“The people in charge of teams only think about buying players,
but the goal should be to buy wins.
And to do that you need to buy points… they ask all the wrong questions,
and if I tell someone this, I become a leper”.
So said Peter Brand in the film “Moneyball: The Art of Winning”, speaking for Paul DePodesta, Harvard graduate in economics and a baseball and football executive. In short, it highlights that data can lead to obtaining wins and it is readily available, but unfortunately the focus for most remains anchored in a non-systemic vision, based on individual players, with the belief that buying champions means winning straight away. What matters more however is the combination of characteristics held by individuals in a complex, dynamic system. This was a problem in 2002 that today, almost two decades later, we are able to solve.
Data has always been viewed with suspicion by sportsmen, seen as a threatening unknown which can only take away the job of decision makers. In Italy, we are still a long way from the winning models across the Alps and across the Ocean, and the recent (temporary) farewell to the world cup by the Italian national team is just one of many examples that show conservative management of the hive of sports talent. Comments poured out on social media for weeks, attributing the elimination to the most disparate causes (racial, economic and others that are even more incredible), but I believe rather that the cause points back to a limited understanding, of the potential of a good system of athletic evaluation and tracking, which is no longer adequate for competition in the modern landscape.
Why is Italy behind?
A large part of the problem stems from the determined, even admirable, defence of the human factor, of the talent or artistry, impossible to track, that a player can put on the field and which always shows that something extra that can never be completely gathered from numbers, calculations or artificial intelligence. Even if this is true, it is nevertheless clear that these intangibles now form only a small percentage compared to the competitive information which is actually measurable. It is the latter that is rejected, out of fear that numbers will ruin the sport, along with protectionism and widespread near-sightedness, given that sports analyses have the same impact on the player and the game as the radio announcer: they merely describe a situation!
In the era of digital transformation, by now the importance of data for business purposes is well established and for some time companies have been making significant changes in that regard, aware that a medium-long term vision needs information of value to build upon. In the Italian sports system, this type of innovation has met with resistance, just like any real change which it does not feel it can fully control. Sometime back the sports sector became a true industry; however it cannot continue with strategic choices based (only) on whims and feelings: managing and analysing data (with the right processes and methods) is the only sure path for reducing risks and making knowledgeable decisions.
A clear example of this is the 7-1 score with which Germany humiliated Brazil in the 2014 FIFA World Cup: Germany came to the world competition after five years of maniacally tracking the best German footballers and basing its strategic/tactical choice on that data and the relative analysis. Another legendary case is the true-story of Billy Beane and the Oakland A’s (2002), recounted in the book Money Ball, when the general manager brought American baseball’s “punching-bag team” from the bottom all the way to the championship, using data to find the right players – not the most famous. Beane based his acquisition campaign and game strategies on statistics, finding the best way to win match after match.
By examining concrete cases one realizes straight away the potential of this new world. There is an algorithm for evaluating sports talent, developed in Italy for basketball in 2011 and then brought to MIT in 2016 which, on the wave of Billy Beane’s revolution, seeks to find the “right” player not “the best.” That is to say, every coach can easily name the best or worse player on his team and the opponent’s, but he would have a hard time drafting an objective ranking of the rest of the players. This is why he evaluates them subjectively and not objectively. A talent scout does the same, even if it happens unconsciously. An objective analysis, on the other hand, allows one to better “catalogue” an athlete, learning his pros and cons, to understand and forecast performance and to describe him in such a way as to know what type of player he embodies. Thus it will be easier to find out with whom he could be substituted or teamed up when constructing a new line-up, in market searches, in developing new strategies, and so on.
New descriptive roles
In regards to players’ roles, to cite another example, through machine learning based on an SOM (Self-Organizing Map) one can designate new roles, that are descriptive and not mere names, for basketball players [min 18:20 di questo video]. Thus we put behind us generic terms like “power forward” or “playmaker,” and move to “rim defender” (a player with excellent defensive skills, who is often stationed in the painted area, and does not throw often) or a “role player” (a player with good offensive percentages, one who is usually very tall and leads team performance to improve when he’s on the court, positioned in the three-point area. It’s not just a classification, but an objective definition of the athlete based on actions accomplished. As on can easily understand, for a coach, being able to find a player who has those precise qualities (as we said, the right player) is immensely more helpful than generic research on a “centre” in a type of game against the “centre” with whom we wish to replace him.
Let’s not get frightened
The subject is complex, it’s frightening, but it adapts to every discipline and it makes available, to the person who discovers the strength of data analysis, a good number of services which already exist on the market. The ultimate real-word best practices comes from Canada (an analysis support platform for winter sports) and the Netherlands (€6 million next year for prevention of injuries through Big Data) but the list is a long one. A few weeks ago the news reported how the company VolleyMetrics and Giuseppe Vinci (Italian, ex-analyst for the Italian national volleyball team) had acquired the leading business in personalised match-analysis: Hudl.
The most recent and famous Italian example of virtuosity connected to sports data is definitely Prof. G.P. Cervellera who overcame the organisers of the Manchester City Hackathon #hackmcfc. In our country, we have many young businesses and much talent in the world of sports innovation; anyone looking for it can turn to incubators and expert analysts like WyLab, Vittoria Gozzi’s Incubator, or Francesco Mantegazzini of Infront. However nothing systematic or incisive has truly been launched. That notwithstanding, ever more knowledgeable discussions are being had on the subject. We are beginning to ask the right questions.
Do or not do. There is no try.
You need to believe this. Tackling a market, a situation, truly supported by numbers is an arduous mission but nothing impossible. The real question we should all be asking ourselves before starting down this course of change is: do you believe in this story or not?
I have no doubts.