Danny Palmer talks to IBM, Opta and the RFU about their use of data analytics to track every detail at the recent Six Nations rugby tournament
England fell just short of winning the rugby Grand Slam, losing the game – and the Six Nations title – to Wales on the final day. The Welsh win in Cardiff might have come as a shock to England supporters, but data analytics provided to the Rugby Football Union by official partner IBM allowed fans using the IBM TryTracker tool to gain insights into the game from up-to-the-minute stats, which made it clear England looked likely to end up losing this key fixture.
The TryTracker application, available to use live alongside all of England's Six Nations fixtures, provided real-time insights, including information about individual player performances, boosting understanding of what England – and their opponents – needed to do to have the best chance of winning the game. The TryTracker, a joint venture between IBM, Opta and the RFU revolves around data collection and analysis both in the run-up to and during matches, with the latter requiring both processes to be done in real-time.
Nick Shaw, head of digital at the RFU, told Computing the project was born out of the desire to be more innovative with the data produced during rugby matches.
"A lot of our steps into the digital world are based around innovation and personalisation," he said. "We tried to be more insightful, to actually bring the data to life and tell the story of what's happening in the game."
That is done through TryTracker's "Keys to the Game", three crucial targets for each team which, if achieved, greatly increase their chances of winning. They are decided through examining past performances of the two sides with data provided by Opta, before being analysed throughout the match to indicate if they are being achieved.
"It could be about tackles made or metres gained; there's a whole bank of 26 of these," said Shaw. "Then in the week prior to the game we look at the historical data about England, what have been the main keys in the past that have made either team win.
"Then we come up with three which take on board both teams and that comes out with these three keys. It's not definitive, it's not like if you achieve these you're going to win. What it says is if you achieve these three keys which we have predicted, they have a better chance of success."
[Turn to next page]
The importance of the keys was demonstrated at Cardiff's Millennium Stadium, when Wales beat England 30-3, successfully achieving two of the three targets in the process. According to the keys, Wales needed to force 15 turnovers, win 86 per cent of their own set pieces and make an average carry of 6.2m when the backs had the ball. The team forced 18 turnovers and won 90 per cent of their set pieces, putting them well on the way to winning. England, on the other hand, only achieved one of their targets while failing to complete the required 91 per cent of tackles or attempting the nine offloads the TryTracker suggested was needed for success.
For TryTracker to fulfil its purpose, large amounts of data required processing, with that task falling to sports data expert Opta, which has previously worked with IBM on the tennis-based Slam Tracker. Angus McNab, head of professional rugby at Opta, told Computing that it is still fundamentally based around "the science of wins", with all on-the-ball action tracked using X/Y coordinates and player identification numbers, which stay with that player for their whole career.
Opta's data collection process during matches falls to a team of five, consisting of two callers, two inputters and a checker, whose titles describe their roles. Each caller describes the on-pitch action for one team, with the data immediately entered by each side's inputter. The checker is able to clarify uncertain information using replays at different angles, ensuring high standards of accuracy.
"Having that fifth guy allows us to make sure that in a live environment everything is checked and verified and we get to as high a degree of accuracy as possible for us," said McNab, who emphasised the importance of accurate data collection to the process.
"The consistency of data is what's very important; there is no room for interpretation. We set our definitions and stick to them. Whether you agree or not with that as a definition, that is what you input."
Such is the strength behind the IBM-powered solution, data was entered and pushed through to the TryTracker within seconds, which allowed fans to keep up with analysis of the match in real-time.
"It's inputted into our system as soon as we see it," said McNab. "You're talking less than a second for it to enter our database. Then it's more on the speed of feed delivery, because there are so many data points that need to be processed. If we sent everything individually, they might not arrive in sequence. So what happens is every 30 to 40 seconds, there's a data pool from the IBM cloud to push the data."
[Turn to next page]
The pre-match analysis of historic data provided by IBM-produced algorithms, which – with the use of the Opta-provided data – judges each team's ongoing performance, giving insights into its momentum and chances of winning, as IBM sports analytics lead Michael Nicholson explains to Computing.
"For rugby we've got every international match that was played in the last three years, which gives us a large database of matches we can analyse," he said.
"We took that data and analysed which events were correlated with scoring points and winning games, and so that allowed us to build an algorithm that said if this event happened in that part of the pitch, that contributes to the team's overall momentum."
While professional sport might not appear to have much in common with enterprise IT, Nicholson believes TryTracker provides insight into how organisations could benefit from using big data.
"This is a great example of where sports is a metaphor for business, because using the keys to the match analogy, there's no reason why a business couldn't go through its historical performance and pick out the exact measures indicative of them doing particularly well and then tracking those as the calls come in, or as the sales leads come in and getting a real-time indication of how their business is doing," he said.
"If you can put in a process like this which would allow you to measure what's important right now, you could react much more quickly. And that's applicable to pretty much any business."
But while he admits there is still some scepticism surrounding the true value of big data analysis undertaken in real-time, Nicholson believes TryTracker shows that it works.
"What the TryTracker proves is that we're actually quite accurate as to what's important," he said. "So, for the England v Scotland game, England achieved all of their keys and Scotland none of theirs. For France it was 2-1 in England's favour and England won the game.
"We do have evidence to show we can pick out what's important and we can identify results and performance with very simple calculations that are carried out in real-time. So hopefully, the stuff that we're doing in sports can convince businesses that it is possible to do and we can remove some of that scepticism."
According to McNab, evidence that big data analysis can benefit businesses will only grow in the coming years.
"If you look at analytics, it's predictive models, and those kinds of things are used in insurance every single day," he said. "What is the science of wins, what makes a winning team? All of that is happening now and teams are looking at that now. In the next one to three years, I'd say, it will explode."