Analytics are used for predictions in every facet of the industrialized world. Your claim isn't based in reality.
The highest scoring teams in the league attack the blueline with speed and generate chances in transition. 75% of goals scored off zone entries in the last two years have come within ten seconds of gaining the attacking blueline. A couple of ways to create offense quickly in the attacking end:
- Gain the blueline with possession – like the Lightning and Penguins, who led the league in controlled zone entry success rate and also ranked 1st and 3rd in goals. Ottawa ranked 20th in entry success rate, 25th in goals.
- Get the puck behind the goal-line, win a battle and create a quick scoring chance. As mentioned, the Senators ranked bottom-3 winning puck battles in the attacking end last season.
But it's not like we're trying to predict the score of a single Buffalo - Ottawa game in December. At the macro level there are general trends that are easy to spot and that you can use to predict the future, especially when it's for aggregate events such as the final ranking of the team after 82 games. Hard to know if it will snow in Ottawa on January 8th, but I bet you it will be pretty cold.Your understanding of modeling/analytics is what is detached from reality. Seriously, you don't change 15 variables and expect the data to tell you anything about the future. Industry tries to remove all variables to maintain quality, hockey has too many for any kind of reliable predictions. Their use in hockey is to analyze the past.
No need for Bayesian statistics though to come up with this equation:
Bad team in 2017 - Karlsson - Hoffman + scrubs + n00bs - $$$ = Bad team in 2018