Where there is a wealth of information, there is a poverty of attention– Herbert Simon.
In 2003, award winning non-fiction writer Michael Lewis published ‘Moneyball’. In his book, Lewis tells the tale of Billy Beane, the cash strapped manager of the Oakland A’s who orchestrated the longest winning streak in pro baseball history, pioneering the use of data science and modeling in sport.
Along with others in elite sport, I took a deep interest in the book’s ideas. Digging deeper into literature and history to understand how this new technology can and is impacting outcomes in elite sport, and the opportunities and challenges it creates for leaders, coaches and athletes.
Simply defined, technology is anything that allows us to do more with less. When framed in this manner, technology as a concept in sport has been used for thousands of years.
Greek Olympians of the Ancient era had their training and nutrition planned by physicians, Bill Bowerman of Nike fame created new technology with his waffle shoe sole, and larger racquet heads in tennis changed the game.
Though technology in sport is far from new, Information technology and the skillsets (such as data science) that come with it is still in its infancy in elite sport and for many teams it has yet to understood or used effectively.
The challenge then for leaders of ambitious organisations in sport is how best to respond to the coming tidal wave of technology in a manner which creates a competitive edge without detracting from core objectives.
The highly competitive nature of the sports industry has led to an emerging trend of pro sports teams and professionals relying more and more heavily on data science and data scientists.
In 2002 AC Milan invested millions to design and build the state of the art Milan Lab, a sophisticated science and research facility designed to extract vast quantities of data related to every athlete and predict injury before it occurs.
In 2014 Super Rugby Premiers NSW Waratahs partnered with IBM to combined data science and modeling with a new movement diagnostic tool called Sparta Trac to better manage their athletes.
It seems every year there is there more technology creating more data, so much so that demand for database software as service companies such as Smartabase has exploded and created a whole new industry within the broader sports context.
As a physical performance manager I now access and interpret thousands of data points each tied and tagged to every athlete I am responsible for daily. The crazy thing is many of these advances have been made in the last five years.
Of course, this all makes perfect sense when you consider the rising costs of athlete salaries, match payments and bonuses combined with the inherent risk of injury that comes with increasing exposure to physiological demands at or close to ones limits.
It can only be prudent to try to use whatever means available and legal to find a competitive edge, and attempt to mitigate as much risk as possible to the most valuable assets in any sporting organization, its athletes.
This new technology related data and data science has given rise to a new kind of expertise, and new questions like:
- What are the implications of data science for elite sport?
- How is data science and the technology that enables it best used in sport?
This article attempts to answer these questions in order to stimulate intelligent discussion around the subject within teams and among industry professionals.
The Implications of data science in elite sport:
In his groundbreaking book ‘Good to great’ Jim Collins identified ‘technological sophistication’ as one factor that consistently differentiated the best-performed corporations from their competitors.
Yet Collins is quick to qualify his assertions by explaining that great organisations are rarely early adopters of new technology but carefully considered in choosing what and when with regards to technology.
Collins goes so far as to suggest that the best organisations and the people within them have a healthy disregard for technology, and rarely credit it with any significant influence in overall success.
“Thoughtless reliance on technology is a liability, not an asset. Yet when linked to a simple, clear and coherent concept rooted in deep understanding – technology is an essential factor in accelerating forward momentum”.
But of what relevance is this in elite sport?
In truth, elite sport is almost a decade behind big business and many other industries when it comes to data science. In many ways big organisations built these tools to improve processes and outcomes by using technology to better describe their environments and variables within them.
Probability & Psychology
Data science in elite sport is the new shiny tool that many of us are currently besotted with. The promise of big data and predictive analytics is that we will be able to accurately forecast the incidence of injury or the suitability of an athlete for selection and intervene proactively to ensure the outcome is a desirable one.
However data science and its findings can only ever really point to correlation, not causation. Yet as humans we find it quite hard to distinguish the two, and our evolutionary instinct is always to take the easy way out by ‘trusting the numbers’.
