Coaching in the World of Analytics
As the ideas and methods from the scientific community gain acceptance in sports, it’s natural to wonder how their adoption might influence the role of the coach or manager. After all, the point of implementing an analytics program is to help make objective decisions in order to improve performance of a group of athletes. This is essentially the coach’s job description! So what precisely is the role of a coach in a modern day sports analytics program?
We first point out that without a coach, there is no analytics program. Saying that analytics can replace a coach would be no different than saying the scientific method can replace the scientist. One cannot exist without the other.
This may seem contrary to some current claims made in the field of “Artificial Intelligence” (A.I.) where it is often implied that one simply dumps data into a computer and decisions about how to manage complex systems are spit out. This is, of course, not the case as those who perform research in the field of A.I. will attest!
The main role of the coach in the context of analytics is to develop a clear, specific set of objectives, e.g., improve shots on target, maintain possession longer in the attacking third of the field, give up fewer goals from direct play over the back line, etc. Without such objectives, scientific tools are of little to no value. These objectives dictate the type of data that should be collected and the analytics tools that should be used.
The second major role for a coach, which follows directly from the first, is to assign value to the data. After all, if one is to measure improvement there must be some type of assessment of what constitutes “good” or “bad” performance. Consider a player’s decision making ability on the field, an important skill in most sports. Over the last several years we have asked both youth and professional coaches what they value in a decision, and the answers are frequently different. One coach’s “good” decision can be another’s “bad”. Neither is wrong, of course, they are simply valuing decisions differently depending on their respective tactics, objectives, and experience. As we pointed out in our prior post, so long as each coach applies their chosen measurement “scale” consistently, even abstract quantities like decision making can be objectively quantified, tracked, and predicted.
Lastly, a coach must deal with the sometimes large uncertainty that is present in the data. If the numbers suggests one player is a slightly more proficient goal scorer than another, but the degree of confidence in that numerical assessment is low, the coach will likely bring to bear other pieces of information. For example, in deciding “who to start?” the coach might use knowledge about the players’ past performance against the team they are facing. Such prior information, gained from coaching experience, is extremely valuable and should be augmented by analytics, not replaced.
In short, coaching with the help of an analytics program requires the same basic functions as coaching without one. Specifying objectives & tactics, assigning data value, and managing uncertainty are all still very much the job of the coach or manager. Coaching experience and prior knowledge are still heavily leveraged.
What changes is that the resulting decisions are all now informed by data! If implemented properly, analytics-based decisions will be much less affected by human bias and subjectivity. The coach can also now directly assess how effective his/her tactics really are. The performance of certain players or groups of players can be unambiguously quantified. Uncertainty in coaching decisions can be captured and subsequently mitigated. Each of these items represents a big step toward improving the performance of athletes and teams, and these improvements are the reason why the topic of sports analytics is receiving so much attention.
The real challenge therefore lies in implementing an analytics program with the flexibility to accommodate a coach’s unique set of objectives and tactics while providing the benefits we have just mentioned. This has been the explicit goal of the DSA program since our inception, and something we continue to work toward daily.
