After reading the Grantland article by Bill Barnwell it seems that the NHL needs some drastic infusion of good applied sport science. While Moneyball has limits (I will blog later about current problems), the article by Bill was too crude in scoring to show a real practical understanding of the relationship of winning and team size. First the size of player weight was just too raw to create proper context of what is going on. Clearly speed and size were correlated that smaller equals faster in the article, but I am not believing that this is a law, just a temporary dogmatic concept. Even greyhound 100m dash athletes are slightly larger than the past, and Usain was at an optimal weight in 2009. Optimal size is what that athlete can produce the most effectively performance wise. Size does matter to a point, but usually a combination of speed and size will be what is coveted with teams.
Skills are not what I will get into, but all things being equal (I know sorry for the repeated mantra of many), repeated power wins. So if I was doing the article I would contact all the strength coaches and ask about jump performance and speed performance and see what players are doing well. Some teams may test better with more distribution of minutes and having smaller athletes (favorable for recovery), but one team will show that they are able to have output high with regards to conditioning and be able to skate faster via explosive power. Add in more functional hypertrophy for a weight advantage and or playing even size wise and you have a clear advantage.
Measuring Ideals requires TMG analysis to estimate fiber profiles, HRV for dealing with fatigue with larger bodies, body composition to see if diet and metabolism is optimized, ESD testing, and getting power tests and speed tests. After all the available information is collected, then GMs, team coaches, and medical/performance staff can start seeing how they can get an advantage training wise. Without the data, nobody knows what to do besides just keeping them healthy and working out without direction.