RE: if probabilities are known, then "luck" can be measured...

Posted on: January 29, 2024 at 13:27:16 CT

Luck is a measure of the deviation between a team’s actual winning percentage and what one would expect from its game-by-game efficiencies. ”Lucky" teams win more games than you'd expect them to based on their statistical profile and "unlucky" teams lose more than you'd expect. This is essentially a measure of the difference between a team's actual record, versus what Kenpom's rankings would predict their record would be, and he calls teams that overperform his predictions to be "lucky" and teams that underperform "unlucky". It basically represents the difference between his model and what actually happens.

To put in laymen's terms, 'luck' in many ways quantifiably measures coaching. A team that underperforms its talent and fails to realize the potential of their assembled roster is unlucky. And a team that exceeds its player's abilities and collective output is lucky. Its not quite this simple as teams like ku that have unfair officiating advantages and protected homecourts often show up high in the luck metric as well irregardless of poor coaching. But as a coach, having a high 'luck' metric should be a badge of honor because it means that your team is outperforming what the statistical model predicts it should be able to do based on your players. Coaching isn't easily quantifiable, but 'luck' may be one of the best indicators of capturing it.