Welcome to Sunday Notes, where we try to rehash important developments occurring on Hockey Graphs and elsewhere in the CORSI twitter league in less than 500 words. I’m sorry if we forgot about your post, or misconstrued what you said. We don’t care. Don’t @ us. Just do better next time. – Asmaen
It’s about the numbers
…but please express them in terms of TOI, especially if you’re going to compare players. Why? Here’s what Conor found:
- TOI and games played (GP) do not increase linearly with each other, and so expressing anything in just point totals and games played ignores the fact that players are given vastly different opportunities.
- Production is a function of TOI: the higher the TOI, the higher the points accumulated. This relationship is more pronounced for forwards.
So not only do rate metrics reflect the different opportunities players are given, they are in fact a superior way to measure contribution. To say otherwise is in glaring opposition of the evidence.
On predicting contribution
Necessity begets the use of analytics, or so goes an ancient hockey twitter saying.
That’s the challenge Jack Han faced recently, where he needed analytics to predict the impact that losing some key players would have on the McGill Martlets. So how did he predict production? Jack Han found that by using a combined video tracking and quantitative approach, past offensive contribution can be quantified to approximate the impact that the players’ departures would have on the team the following year. This information can be used to guide decision-making at several levels, especially in helping the coaching staff make practical adjustments to compensate for the leaving players.
Domenic Galamini from Own The Puck also came up with a method to predict production, but this time using a strictly numerical approach. His general approach wasn’t novel, as it was based on Tom Tango’s baseball Marcel Forecasting System. Overall, his method relied on the same 3 steps Tango used, which are: 1) weighting past performance, 2) regressing to the mean and 3) adjusting for age. The novelty is found in the way he chose to weigh past seasons in step 1), which is outlined at length here.