The 2015 NBA Finals were a tight one, stretching out to six games with notably close scores. For the first time in finals history, the first two matches both went into overtime between the Cleveland Cavaliers and the Golden State Warriors.
Amazingly, both teams had head coaches that were just starting their first year in the position with that team.
How did it happen that two freshmen coaches made it all the way to the Finals? In large part, it happened on the back of a new wave of data-based analytics that has been driving basketball strategy across the league. The Warriors were the league’s best three-point shooting team during the regular season, and the Cavs cracked the top 5. It was no accident—statistical analysis told both coaches that three-point shots had a better payoff than traditional pushes into the paint. Data science was helping them win games… and helped push the Warriors on to victory.
There’s nothing new about statistics in sports. The element of surprise that comes from using statistical analysis to select players and hone strategies went away after Moneyball, the book and film describing the surprising success of manager Billy Beane of the Oakland A’s in using sabremetrics to draft undervalued players and build a successful team for less money, came out.
But just because the secret is out doesn’t mean the technique is dead. Instead, the sports industry has become more dependent on statistics than ever. No professional sports franchise can afford to ignore them, and no athlete today trains without an investment in data collection and analysis performed by master’s-educated data scientists.
Training For Success By Playing The Averages
Records are made to be broken, as they say, and you can see it happening in the record books all the time. Every year, some previous height of athletic accomplishment falls to a stronger, faster, more capable record-holder.
Those continuous incremental improvements don’t happen because humans are magically larger and stronger than they were twenty years ago. Instead, the edge comes from training and technology.
As more and more studies have been done of human physiology, athletes and trainers have had more to work with to optimize workouts for performance. But people are unique. What is maximally effective for one athlete in a controlled study may not be exactly what works for another in their regular training regimen.
These trends are too subtle and complex to spot without in-depth analysis. Data scientists today rely on sophisticated wearable personal monitors and other information-gathering training systems including:
- Heart rate monitors to gauge exertion
- GPS motion trackers
- Accelerometers and pedometers to track strain and effort
- Biometric markers to analyze body position
The data brought in from all that monitoring is compiled by programs to provide an accurate day-to-day picture of athletic capability.
Months or even years of analysis can spot trends and point to key elements that coaches and trainers can use to bring players to their absolute peak capability at the exact moment they need it. Working with trainers, coaches, nutritionists, and other sports experts, data scientists help orchestrate workouts and training regimens for athletes in almost every sport.
Sports Analytics Boosts Athlete Safety As Well As Performance
It’s not all about peak performance, though. Data scientists are also leveraging better understandings of physiology and masses of performance data to help decrease injury rates in pro athletes. The Toronto Raptors, for instance, have used wearable devices to track intensity and acceleration rates exhibited by their players to unearth trends that can show overexertion or poor form that is likely to lead to unnecessary injuries.
Some professionals feel that this is the real killer app in sports analytics. Although improving performance generally is a valuable goal, injuries among highly paid athletes costs teams—and the athletes themselves—millions of dollars each year. Reducing even a small percentage of those both improves efficiency and keeps athletes out of the hospital.
Data scientists working in sports analytics aren’t just there to support the professionals, however. A burgeoning market in personal fitness devices like the Fitbit, or multi-purpose wearables like the Apple Watch, have brought sports data analysis to the masses. Data scientists build the backends to store and collate the oceans of information coming in from those devices and create the algorithms and interfaces that help interpret it for day-to-day users.
Managing The Game With Real-Time Data Analysis
No NFL sideline today is complete without stacks and stacks of visibly branded Microsoft Surface tablets. The off-the-shelf hardware may be all part of a marketing deal, but the information coming across those devices is anything but standard. The Sideline Viewing System gives coaches and players networked, collaborative images of previous plays for analysis and discussion, replacing the slow, cumbersome paper picture print-outs used before.
That’s just a technical update, essentially, nothing to do with data analysis, but today the Surface’s get more of a work-out than just serving as mobile picture frames. The league is now allowing trainers and doctors on the sideline to use the tablets for medical purposes. A concussion assessment app from X2 Biosystems can allow medical personnel to compare recorded pre-game answers from any player to their answers after taking a big hit. Memory and balance evaluation are key components of the concussion protocol, and rapid analysis can help keep any player that might be at risk of traumatic brain injury out of the game when it’s most dangerous. Player safety is always a big component of sports analysis, but strategy also gets a lot of attention.
STATS Technology put together a system that uses real time video footage of the field to generate statistical models of game play and suggest different strategies or tactics that can be used against trends revealed in the opposing team’s play. The system pulls in up to ten data points per player per second, which can come to more than a million individual data points in a single match.
It’s revealing that the complexities of sport – whether soccer or American football, make sabremetrics look like a grade-school math class.
Devising formulas and techniques to evaluate all that big data is a big job, and data scientists need big-time education to figure out how to handle it. Writing up algorithms and processing the information to present to coaches and players in a time frame that makes it usable is cutting-edge analytics. You’ll need to understand:
- Probability theory
- Regression testing
- Research modeling
- Programming in Python, R, or another data-friendly language
Advanced education is really the only sure way to be qualified for the complexity of what sports analytics entails.
Where the Jobs are in Sports Analytics
Most positions for data scientists in the sporting industry are with the manufacturers of monitoring devices and the software devices that interpret their input. Although there are positions working directly with professional trainers and sports teams, in most cases the work is being outsourced to companies like STATS or Catapult, and that is where most jobs are found.
There are far more people interested in working in sports analytics than there are positions today, so boosting your resume with an advanced education is a must. It also helps, of course, to be a sports fan. Entering the industry already speaking some of the language and understanding the imperatives is a much smoother path than having to understand games and training from the ground up.
It’s an exciting and unusual career path within data science, but if you master the tricks of the trade are lucky enough to land one of these positions, you’ll be a god among men in your fantasy football league.