have to say about working at Sportsbet. Get to know them and some of the cool project work they’re doing, below!

Andre Easom, Head of Data Science

The one that really stands out to me is the launch of Same Game Multi, for AFL.

My role was to build the model that powers the pricing.

It was a whole new type of betting experience that we were putting in customers’ hands, so figuring out how to build and test the model came with a range of challenges. Happily, it has been a runaway success with customers, which many competitors have rushed to copy.

I still love using the app to generate prices for all sorts of crazy combinations, knowing it is all powered by a model that I built!

John Hannebery, Data Scientist

I am blessed to have been heavily involved with the world of Safer Gambling.

We are industry-leading with our multiple machine learning models that power our approach to Safer Gambling. This enables us to really scale our approach in ensuring customers have a great but safe experience. 

The concept of Safer Gambling from a modelling perspective is extremely difficult, so we are always thinking about how best to improve our approach. We understand the importance of owning our models once they are productionised, especially for a use-case as important as Safer Gambling.

A big win was setting up a proper MLOps approach. This includes but is not limited to ensuring data is available for the model to run and make as accurate predictions as possible, and if something goes wrong with the model, we have a system set up so that on-call engineers will be notified and we can solve the problem immediately and collaboratively.

Mia O'Dell, GM Data Science

We're on a journey of increased personalisation for our customers at Sportsbet, and the pinnacle of personalisation is through machine learning and data science.

So much of what we do has a huge impact on the business.

We build models that drive industry-leading app features. We build the core models that drive the odds and prices that our customers can wager on (i.e. what are the odds that the Yankees beat the RedSox tomorrow?) Most importantly, we build the machine learning models that help keep our customers safe.

Kaushik Lakshman, Head of Data Science

The most exciting project I've been a part of so far has been Powerprice.

A new and unique generosity mechanic that only Sportsbet offers.

Our customers love it and tell us they’re amazed that we give them amazing offers on what they love to bet on. Due to the unique nature of our domain, it was quite tough from the data science perspective, so the fact we ended up delivering such a great experience that customers can’t get elsewhere is a fantastic feeling. 

To build such a product requires an incredible amount of collaboration, and while this has been challenging, we’ve got a great collection of people who all have a common goal, which certainly made it much easier.

Paul Foerste, Senior Data Scientist

The first project I ever worked on at Sportsbet, 12 years ago was building an in-depth AFL model.

Prior to this, the AFL offering was a handful of markets that needed to be manually priced by the traders, and once a game was live there was only the ability to offer two or three core markets. This meant that any price changes (due to an injury for example) were extremely laborious.

After a collaborative effort with the AFL traders and developers, we delivered the AFL model, which was the first model of its type in the group delivered from Australia at the time. Once the model was productionised, overnight Sportsbet's AFL offering went from the handful of manually priced markets, to over 200 derivative markets – all automatically priced, with the majority of these available to be priced when the game was live.

It was the first time I had built a model of such complexity so there were definitely challenges and plenty of second-guessing along the way but with the help of colleagues both local and global we were able to deliver a model used by the business for many years.