Dr Bill Gerrard looks at the analytical revolution in elite sports, outlining some of the pitfalls and the promise of this burgeoning field.
The difference between winning and losing in elite sports is often very small. Gaining a competitive advantage can come down to what Sir David Brailsford, the coaching guru behind the recent successes in British cycling, calls “marginal gains”. And one source of potential marginal gains these days is data analytics. Statistical analysis of performance and physical data from both competition and training is increasingly being seen as providing invaluable insights to support a whole range of decisions by coaches from player recruitment through to team selection, devising tactical game plans, and setting training priorities and workloads.
Indeed the analytical revolution in elite sports has even generated an Oscar-nominated Hollywood movie, Moneyball, starring Brad Pitt. Moneyball is based on the 2003 bestselling book of the same title by Michael Lewis and tells the story of how a small-budget team, the Oakland Athletics in Major League Baseball, made themselves competitive with the New York Yankees and other big-spending teams by using statistical analysis to identify the most cost-effective players.
Scrape away the drama and passion of pro team sports and ultimately it is all about transforming a financial budget into sporting success. If you do not have a huge fanbase or a very rich owner, then you cannot compete by outspending your rivals. You have to be smart, spending less but spending better. And that is exactly what Oakland did under the leadership of their GM, Billy Beane, played by Brad Pitt in the movie. It is a classic David story, taking on Goliath by doing things differently.
Moneyball has been a real game changer, putting data analytics on the agenda as a source of competitive advantage in elite sports. Initially the emphasis has been on data analytics applied to physical data. The development of tracking systems including GPS and other wearable technology is producing a huge amount of data on distances covered and speeds of movement to go along with masses of other physical data on heart rates, muscle size, weight, blood, urine, muscle soreness, sleep patterns, well-being and so on. In addition teams have access to detailed event data produced by UK-based companies such as Opta and Prozone who record every action performed by every player during a game.
The problem that coaches now face is data overload. There is just too much data available. This puts a real premium on being able to determine what is critically important to produce a winning performance. It is a signal extraction problem, trying to identify amongst the noise of myriad data sources the really key performance indicators on which coaches should focus.
The answer is likely to lie in one of the most consistent findings of data analytics across a number of team sports, namely, it is not the activity levels of teams that counts but rather the effectiveness of what is done; how well rather than how much. Elite athletes are exceptional physically, able to maintain exceptionally high activity levels in their sports. But even more so they are exceptional decision makers, able to make the best decisions instantaneously on what to do and then able to execute those decision with exceptional technical skills.
Ultimately it is the expert evaluation by coaches of the decision making and technical skills of elite athletes that provides the most useful data and a real source of marginal gains in the future.