Over the past two decades or so, statistics in sports have evolved dramatically. Technological advances have made it possible to collect more data than ever about what is happening on the field. But how do we connect the dots of easily quantifiable data points to see the underlying performance indicators? Dr. Bill Gerrard was one of the early pioneers who sought to understand what really made sports tick.
Dr. Gerrard is a Professor of Sport Management and Finance at the University of Leeds, in the UK. His research and professional work have dealt with some of Europe’s most popular sports: cricket, football, and rugby. In addition to teaching, he is currently working as a data analyst for AZ Alkmaar in the Eredivisie, Netherland’s premier football league. He has also worked with some of Europe’s top rugby leagues and teams and other football clubs in similar capacities. Beyond performance data, Dr. Gerrard is also a leading expert in player contract valuation and league financial analysis, drawing from his background in economics.
We got a chance to speak with Dr. Gerrard about the boom of the sports analytics movement, his transition from economics, to academia and finally into the world of professional sports, and about his predictions for how this field will continue to grow and evolve in the years to come.
SportTechie: Prof. Gerrard, let’s start at the beginning. Can you tell us a little bit about how you first developed your love of sports, and especially for the analytical side of the game?
Dr. Bill Gerrard: My love of sports dates back to my childhood. I wasn’t particularly gifted with skill but I loved playing all sports and was hugely competitive. I mainly played football but I also played cricket, basketball, tennis, golf, pool and snooker. And I followed everything including F1, rugby union, rugby league, athletics, cycling, etc. I always took an interest in the numbers in sports – I acted as a scorer for the local cricket team when I wasn’t playing, and religiously tracked the football league tables. I studied economics at university with a strong statistics component, and my first job was as a business analyst at Unilever. After I moved back into academia as an economist, I had greater freedom to pursue my own research agenda and gradually realised that there were possibilities for applying economics and statistics to sport. And that is how my involvement in sports analytics started in the mid-90s.
Growing up in the 70’s and 80’s, many sports were still unaware of the advantages that advanced statistics could bring them. What was it like to see that movement begin to take hold across different sports and leagues around the globe?
BG: In the 70s and 80s there was very little interest in the analytical side of sport other than tracking the basic stats particularly batting and bowling averages in cricket. I remember seeing an article when I was at Unilever on modelling gate attendances at John Player Sunday cricket matches and that fascinated me. It was probably reading that article which first sowed the seed for me of the possibilities of applying data analysis to sport. To see the growth in analytics in sport over the last 30 years from virtually nothing to now being seen as a vital source of competitive advantage in a huge number of sports is truly amazing.
Get The Latest Sports Tech News In Your Inbox!
In fact, you’ve worked with one of the earliest pioneers of sabermetrics, Billy Beane, the subject of Moneyball [by Michael Lewis, published in 2003 and adopted into the 2011 film of the same name]. I’m sure that was an exciting opportunity. Any stories you’d like to share with your time there?
BG: I’ve been very fortunate to work with some great people and great organisations over the years. Billy is a great guy who I hold in high esteem as a professional and as a friend. His advice and encouragement have been incredibly helpful. Seeing the real story behind the scenes is quite different from reading the book and watching the film. People forget that the book and film are only “based on a true story”; there’s a lot of dramatic license in both particularly the portrayal of the relationship between Billy and his scouts as very conflictual. In reality, Billy uses all the available evidence – numbers and scouting reports. Moneyball is really about being systematic in using all the available evidence rather than relying on conventional wisdom and gut instinct.
I’ve never seen Billy throw a chair or chew as much gum as Brad Pitt does in the film, but the film was bang on about Billy not watching games. I sat in Billy’s seat watching my first game at Oakland while Billy went to the weights room. Billy’s rationale is that he wants to watch the game devoid of emotion knowing after the event. He tracks the scores during A’s games but never analyses the video and stats until afterwards. As a result, when I worked at Saracens, I made a point of never listening to or reading the coaches’ post-match comments if at all possible. I wanted to remain independent and analyse the data unaware of what the coaches thought I should find in the numbers.
A lot of your academic background focuses on business and economics; areas that aren’t necessarily as obvious on the field of play. Can you tell us about some trends with how players and teams are reacting to increasingly global fan bases? How have things like media deals and player salaries reacted to one another in this macro-economy?
BG: It’s true that my early work focussed more on the economics and finance of sport but for the last 10 years or so I’ve only really worked on analysing player and team performance. But much of my findings and projections on the business of sport still remain valid, particularly the trends in player salaries and transfer fees in football. Salaries and transfer fees are driven by team revenues and increasingly these are driven by the value of media and image rights. Teams should always work on the basis that if their revenues go up by x%, then expect player salaries to go up by x%. It’s not rocket science to run a pro sports team the right way.
Teams must ultimately balance the books by ensuring that their total costs are sustainable given their total revenues. My local football team, Leeds United, forgot that basic lesson, borrowed excessively to chase the dream of sporting glory, went bankrupt and 15 years later are still paying the price of crazy economics.
Outside your role in academia, you’ve kept yourself active in professional sports, either with rugby (Saracens, Super League) or currently in football, at AZ Alkmaar. With the amount of new data constantly being collected, how do you search for relevant correlations and present that data to teams and coaches?
