Injured players cost sporting clubs huge amounts of money and can result in lost games.
Data and predictive modelling are helping some teams forecast injuries before they occur.
“Big Data” is a huge buzz-term in business at the moment – many industries are furiously trying to understand how it can benefit their business.
We found out how it works in rugby, thanks to New South Wales Waratahs’ sports science and research manager Maria Nibali, who spoke at tech conference CeBIT in Sydney this week.
The main ways the team uses data are to prevent injuries, decide on tactics and recruit players. They generate 9,000 or more data-points each game, and almost the same amount from training sessions between matches.
It’s mind-blowing. Here’s what the club is doing.
Rugby is a high contact sport which requires strength, agility and a lot of power.
The nature of the game means players are injured frequently. In the 2013 season on average about eight players were on the bench due to injury each game.
In 2013 there were 24 injuries from 18 players, between three and nine players were unable to play each week and a cumulative total of 78 games were missed as a result.
“Of our 18 injured players last season, eight of those were current Wallabies members,” Nibali said. “Across those eight players they averaged 3.25 games missed.”
Nibali said this injury rate costs the club, based on players’ salaries, about $111,500 per game, or about $2.9 million over competition season.
“Injuries are a huge concern for us,” she said.
The club classifies injuries into two categories: contact and non-contact.
“One of the areas that we’re sort of struggling with is what do we do about contact based injuries it’s very difficult for us, one to prevent them, and two to predict them,” she said.
However non-contact based injuries including muscle strains and bruising, are becoming easier to predict. The Waratah’s understanding of what each player’s injury risk profile is and how to manage that risk is becoming clearer.
“There are several monitoring areas that we have in place currently,” Nibali said.
The players wear GPS units at every training session and game.
“We start to get statistics on number of accelerations, decelerations, metres run and times spent in different velocity zones and we can also start to look at collisions,” she said.
The club is also using force profiling to show an athlete’s ability to both produce and withstand force as well as providing information about the status of the player’s muscular system.
“All these measures collectively give us an idea of the work load the players are experiencing either during training or during games,” Nibali said.
“This information is also used by our strength coach to prescribe exercises for the following week for those players in the gym.
“Everything that we’re doing is very much at an individualised level.”
The computer modelling looks back at the previous 21 data records to understand a player’s rolling average.
“Instead of trying to get a snapshot of when an injury occurs, we’re looking at what’s happened on the day, we’re looking at what’s happened in training or game situations three weeks prior to that,” Nibali said.
“What we’re then asking the model to do is identify injuries that are in our medical screening database and put that through the model and try to identify which elements or combination of variables are most sensitive to being able to detect the prevalence of injury.”
In 2012 the system was able to predict three key players, who were all Wallabies players, were were likely to be injured. All three were injured in that season.
“It was really exciting for us to see, I guess, the capacity of predictive modelling and what it could be doing for us,” Nibali said.
Being in the right position at the right time and reading a game correctly is a huge advantage in rugby.
Nibali said being able to use data analysis is improving the team’s performance, helping the Waratahs make more line breaks, out run opponents, lift tackle numbers and to an extent predict wins.
“For every game our performance analyst ends up with over 9000 data points,” she said.
“The magnitude of the data that we’re getting out of the games is quite insane.”
Nibali said analysts are looking for patterns and are closing in on key variables which indicate a winning outcome.
“Our style of rugby in 2014 and really in the last 18 months is an attacking style of rugby,” she said.
“We’re seeing [anecdotally] that is a key variable in predicting wins.
“Our coaches are onto a winning thing here with the attacking style of ruby.”
The club’s data analysts are also examining other variables to predict wins including how winning scrums and line-outs in different areas of the field effects results.
The Waratahs is using data points to select the right players for the team when recruiting, not unlike what played out in the movie Moneyball which used data to acquire the types of players needed to win games.
The organisation is combining game and injury prevention data to help make better player selections but Nibali cautioned “there needs to be some intuition behind it”.
“We need to be cautious of how we interpret the data,” she said.