Greyhound Racing Victoria (GRV) is the statutory body that regulates greyhound racing in the state. GRV’s core system is proprietary, based on Microsoft Azure and specifically designed for the industry, the niche nature of which makes obtaining commercial off-the-shelf applications difficult.

The organisation relies on the platform to support revenue generation, most critically through the provision of data generated by around 1,300 races in Victoria annually, which is then used to drive wagering either directly or through partners such as Tabcorp.

Data on every race, including dates, times and outcomes, is distributed via a system called the National Data Exchange, which forms the nucleus of the Greyhound Racing Regulator industry in Australasia.

When chief information officer, Kate Romanova arrived at GRV in 2021, the organisation’s systems were still using SFTP file drops and CSV files for data transfer. Wagering partners would log onto an SFTP server and load files that were then picked up by a SQL Server process and imported.

“As there was no feedback mechanism, partners would then wait and hope that the data was picked up and processed. At the same time, participants would need to log onto another SFTP server and pick up data files dropped there by another process, to process their own systems,” she recalls.

The ageing systems and processes had not kept up with changes in technologies and methodologies and were suffering from a large amount of technical debt, costing GRV on average one day of a person’s time every week to maintain. There were also instances where a failed process cost GRV millions in revenue because data was not available.

New innovations

Innovations introduced by Romanova and her team have helped the Victoria racing industry move towards its strategic goal of acting as one holistic entity. This has previously been challenging, in part because most jurisdictions are funded by different government bodies with substantial differences in architectures, technology stacks, capabilities and resources.

One key development, Romanova adds, has been a rules engine which can be used to determine whether a greyhound is eligible to race. For instance, a dog that runs today can’t run tomorrow in its own or any other state. The bespoke system was written as a proprietary offering only after all other options were exhausted.

“For the time being, it does not enforce the rules of racing because these are based on human-determined factors. For Phase 2, we are investigating whether AI can assist in determining the outcome of a rule,” she explains.

“The only reason this is possible is because we have designed the rules engine architecture in such a way that it is technologically and system agnostic.”

A new set of APIs operate in concert with each other to provide services collaboratively or separately. 

The team has also deployed the Azure Confidential Ledger (ACL) blockchain and as more data is gathered and AI and machine learning capabilities are rolled out, which, in conjunction with the rules engine will enable the organisation to review outcomes of previous decisions in order to make better ones, Romanova says.

The innovations were designed so that each component could be used either on its own to perform a single task or together to accomplish a more complex task. Each component can be used, not just by specific applications, but by any application that can call an API, making them highly flexible and system agnostic.

“Substantial value is derived from this design and what is particularly unique is that each API has a user interface (UI) associated with it,” says Romanova.

“This means that not only can we implement, as an example, a rule of racing in any system we choose, but we can also use the UI to test that rule against real life scenarios. This means rules can be checked for correctness by the people that are responsible for managing them”.

In addition, other jurisdictions can use the interface to validate that GRV, as a regulator, is implementing the rules correctly or alternatively that their interpretation of a rule matches with GRV’s.

“Where there is a mismatch, we can customise the rules engine so that it supports more than a single version of the rule.”

One aspect that was missing was the ability to retrospectively view our data and determine whether it had been tampered with.

“Every system we looked at had some element of vulnerability inherent, but the blockchain gave us the level of comfort we wanted in ensuring that any data that entered our systems was captured in an auditable system and could not be realistically tampered with,” she says.

Support requests down, industry perception up

GRV has reported that support requests from internal and external users are down from two to three per week to less than one per month – usually an external technical team requesting assistance to implement rather than a system fault.

“Not all of the requests we previously received were due to system failures, some were due to timing issues with other states, but they all impacted our spend and reputation. Now, we have an ecosystem that has not failed in more than 8 months, and which costs less than one hour of FTE per month,” says Romanova.

She adds that GRV has also been able to retain tech staff for longer because it has been able to offer greenfield development in systems that have a demonstrable benefit to the business, as well as encourage the team to explore interesting technologies like AI and ML.

Resource and cost efficiencies have also come from the reduction in the number of additional applications required to support the business. The system is primarily maintained using internal GRV resources which, in turn, has reduced dependence on external vendors and associated costs.

“Intangible benefits include the shift in how GRV is perceived by the industry and by people that use our data. We are now perceived as being more open and transparent and as innovators,” says Romanova notes.

Byron Connolly