In the lead-up to last year’s Australian Open men’s final, Stanislas Wawrinka had a one in six chance of defeating his opponent, Rafael Nadal. If he could win 32 per cent of first serve return points, at least 46 per cent of medium rallies and more than 68 per cent of points on his first serve, Wawrinka’s chances would greatly improve. He hit those targets out of the park.
Like Wawrinka, the Australian Open pulls some impressive numbers. In addition to the more than half a million spectators who attend the event’s annual two-week run in Melbourne, Australia, there is an astounding breadth of data collected, analysed and shared.
Before each match, the event’s governing body, Tennis Australia, has its technology partner IBM analyse 41 million data points collected from eight years of Grand Slam matches.
So how is it done?
Mining for data
IBM spends an intense two weeks gathering data from three main sources – historical data from international Grand Slams, statisticians on court and the Australian Open umpires. Each umpire has a PDA (personal digital assistant) called a CHUMP, where all the action on court is recorded.
The CHUMP sends real-time data to the tournament server room, located metres from Rod Laver Arena. It’s a small room with lots of computers and a few IBM staff. From here the number crunching begins.
A global private cloud computing system allows Tennis Australia to scale this data in real time and ensure continuous availability.
In the finals match between Nadal and Wawrinka last year [data], showed that Nadal needed a certain percentage of first serves and return points. He didn’t meet the crucial actions and he lost.
“We have three service delivery centres in the US that support the cloud, so if one falls over the other two can pick it up as part of a disaster recovery plan,” explains Angela Gallo, corporate sponsorship and hospitality leader at IBM Brand Systems, Australia and New Zealand.
Made for sharing
Data from each match is shared with multiple sources, including the scoreboard, the Tennis Australia website and app, and the Australian Open Media Room, where hundreds of journalists prepare their stories.
“They receive information via a system called AO Vision, which enables them to see real-time and historical data and video footage from each court so they can pause, rewind and watch any match,” says Gallo.
The data can also give players a winning edge. Each player receives a summary of analysis via an intranet to help them fine-tune their game.
Data from each match is analysed and sliced into bite-sized points of interest. Each player’s momentum is visually mapped on the Australian Open website by IBM SlamTracker, which displays faults, serve speed and other key statistics in real time.
“If someone has scored an ace, for example, we can also show what their history of aces has been,” says Gallo.
The past eight years of Grand Slam data is also mined to produce “Keys to the Match”, built on IBM's predictive analytics technology. Before each match, the system runs an analysis of competitors' historical head-to-head games and comparable player styles to show audiences what a player needs to do to win.
“In the finals match between Nadal and Wawrinka last year, Keys to the Match showed that Nadal needed a certain percentage of first serves and return points,” says Gallo.
“He didn’t meet the crucial actions and he lost.”
Data is also analysed to track social media winners. Ana Ivanovic’s win over Serena Williams last year, for example, drew 123,000 tweets during the match at a rate of 7,232 tweets per minute. The final match result received the most retweets of the tournament from the official Australian Open account.
“Tennis Australia likes to encourage fans to get behind their favourite players and we have a Social Leader Board based on the most positive tweets. This helps Tennis Australia to identify the most popular player,” says Gallo.
“It’s all reflected in the data.”
643,280 people attended the Australian Open in 2014
4,200 racquets were strung at the Stringer hut in 2014
More than 35km of cabling is wired throughout the stadium
8,595,362 Twitter references about players were tracked during the 2014 tournament.
The fastest serve in the 2014 men’s final travelled at 217km/h
The Australian Open website attracted more than 15.5 million unique visits during the 2013 tournament
More than 41 million data points identified patterns in players and their styles
An average of 60 IBM staff work on the Australian Open
The 2015 Australian Open overall prize money will be a world-record $40 million
Where the Australian Open's $44 million purse fits in the global prize battle