NEWS

Retaining a human touch in sport and in business

Eleanor Dallaway, Editor of Infosecurity magazine

Professionals in both the business and sporting worlds increasingly exploit data analysis to inform team selection. But how exactly does the technology work, and what does the future of data-led team selection look like?

Bill Gerrard is professor of business and sports analytics at the University of Leeds. Coaches at both national and club level have a huge amount of data available on player performance, Gerrard says, including tactical data on player actions and fitness data: “This allows coaches not only to assess player performance in individual games but also to detect potential trends, like a loss of form.”

This, of course, is integral in aiding team selection. But, explains Gerrard, this is far more useful for national coaches who have a bigger pool of players to select from: “Data analysis offers a first cut to identify potential national squad players whom the coaches should scrutinise in more detail.”

In non-team sport, data analysis is used in a different way. Jonathan Leeder, physiologist at the English Institute of Sport (EIS), describes how they use data and technology to “really pull apart what it takes to win an Olympic medal”. This intelligence is then used to craft the training and resourcing of that athlete “to better prepare them and increase their chance of winning”.

Unlike in team sports where competitor analysis is key, single athletes use data in a different way. “You’re trying to get from A to B as fast as you possibly can, and aren’t too bothered about what other nations are doing – you just need to be the fastest.” In this instance, Leeder explains: “Historical data is used to predict the winning time for the next Olympics. You then invest resource into what it takes to do that time and train athletes appropriately. It’s you against the clock.”

No sufficient algorithm

Using data to recruit high-performance teams in the business world is arguably less reliable.

Advertisement

At Propel London, a specialist recruitment database technology is used to segment candidates according to multiple factors, which allows high volumes of candidate profiles to be sorted in a timely manner. James Webb, the firm’s managing director, does counter that “relying solely on technology to pick out specific data points would run the risk of the exclusion of certain people who may not have presented the information in a specific way”, which is why the human element in recruiting frequently outweighs the data.

Scott Ross, chief technology officer at global marketing and technology agency, DigitasLBi, describes how their recruitment team utilises commercial tools and proprietary extensions to simplify the process of candidate data aggregation, and to provide scoring on basic metrics such as skill proficiency, demand potential, and acquisition cost. “We also combine this data with tenure analysis, helping our recruitment team identify candidates who may not be actively searching, but are on the lookout for their next opportunity,” he explains.

Ross admits that his use of data to recruit is largely limited to human-driven interpretation of candidate performance data. “Simply put, there is neither a sufficient quantity, nor more importantly quality, of data points to gain real benefit from an algorithmic analysis,” he states.

Behind every computer is a human

The importance of human-driven interpretation, as raised by Ross, is one not to be under-estimated, and it’s worth exploring how, even when relying on data analytics, human involvement is still required.

“People are trying to make it so that a computer can do the analysis,” says Leeder, but “as soon as you lose the human interface, that’s where it falls down. You still need an expert to interpret the data, you need a detailed understanding to realise what the data needs.” It’s for this reason that Leeder maintains that the requirement for human interaction will never be lost.

Coaches need to not just interpret the data, but also help identify where additional information is needed. It is the role of the technical teams and analysts to identify where additional information could come from and then identify patterns in that data that could be of use to the coaches.

Harry Jones is the co-founder & chief technology officer at Top10. As he looks to recruit a high-performance team, he places only finite value in algorithms. “They only tell you half the story. Humans will always be needed to analyse other humans for job suitability,” he explains. “No matter how technical recruitment gets, personality will always be important.”

Alec Dent, partnership manager, at BlaBlaCar, explains that it’s impossible to replace the human element in recruitment. “It’s all about fit – and an algorithm just can’t predict that yet. Still, we’re optimistic that as they develop, algorithms could eventually play a bigger role in our recruitment process.”

Moneyball: the future

But Dent’s vision seems to be, at present, just that. Consensus suggests that data analytics in business team selection is still a fairly immature field. While many CTOs and recruiters are using technology and data metrics to identify relevant candidates, more in-depth candidate selection – and indeed, hiring decisions – remain very much an inherently human process.

“In the future, I’d love to see a candidate equivalent of a slugging percentage, or combination of multiple discreet statistics that point us towards not just qualified candidates, but the right candidates for us,” Ross envisions.

In sport, experts project that new technical advances will continue to increase the odds of success, and the influence of data in helping to define team selection will continue to grow.

“The future of data-led analysis is sort of ‘Moneyball’; all done on probability,” predicts Leeder, who is quick to counter that this doesn’t work for everyone. “Mathematically minded people will love Moneyball-style stats and probability, whereas others will be more into changing the culture and knowing their art.”

Love it or hate it, data is here to stay, and Seddon predicts that as wearable technologies become less intrusive to players, real-time team performance analysis will increasingly play an instrumental part in shaping tactics.

Evidence-based coaching data analysis that supports the decision making of coaches is still in the early stages of development, says Gerrard. However, “coach-led analytics will grow, as will the quantity and quality of the data available to be analysed”.

The future of data-led team selection, according to Joseph Turley, director of marketing and communications at real-time performance analysis vendor, Performa Sports, is in “encompassing a greater appreciation of the behavioural aspects of performance. An interesting dimension in this regard will be the ability to measure and assess player compatibility across a spectrum of key performance indicators to map out the optimum combination of individuals (by type) for the most effective team collective.”

The final word must go to Angus McNab, head of rugby at sports data analytics organisation, OptaPro, who brings it back to the human element. “Selection isn’t quite like Championship Manager – these are people we are dealing with, and we are a long, long way from being able to factor certain elements into an algorithm.” The real advances in the future, he predicts, will be in injury prevention, player management, and welfare, to ensure the best team can get out on the pitch as often as possible.

More News