The problem is, the numbers don’t tell us what is going to happen, they only offer us a closer look at the probabilities of events that MAY occur. Where we get into trouble is in the way we treat the numbers, like gospel.
Daniel Khannemans book ‘Thinking fast and slow’ describes two ‘systems’ we use for decision making; system one and system two. System one is fast, automatic, instinctual and gullible. System two is slow, deliberate and calculative however much more effective in decision making.
As humans we have a tendency to lean on system one because it requires less energy and concentration to do so, allowing us to preserve our reserves for dangerous situations that may require action.
We look to preserve energy because we have evolved to be highly motivated to avoid pain or the possibility of painful situations (like making a bad decision). This instinct is called loss aversion, and up until recently it has been our most important survival mechanism.
The combination of these two very human habits; being lazy in critical thinking and instinctively avoiding pain or situations that might cause pain, mean that we frequently make mistakes when it comes to interpreting and acting on data.
The danger that data science presents is that if we the users are not aware of our own limitations and our inherently human flaws, little can or will be done to mitigate them.
This will cause us to rely too heavily on technology and ignore or overlook information from alternative sources while unconsciously moving away from crucial aspects of our work. Consequently we will actually be more prone to error than we were without this technology.
In the Justice system we can see how the use and the users of data science has evolved in this way, at first it was used effectively and ethically, however more recently these tools are increasingly being utilized in troubling ways that significantly impact peoples lives.
As far back as 1990s the police force used data science to make decisions on where it would deploy its resources based on past trends and new information. Seems like smart use of resources right?
Fast forward to the present and we can see where our evolutionary instincts can lead us; Judges are now using data science to determine whether a felon has been properly rehabilitated and should be released back into society.
If this seems like a good idea to you, it’s probably because you have confused correlations with causation, or you too have fallen victim to your evolutionary instincts. The implications for incarnated people with the certain background experiences and poor socioeconomic status could be damming.
In 2002 Tom Cruise starred in the action drama Minority report, in the movie Cruises character was tasked with arresting people who were allegedly about to commit a crime. Just over ten years later and we are seemingly not far from that reality.
In sport, the dangers are less severe but still worth considering. By mistaking correlation with causation and relying too heavily on technology we may never have experienced the greatness of athletes such as Tom Brady who’s poor showing at the draft almost saw him lost to the game.
Should we fall prey to these flaws as managers, coaches or trainers we are less likely to be aware of those small but crucial observations and interactions that provide additional ‘data’ for decision making.
Possibility and mastery are the very tenets that sport rests upon. Sports stars symbolize what is possible when one sets their mind to task and carries it out with discipline and focus.
American football fans watch Tom Brady not just because of what he has achieved, but because of what he has overcome and who he has become in the process. This is what sport is all about.
By its very nature, the highest achievement occurs outside the bell curve, it is the outliers who defy the odds when people least expect it. Imagine what world we would live in if Roger Bannisters coach felt it necessary to educate him on the odds of his success?
Science can help us explain the world, and the best minds have helped advance humanity through the scientific method, but we must remember that the ultimate scientific instrument is the human mind. It was after all Einstein who said: ‘Look deeply into nature and you will understand everything better’.
Polynesians learned to sail hundreds of years before Columbus famous journey and the invention of instruments like the compass. Using deep observation they understood patterns in the direction, temperature and strength of the wind, the colour, shape and content of waves and water its currents and contents.
So advanced were their methods, that when the Spanish first discovered their culture they simply could not comprehend or accept that such primitive people could be technically superior navigators of wind and waves. As a result the superiority of Polynesian people as sailors was ignored for centuries.
So what does all this mean?
The future of data science and technology in elite sport is dependent on its users, like most tools it can be used wisely by well meaning people, or it can be used poorly by people who fail understand themselves and fall into lazy ways of being.
The tool itself is not good or bad, but the way it is used can result in either.
