Ensuring Predictive Analytics Actually Predict

MIT Sloan Sports Analytics Conference

Football analytics guru Brian Burke told the Sloan gathering today that there are no significant correlations between success in the NFL and players’ performance at the analytics-laden Scouting Combine. Really? It was a rather shocking disclosure given how fervently NFL personnel evaluators monitor and measure every aspect of performance in the 40-yard dash, vertical jump, and other competitions.

Burke was joined by Texans' GM Nick Caserio, Rams' COO Kevin Demoff, and Ringer Network’s Kevin Clark on a panel that raised the question whether the NFL is measuring the right things. After all, we heard that the 10-yard burst is far more important for defensive linemen than the 40-yard dash and that vertical jumping ability is of little value to offensive linemen. Even some standards about height and weight, which would never have found the 5’10” Cardinals’ quarterback Kyler Murray the first choice in the NFL draft even 10 years ago, seem to be receding in importance, as well. They are surely evolving toward more nuanced, position-specific metrics and away from one-size-fits-all measures.

It’s a reminder to us all that, no matter our line of business, metrics serve us well when they deliver situationally relevant, finely gauged predictive value in the real world. Otherwise, they can be an illusion that actually undermines decision making. Ask yourself, are you measuring the right things and are you using the right tools to do so?

Image courtesy of BizTech Magazine.