Data Science in the Sports Industry
The sports industry generated $90 billion in revenue for 2017 and is a world completely saturated with statistics: What is the likelihood of team A defeating team B with its current roster of players? Since player X has been injured, how will that impact the probability of team A making it to the finals? As a sports statistician, you will be responsible for building predictive models based on an array of data, e.g., individual player performance, team performance, etc. Much like being a data scientist, you’ll need to have some programming knowledge along with advanced statistical analysis skills.
One of the most well-known sports statisticians, Nate Silver, has made a lucrative career from the sports analytics industry (it doesn’t hurt that he wrote a New York Times Bestseller, “The Signal and the Noise”). But, does that necessarily translate into success for other statisticians and data scientists who are interested in entering the sports analytics industry?
Career Outlook and Salary for Sports Analytics
Merging what we do for daily work with our “passion” is usually the ideal. However, being able to keep a roof over our heads and food on the table — more often than not — takes precedence. If you love sports and statistics, then you may be one of the lucky few who has the opportunity to combine both.
The U.S. Bureau of Labor and Statistics (BLS) does not have a separate designation for “sports analytics,” and sports data analysts are more commonly known as sports statisticians. Thus, sports statisticians are classified as simply “statisticians” by the BLS. Statisticians can expect to earn a median salary of $84,760. Per Payscale.com, the average sports statistician salary is just under $72,000 annually. As a whole, the career path for statisticians is extremely positive: 33% growth between 2016 and 2026, which is much faster than average when compared to all other occupations.
On the other hand, mathematicians can expect to earn a median salary of $103,000, but this is likely to be adjusted downward considering the average sports statisticians salary mentioned above. If you decide to earn a degree in math, and your goal is to enter sports analytics (or any other analytics career path), then you should take stats coursework in addition to your math requirements. While the basis for stats is math, most mathematicians and statisticians will tell you that statistics is a different math-based specialty (or that it isn’t even a math sub-discipline at all). What are the main differences?
- Stats relies on the context of data, e.g., the what, where, how, and why of collection and analysis;
- Mathematical reasoning is far more deductive than statistical reasoning (which is primarily inductive);
- While mathematical statistics courses are available, math is merely a tool for measuring uncertainty (probabilities) rather than going through the definition-theorem-proof protocol.
So, they’re intertwined but have distinct, and important, points of departure. Both are rigorous programs regardless of which one you choose, and you’ll be positioned for an analytics career in just about any industry. The trickier decision comes to pass if you’re specifically seeking a Bachelor’s or Master’s Degree in Sports Analytics.
Degrees in Sports Analytics
Sports analytics isn’t a well-represented undergraduate degree. Most sports statisticians either major in math or statistics, and then take supplementary courses (or choose to minor) in sports management. You’ll find that a majority of sports-related bachelor’s degrees focus on sports management, which can include analytics, but mainly focuses on students learning sports marketing, brand management, and the business side of sports in general.
Bachelor’s Degrees in Sports Analytics
Syracuse University is one of the few academic institutions offering a Bachelor’s Degree in Sports Analytics. Over the course of the four-year program, you can expect complete coursework in game theory, advanced math (through Calculus II), sports data analysis, programming for sports analytics, and sports management. Notably, Syracuse University’s degree program is on-campus as there are no online bachelor’s in sports analytics options at this time. Indeed, many sports statisticians major in statistics or math and then apply their learning to the sports realm. From there, they leverage their knowledge to find a job within the industry by either taking the usual route (searching job listings at major job sites and applying) or through starting a blog (such as Nate Silver’s FiveThirtyEight).
Sports Analytics Master’s Degrees and Ph.D.s
A few more options exist for master’s degrees in sports analytics. Syracuse University, Indiana University, and Northwestern University all offer graduate degree programs (certificate or master’s level only) in sports analytics. Applicants should already possess either a STEM undergraduate degree and/or have experience in sports analytics. For Ph.D. level studies, the primary option is to complete a Ph.D. in statistics or math.
Data Science or Sports Analytics?
Data science and analytics are frequently seen as interchangeable job functions. While data science is an analytical discipline, and analysts do perform some of the same work as data scientists, there are subtle yet important distinctions. As a data scientist, it’s expected that you’ll be part data engineer, part data analyst, and part data engineer. The field is in a continuous forward motion as employers are still sifting through the who, what, how, when, where, and why of their data teams. Sports statisticians (or analysts) will perform data science work: building mathematical models for predictive analysis which will include machine learning and AI. The main differentiation is the type of data they are working with. Consequently, sports statisticians are data scientists for the sports industry.
How Data Science is Transforming Sports
From the Summer Olympics and Soccer to the Super Bowl, billions of people tune in to watch their favorite teams or sports stars compete for the top spots in their particular sport. Between merchandising and highly profitable endorsement deals, garnering a wide viewership is critical for continuing cash inflows; after all, the sports industry is ultimately a business.
