With the continued rise in sport betting Gaming Eminence reached out to Roger Creyke Chief Technology Officer and Co-Founder at Bettormetrics as he answers questions for us around operator data challenges, infrastructure, governance and the relationships across external data providers, internal algorithms and staff.
GE) Throughout your time working with sportsbook products, what do you think have been the biggest challenges operators have faced with getting value from their data? How do you think these challenges have been resolved and complicated during recent years?
RC) Our industry has a real problem with obtaining and mining its data efficiently. Third party data silos have always been an issue. It's rare that a sportsbook operator has control over all of the products and services they run. It is very common, even for the biggest operators, to tie together many disparate systems from suppliers. Our core product (betting opportunities) is outsourced. Scout data is obtained from live matches by one company, passed on to another to generate algorithms, and another to add prices. These companies often specialise in a small number of sports. The web of dependencies goes deep and broad. How these products are generated, what the inputs are, how the decisions were made, all of these things are often hard to obtain. There is no standard format, and no guarantee of access to this data by the sportsbook operator.
In highly regulated jurisdictions, for example physical casinos, it can be tough to even get access the data. There is huge variety in the structure of data that represents the state of a sporting event. Scoreboards for individual sports differ hugely so therefore domain expertise is required to mine these efficiently. Some of the biggest platforms in our industry are over 30 years old and have reached a level of maturity where extending them requires a very compelling business case. We have a lot of catching up to do with fintech.
GE) How would you advise a sport betting operator to look at its approach to data infrastructure? What does bad vs good look like and what are key areas operators have to address to have a robust structure as it relates to data governance?
RC) There seems to be a push towards getting all the data into a lake rather than making all the data accessible. I would recommend that technology executives in these industries stop telling all their teams where to put the data and in what format, and instead incentivise the instinctive exposure of it proactively. Get everything into a stream so when you need to build things people can access it. Kafka, AMQP, whatever it takes. I would advocate for moving systems much more towards eventual consistency, which forces distributed logs and messaging systems into the architecture. This opens it up for introspection and plugging more steps in later, as well as peeling away data streams for analytics, ML, regulatory steps or future components of the system.
Don't put all your eggs into one basket. Lambda, hotpath/coldpath, Kappa architectures, these are all valid but none solve all use cases. Empower your teams to deploy technologies not just stream processing jobs or tables. Bad to me looks like a data lake filling up and teams having no access to data until it is at rest for batch jobs. This kills real-time use cases. Good looks like a positive, open data initiative, with sensible PII requirements. If you don't proactively expose data, teams simply won't make it available efficiently when you need it, and progressive, data driven projects won't get off the ground when then business wants them because the quotes are too large and everyone is too busy. You have to make a culture of data abundance.
GE) What do you see as the real rules sportsbook operators need to think about regarding data governance?
RC) Of course I could go on about ISO 27001, GDPR, best practices etc. I will assume everyone is at least trying to follow these rightly pessimistic and well documented defence mechanisms.
On the side of optimism and opportunity, please make any data that is not highly sensitive available to your teams. Don't be afraid of them knowing what your biggest sport is, or territory. Engineers are great at prioritising work and product design but they need as much context as possible. They will automate more, and the product will appear more intelligent to your users because of it. If your engineers do not understand your priorities and the business context of your priorities, they will not consider it in every small decision they make on a daily basis. It's these small decisions, over many months, that take your platform and operation to the next level of data driven proficiency, and prevent the need from just throwing more and more humans at problems such as player segmentation, customisation, product optimisation, and of course responsible gaming.
GE) Where do you see the importance of how operators approach the dynamic relationship of their external data providers, internal algorithms, and internal staff to ensure a successful sportsbook product?
RC) It's critical that these relationships are maintained and well oiled. A sportsbook is often like a supermarket. It has no products other than the ones your suppliers give you. If you want to keep the right things on the shelves for your customers you of course need to manage this web of dependencies carefully. I would say, ensure you don't assume everything is operating nominally just because it is outsourced. You are judged on the quality of the product and your profits are dependent on it too. There could be money on the table. A chef checks their produce regularly. So get great monitoring tools in place. Then you can put healthy pressure on your suppliers to keep the quality up. Never assume the quality remains consistent over time, nor that the rest of the industry isn't moving forward relative to you. The more data and feedback you can give to your suppliers, internal and external, the better their products (and your profits) will become.
About our contributor
Roger has spent 15 years building software, teams, and studios in the gambling industry, with the last decade focused on sports betting. He is a co-founder of Bettormetrics, an analytics company which provides big data observability and insight to the trading departments of some of the worlds biggest gambling companies. Visit https://bettormetrics.com/