Now that we have explored the implications of data science, and learned how to be in relation to this new technology. Next we will learn how to use it most effectively.
We will do this by modeling and adapting the combined experiences and methods of a billionaire venture capitalist/tech entrepreneur, with one of the worlds most decorated fighter pilots.
How to use technology and data effectively in sport.
Billionaire investor and entrepreneur Peter Thiel is by virtue of his creations and experiences probably THE world expert on the effective use of data & technology. In his recent book ‘Zero to one’ he challenges the fear that computers will cost all people their jobs by describing the fundamental differences between skillsets and abilities of both.
Computers are far more different from people, than any two people are from each other, computers and humans are good at fundamentally different things.
People have intentionality; we form plans and make decisions in complicated situations. Were less good at making sense of enormous amounts of data.
Computers are exactly the opposite: they excel at efficient data processing, but they struggle to make basic judgments that would be simple for any human”.
As an example Thiel points to a project Google invested years and millions of dollars in to build a supercomputer capable of scanning millions of thumbnails in YouTube videos to identify a cat with 75% accuracy.
‘That seems impressive” he writes, “Until you remember that a four year old can do it flawlessly”.
In describing the differences between computers and humans Thiel teaches us how technology and data can be best employed in the sports industry. He also directly shows us, through explaining how Palantir (a cybercrime company he founded in 2004) combines the skills of computers and humans with stunning results.
Thiel’s company has created immense value (15 Billion dollars worth) by producing far superior results when compared to the United States two major security agencies the Central intelligence Agency (CIA), and the National Security Agency (NSA).
The CIA is primarily focused on using human intelligence to solve security problems, they employ a huge network of spies to keep the US safe, whereas the NSA is biased toward computers& technology.
Neither approach is wrong or right but neither approach works as well as Palantirs more hybrid approach.
Palantir merges the gifts of humans and computers to successfully predict where terrorists plant their explosive devices, prosecute high profile insider trading cases and uncover the largest child pornography ring in the world.
Palantir uses software to analyze data the US government tracks, the software crunches the data and flags suspicious activities trained analysts then review. Using this approach, the company played a significant role in finding Osama bin Laden.
Importantly, the computers do not predict or decide anything. The power of the human mind and its contextual understandings is far better suited to this task. In this way, Palantir uses computers to free up people to do what they do best.
Palantir’s example shows us that computers if properly used can create space for humans to do less of what they do badly (crunching data), and more of what they well (planning, deciding and executing well timed actions).
So if computers and humans can partner to produce superior results, how can we know how and when to use each distinct skillset in the context of the rapidly changing, high pressure, high stakes environment that is elite sport?
Enter John Boyd.
Boyd was an Air force fighter pilot who served in the Korean War. Through deep observation and attention, Boyd developed a framework for rapid and effective decision making that ultimately changed the war.
Boyd knew the Americans used bigger, slower fighter planes than their opponents, but he also knew that the US Sabrejets transitioned more quickly between fighting styles than Soviet Mig jets.
This meant that the Americans could perform more maneuvers in less time, allowing them a competitive advantage were they to recognize and exploit this information.
“The adversary that can move through cycles faster gains an inestimable advantage by disrupting his enemies ability to respond effectively” Boyd explained while speaking before the housed armed services committee in 1991.
Among other contributions like helping design superior fighter jets Boyd created a framework to teach and guide effective decision making in dynamic, high pressure combat he called the OODA loop.
The OODA loop is an acronym for the four major principles of the framework he spent several years developing. The four letters stand for four steps: Observe Orient Decide and Act.
- Observe what is happening and process as much information from as many sources as possible
- Orient those observations by distinguishing the relevant from the insignificant
- Decide on a course of action and select one path.
- Act to execute the decision, while bearing mind this action is not the end since the loop flows continuously.
