Business

EXCLUSIVE

Australian cricket looks to tech for competitive edge with new Microsoft deal

Australia's cricketers will soon get a much-needed tech-led competitive advantage after Cricket Australia signed a deal with Microsoft to become the first cricketing nation to use its new team and player performance platform to assess everything from form and fitness to strategy and team selection.

The deal was sealed on Tuesday as Microsoft's cricket-loving chief executive Satya Nadella made the Melbourne Cricket Ground one of his first ports of call on his maiden visit to Australia since taking the reins at the tech giant in 2014.

Up Next

Grant: I expect to be here in five years

null
Video duration
00:23

More Sports HQ Videos

Voges and Ferguson's embarrassing dismissals

Australia's capitulation was exemplified in the dismissal of its numbers 5 and 6, with both batsman left looking as bemused as onlookers.

The new platform will be trialled throughout the summer by coaching and support staff on the Test and state teams, using the latest in machine learning and predictive analytics technology to make practical use of the mountains of player data being tracked on performances and their level of preparedness for big matches.

While cricket fans have become accustomed to seeing players and coaches sitting on the balcony outside the pavilion looking at tablets and laptops, Cricket Australia's technology head Michael Osborne said the new system was a lot more sophisticated than anything that had been used before in terms of giving relevant context to data.

Mr Osborne said high-performance coaching teams led by Pat Howard had collected huge amounts of data in recent years on the players, their wellness, training and match workloads and ball-by-ball statistics, but the new platform was pulling the myriad sources together and presenting it in ways that would inform key decision-making.

"Our sports science guys have been doing some very interesting things trying to understand probability of injury, and things like that. But what we're finding is, for a lot of our frontline coaching staff, there's almost too much data, and to get in and understand what the data is saying is a daunting task," Mr Osborne said.

Advertisement

"The new platform takes this vast amount of data, provides an environment for our sports science folks to explore that data and find insights in it, and then provides a very elegant dashboard that will surface the trends and the information that will be impactful to the coaches."

Dashboard warns 'slow down'

In a demonstration of the new platform to Fairfax Media, Mr Osborne showed how a coach and selectors could track the squad of available players, and get real-time information on their fitness levels, as well as recent and historical performances.

Elite players have already become accustomed to providing daily updates through a mobile app on areas including how they slept the night before, and their general health. All of this data will now be drawn into the new platform, alongside performance metrics from Catapult Sports player tracking technology and ball-by-ball match data from Opta.

By monitoring the workload of players in recent games, coaches will have a meaningful, up-to-the-minute ranking of how ready a player is to perform in a match.

The Australian Test side has had particular problems in recent years with fast bowlers breaking down and missing games through injuries, most recently Peter Siddle, who is sitting out the Second Test against South Africa.

The new dashboard is expected to give advanced warning about when a player is passing their limits and needs to be rested.

I don't think the artificial intelligence will get to the point where it's selecting our national team for us.

Michael Osborne, Cricket Australia

The technology teams have been working with coaches, including a three-day session recently at the National Cricket Centre in Brisbane, to get insights into what data coaches would like to have, and are working on ways to demonstrate which players perform best in different venues, based on historical data, current form and variables such as weather and pitch conditions.

This theoretically gives a scientific algorithm, which could statistically say which players should be picked for each match.

'We will always need human selectors'

In a sport that values the knowledge and instincts of it captains and coaches, Mr Osborne concedes that some will be more willing than others to listen to the wisdom of a computer system. He said it will work best as a support, rather than a replacement for human insight.

"I think we will always need human selectors, we'll always need people to make those judgment calls," Mr Osborne said.

"I don't think the artificial intelligence will get to the point, or I don't think we'd want it to get to the point, where it's selecting our national team for us.

"To me, it's about helping improve the conversation so that when people are making these decisions, they have all of the information and the key data points right there at their fingertips."

The team and player performance platform is the most publicly visible aspect of a broader digital transformation under way within Cricket Australia, which also incorporates improved systems for administering cricket clubs down to the grassroots of local sides, and increased online coverage of matches online through CA's website and app.

Mr Osborne and CA's digital head Finn Bradshaw are also working on other areas such as investigating how the coaching dashboard data can be utilised in virtual and augmented reality applications for training.

The applications being used will all run on Microsoft's Azure cloud using its machine learning capabilities. CA is also standardising the vast array of about 24 different notebooks used by staff on to Surface devices.

Microsoft Azure chief technology officer Mark Russinovich said the tech company would learn a lot from its dealings with Cricket Australia, and would be able to apply its lessons to other sport codes and teams.

While Australia is the first cricket team to get on board Microsoft's platform, it has been adopted by Portuguese football club Benfica​ and a number of other early test cases in the US.

"We're working with Cricket Australia to understand what kind of machine learning algorithms can be brought to bear to give coaches better guidance or input into the decisions they're making about which players they put on the pitch and which ones should maybe take a rest and how they stack up against other players in the match," Mr Russinovich said.

"This is early greenfields ideas, with not a lot of past experience to point to or playbooks to pull off the shelf … but it is powerful stuff when a system can look at data on a match in progress and marry that to historical data where Azure machine learning can get insights out of that data that humans just can't get."

Advertisement