But, sports are replete with confounding and uncontrollable variables such as the weather, the unique physiology of the individual players, decisions by referees and judges (which are sometimes politically based), and the choices made by the players during the game or match — to name just a few. Data scientists and sports statisticians definitely have their analytical work “cut out for them.” Fortunately, due to continuous advances in technology for Big Data analytics, acquiring sports data is relatively easy; it’s deciphering what to do with that data which tends to pose the biggest hurdles. This is where the sports statistician/data scientist comes in, and they’re revealing predictive insights for optimal decision making throughout the sports industry.
Game Outcome Prediction
Game theory and combinatorial game theory reign supreme for attempting to predict the outcome of the various decisions affecting the likelihood of a particular outcome in sports. Deep learning algorithms are currently being used, in concert with game theory, to “quantify quarterback decision-making and performance.” Markov models are also being deployed to help model “movement in the NBA” for more robust predictions at the player and team levels.
Sports simulation is often used in sports wagering and fantasy sports leagues, and it’s made its way into sports organizations where sports statisticians/data scientists construct mathematical models based on a combination of various game theory types and perpetual data ingestion from player and team statistics. For example, in 2018, one set of researchers performed a hypothetical analysis of the addition of the team competition for figure skating which was implemented in 2014. Their goal was to identify which teams might have won prior to 2014 had the team competition existed for each of the prior Winter Olympic competitions. Another research group developed a team recommendation system for Cricket, where individual player strengths and weaknesses were analyzed and compared with the same factors as another team to arrive at a predictive outcome.
Since Big Data is a recent phenomenon, and there is ongoing development for technological tools that can measure all possible facets contributing to athletic success or failure, there is much room for growth in being able to precisely predict the winner of a sporting event.
Professional sports organizations invest a sizable chunk of cash into finding and onboarding new athletes. For example, each year the NFL conducts a combine where professional football hopefuls demonstrate a myriad of skills which are then analyzed by team scouts. Each defensive and offensive position is tested separately through running, jumping, strength, and agility drills while team coaches and other decision-makers observe the potential players in action.
Certainly, coaches can draw conclusions based on the combine and the players’ performance stats (either from college sports or semi-pro league play). But, the common thread in Big Data for sports is a lack of certain types of data. While this as of yet unseen or uncollected data is available (insurance industries have been collecting this type of data for decades), the world of sports has been slower to move beyond a relatively superficial player and team analysis.
Data scientists and sports statisticians — if they are well-trained experts — can tease out possible correlations and likelihoods based on copious amounts of data, e.g., the athlete’s current and future health (physical, mental, and emotional), individual vs. team dynamics, game simulation in various conditions, etc. Furthermore, understanding what data is needed and how to collect it is a fundamental data science skill. As such, as data scientists continue to enter the sports analytics industry, they will be able to create new predictive models and vastly improve the data collection capabilities for more robust predictions under highly variable conditions. This will result in a significant increase in choosing the right players for the right team at the right time.
Keeping Players Healthy
If you’ve ever played a sport, whether recreationally or professionally, then you have a direct point of reference for just how strenuous sports activities can be. Professional athletes are pushed to their physical, mental, and emotional limits. The organization and fans fully expect that athletes will compete at an exceedingly high level for every game. Contact sports such as Rugby and American football have additional elements that significantly increase the risk of injury. Team sports place further pressure on athletes as any injuries or illness of one teammate impacts the dynamic system that is competitive sports. This means that athletes must maintain a state of health and healing which is far above the everyday person.
Physiologically, there are plenty of technologies that professional sports organizations and individual athletes use to analyze the correlations between sleep, nutrition, and training which serve as a guide towards sustaining exemplary physical conditioning. However, it’s more difficult to measure an athlete’s mental state during a game unless it’s followed by an observable behavior.
Researchers are exploring the effects of mental pressure on performance, and the long term and short term connections between mental health, emotional health, and how those aspects provide either supporting or negating effects on individual and team performance. With the continued forward motion of artificially intelligent systems, more data can be collected and correlations drawn between the various mental and emotional states that can either directly or indirectly affect athletic performance.
In some cases, the lag in sports analytics is caused by a lack of technological tools that would boost data collection for a wide array of “silent factors” impacting sporting event outcomes. The other repeating issue is parsing through the data as not all data is “created equally” when it comes to making predictions under a collection of uncertain conditions (e.g., individual and team decisions that will shift within an action-reaction context). Additionally, due to a dearth of highly qualified data scientists throughout all industries, those who have attained such expertise are lured by other industries that provide a higher salary and other incentives. The sports industry has yet to feel the impact, and recognize the distinct value, of data science. But, as more sports analytics research is conducted, and deeper insights are revealed, there may be an increased demand (along with an uptick in salary) for sports statisticians and data scientists in the coming years.