Predictive analytics and even automated machine learning have existed in other industries for over a decade. Stevie Cohen uses market forecasting in his financial services, Mark Cuban leverages decision modeling for his e-commerce empire and his new pharma project, and Harris-Blitzer & Michael Rubin surely make use of descriptive analytics for their entertainment empires. All of these groups, when they bought these teams, intended to bring best practices from industries more mature than sports.
Sports teams will surely benefit from leveraging this among all stakeholders, not just R&D departments. How can we make predictive analytics accessible enough that they are more universally understood and leveraged?
Let’s level set on a couple of things:
1: Predictive Analytics are no longer the challenge, getting the right data ingested fast enough to action it is.
2: An algorithm is a formula that weights variables to help calibrate calculations which solve problems through prediction, description, and decisions.
3: Forecasting of outcomes (such as injuries or draft-prospect success) is by definition not binary; algorithms assign probability.
Leveraging the above while exploring causality are frameworks for data literacy. Iterate within this by improving datasets with more detail and contextual data. This can be built on open-source datasets; we’ve published using them! Remember, a simple model with 80% accuracy may be more useful (in practical terms) than a complex one with 95% accuracy. Once predictive analytics become more accessible with modern capabilities like AutoML, better decisions get made.
“Talent wins games, but teamwork and intelligence win championships.” – Michael Jordan
The same is true in business.