BG: The role of an analyst in sport is no different in essence to that of analysts in any business sector. There is a massive problem of data overload in every sector including sport because our ability to collect data is accelerating every year. Analysts are there to solve the data overload problem by applying analytical techniques to extract actionable insights. Every week I receive around 1 million data points just on the Dutch league. It’s a challenge but one that I hugely enjoy. There is nothing more satisfying for a data analyst than identifying patterns in the data that coaches can use in developing a game plan which contributes towards the team winning a game.
A lot of what I use is pretty basic statistics – descriptive statistics, win-loss analysis (i.e. t tests), correlation and regression. But I also have used more advanced techniques such as factor analysis, cluster analysis and logistic regression. I always try to keep it as simple as possible since you must always be able to translate the statistical results into practical advice, and you must be able to provide practical explanations of the patterns you have found in the data. No coach is going to develop a game plan on the basis of statistical significance. The results must have practical relevance.
A growing area of sports data collection is through the use of player wearables. You’ve done extensive work on correlating things like speed, reaction time, distance covered, and many others that can give insight not only into player performance, but into overall health as well. What are some of the biggest benefits, and limitations, of the increased use of wearables in sports?
BG: In the sort of team sports I mainly work, the invasion-territorial sports such as football and rugby, the spatial dimension is crucial. So the particular wearable that has impacted on my work most is the GPS monitor. We now have great data on pitch location, distance covered and speed of movement. Most of this data has been used by sports scientists to measure physical performance. But what I’ve found is that there is no simple linear relationship between physical performance and winning games, in much the same way as there is no simple linear relationship between share of ball possession and winning games.
As always it’s the QUALITY of what players do, not the QUANTITY. So great care needs to be taken in analysing GPS data and interpreting the results. Probably the greatest benefit in the use of wearables is monitoring training workloads and having much better knowledge of the thresholds above which there is a significant increase in the likelihood of soft tissue injury for individual players. This helps teams to create personalised training programmes rather than the traditional one-size-fits-all approach to training.
And how do you approach “old-fashioned” coaches/owners who may be skeptical of this type of research? Do you think it’s possible to accurately quantify everything going on during a match? Or do you still use subjective measures like “the eye test” or “a gut feeling” to balance out what the data tells you?
BG: Data analytics gives the “truth” but never the “whole truth”. The analytics revolution is not about using only analytics but rather about realising that analytics has an important role to play alongside other sources of information. It’s one source of marginal gains. Top-class coaches, like all human beings, are subject to biases particularly when drawing conclusions in complex situations in which lots of different types of information has to be considered. And trying to evaluate the overall performance of players in complex sports where players are performing many different types of activities is incredibly difficult.
There is a tendency to over-emphasise certain key moments in a game and put too little weight on everything else. In team sport there tends to be a bias towards attaching greater importance to attacking contributions than to defensive contributions because attacking success is much more observable. Attack is about creating and converting scoring opportunities. But defence is quite the reverse; if you like, defence is about stopping things from being observed. So a lot of great defence is often ignored because it isn’t headline grabbing in the way that scoring is.
You recently started a blog, Winning with Analytics. What was the inspiration for that? Did the growing awareness of the role advanced stats in sports, not just in teams but in fans as well, play a role in your desire to start putting your work out for the public?
BG: Yes I have started a new blog and I’m really pleased by the positive reaction to it from analysts, coaches and people with a general interest in sports analytics. I’ve learnt a lot over the years on how to be a more effective analyst and the blog is an attempt to share that experience.
I’ve experienced the frustration that all analysts have felt at some point that they’ve done great analysis but no one is listening. But what I’ve learnt the hard way is that the analyst must bear most of the responsibility for being ignored because one way or another they must have failed to build up a relationship with the coaches. If coaches are ignoring the analysis, then it’s very likely that the analyst has not really understood what the coaches are trying to do and what sort of analysis would be most useful.
Being an effective analyst is as much about the soft skills of building great working relationships as it is about knowing statistical methods. Too often analysts, especially business analysts, adopt a sort of Master-of-the-Universe arrogance that they know it all because of their exceptional statistical abilities. Not surprisingly this type of analyst tends to find that their analysis and recommendations are ignored by practitioners. Great analysts, like great coaches and great teachers, are humble, always aware that their knowledge is limited and with a lifelong commitment to learning more.
As a hypothetical, what’s the one variable that technology is still unable to capture to your liking? What statistic(s) do you have the most interest you in further researching?
BG: Technology cannot really fully capture the ability of top players to make optimal decisions instantaneously. Technology can help measure the outcomes of decisions but cannot determine the range of alternative possible decisions. But tactical decision making is an essential attribute of top players. Decision making can only be assessed by the coaches relative to the team’s game plan. What I’m really interested in is trying to quantify the optimal positioning of players at any point of the game and deriving measures of how good players are in maintaining the optimal position both in attacking and defensive phases of games. Spatial analytics is at the cutting edge of data analytics in the invasion-territorial sports.
—
You can read more about Dr. Gerrard’s work at his blog, Winning with Analytics, here. Or, check out this recent piece on the use of wearable tech in sports.