As in most cases the simplest solutions are often superior. If we look deeper into this framework in the context of what we have already explored, these four steps are eminently instructional and practical:
Observe what is happening and process as much information as possible – Observation is a fundamentally human skill and therefore any task or technology that distracts us from this is to be delegated.
Information processing is best done by computers and associated technology. Therefore, it is advisable to invest in technology and associated expertise that facilitates automatic and efficient processing of data for descriptive analysis.
Orient those observations by distinguishing the relevant from the insignificant – Effective, efficient data processing should allow more space and time for professionals who possess domain specific experience and expertise to determine which information is important to consider.
Decide on a course of action and select one path – Once technology has completed a comprehensive descriptive analysis and skilled professionals have discerned what is relevant, the next task is to use this information to inform a decision.
Unlike dog fighting: In sport there are decisions that can and must be made quickly, and there are decisions, which can be made with less haste. In the past I have written about the importance of effective decision making in an athletes return to play. Though it is worth revisiting briefly here.
There are four ways to make decisions, and all four have their various pros and cons. The following tables summarizes each:
|Command: Decisions made with no involvement from others.|
|Very fast way to make decisions.||Typically low quality decision because uninformed by anything happening now but on past experience.|
Command decisions are best suited to scenarios where little or no information is available and time pressure is significant. Making substitutions on game day is one scenario where this is likely to occur at times.
|Consult: Decision maker invites others to influence them before they choose.|
|High quality decisionSafer environment for some people to express their opinions (less group mentality)||Can be slower when involving too many people, and without parameters for their involvement.Can be slower if information hard to access or make sense of.|
Decisions made via consultation are best suited to scenario’s where many people are involved but some people have more informed and valuable opinions than others. The consulting decision maker has flexibility in whom he/she consults, but still retains autonomy of choice.
Deciding on how to manage an athlete during their return to competition and subsequent exposure to load is one scenario where consultation with experts such as rehab coach, treating physiotherapist and doctor is valuable.
|Vote: Majority decides after all facts presented.|
|Fast decision which can be high quality||Colleagues who lack ability, experience and work ethic will negatively affect the quality of the decision.|
Voting to decide is best used as a method where technical expertise is not likely a prerequisite for involvement, time pressure is high and stakes are relatively low. Personally I think this is a lazy mans method and cannot think of any scenarios in using data in elite sport where it would be most appropriate.
|Consensus: Dialogue until everyone agrees on one decision.|
|Highest quality decision||Very slow process and time cost can be prohibitiveConflict likely, frustration probable.
Intelligent introverts may withhold valuable information and opinion in the presence of dominant extroverts.
Deciding via consensus is most suited to situations that are extremely high stakes, relatively low time pressure and have long term implications. For example: which data should we display, to whom, and how should display it?
These four methods are relevant to interactions between interdependent humans, less so humans and computers. However it is assumed that people responsible for collecting, collating and using technology to analyse data will possess relevant information in any situation where data can be used to balance opinion.
Act to execute the decision – Executing is easy and can be automated, but optimal timing requires a judgment and contextual understanding of scenarios that computers lack, therefore this should always be the job of experienced, attentive and engaged humans.
Evolution, Technological Sophistication & Conclusions.
Darwin was misunderstood by those who claim his theory of evolution championed the idea of ‘survival of the fittest’ In actual fact, his findings (and his real opinion) indicated that is more likely the most adaptable or changeable species who survive and thrive in an ever changing world.
Jim Collins, our business researcher observed the same trend in the companies that stood the test of time and consistently outperformed their competitors. Technological sophistication allowed these organisations to avoid fads, see the bigger picture and choose and use technology wisely and well, while adapting and evolving their methods over time.
This article has been my attempt to contribute to a more sophisticated approach to data and science in sport. Or at least to improve awareness and start some intelligent conversations that go deeper than ‘the power of analytics’. I hope to have achieved my aim while also providing some practical and pragmatic information for professionals in elite sport.
If you have anything to add to this conversation please don’t hesitate to contribute by commenting belowFollow