Expert View: How Can Sports Analysts Decide Which Metrics Are Important?

We talk to three industry leaders to find out


Ever since the explosion of sports analytics, teams have been dealing with a frankly ridiculous influx of data. Almost everything on a sports field can be measured, and with each new metric comes a potential new opportunity to gain an advantage over the competition. Sporting competition is so often decided by the finest of margins, and any extra insight analytics teams can offer coaches could be decisive.

With every useful metric, though, there are countless useless ones. Had Billy Beane selected something other than on base percentage to influence his batting recruitment - the first major use of statistical prioritization to make an impact in sports - the success story that is the Oakland As might not have happened. It’s the job of analytics teams, then, to pick out the gems and present them to coaching teams as actionable areas for improvement.

Off the field, too, analytics is having a profound effect on the way sports clubs operate as businesses. Analytics is helping clubs develop a more rounded view of their customer, and offer greater insight into how best to engage audiences and provide a better match day experience. Such is the sheer volume of data now collected by clubs across different sports, that knowing which metrics to focus on can be the difference between success and failure.

We spoke to three industry leading experts to get their take on how best clubs can pick out useful metrics.

Jeremy Loeliger, Chief Executive Officer at the NBL

Virtually everything is important… it’s a matter of figuring out how to use the information to maximum effect. As a case in point, we are looking at all kinds of biometric data that can be utilised live in broadcast – but the information in isolation isn’t necessarily interesting to a viewer (although it may be compelling for the team’s trainers). It is a question of understanding what stories you can tell with that data that will be of interest to viewers.

Robert Sorenson, Founder at Litespeed

I started in an area of more non-traditional sports such as speed skating, skiing and surfing. These sports have three facets to overall performance: athlete fitness, technique and equipment choice and setup. Each one of these contributes to the data in their own way and untangling, for example, whether an athlete has a technique deficit or poor equipment choice or setup is incredibly difficult.

These sports have no hard and fast rules and the parts discussed reside to some degree or another in tribal and experiential knowledge. Decisions are made based more on gut feel and the quality of relationship between the athlete and coach.

Using data to find that extra bit of performance, correct the right problems and measure ongoing progress is going to be the most difficult aspect sports analytics will face. As exciting as it was to first get the data, the unexpected realization of how to build an effective collaborative framework to use it is a daunting challenge.

Adam Karg, Senior Lecturer and Course Director at Deakin University

There is some important background here around the use of quality data, valid cases and investing in the right expertise and techniques. Often a danger can be working on poor or limited data which impact results, and resultant strategies. Using data mining and predictive modelling approaches can manage extraordinary numbers of variables – so volume of data presents less of an issue than in the past. Likewise, open text and composite metrics become more relevant here to add value to current modelling.

Most vital is balancing an approach that both confirms logical or known patterns (i.e.; the variables we’d expect or know are important), but also combines exploratory techniques to uncover hidden or less obvious patterns and metrics that can assist marketers. Above all, one size does not fit all (i.e.; not all fans or markets behave the same) therefore adopting non-specified models or insights from other contexts can be dangerous.

Vision small

Read next:

Big Data Forecasting In Pharma