Quote:
Originally Posted by Itse
If those models for goalies were reliable, everyone would use them and no one would sign a goalie that doesn't work out or find a hidden gem or keep the wrong goalie.
The idea that you can accurately predict the future by mathematically analyzing the past is ages old and supposed breakthroughs have so often been just around the corner.
Baseball is a much better sport for stats based predictions and they've been on this a lot longer. Same with football. Soccer has been studied at the university level for a long time in Europe. Yet somehow the number of bad signings or hidden gems popping up hasn't really changed that much.
People are not frictionless objects moving in a vacuum. Hockey is complicated. People are complicated. The number of potentially significant variables is infinite. No matter how good you become at measuring some things, you will never run out of things you can't measure or just didn't think to consider in that instance, and when things go wrong, it's often impossible to know for sure what you missed.
People are also constantly changing. How do you know when Tanev is going to lose that one step too many?
It's always going to be an educated guess.
|
I think more and more of these models will be adopted over time, and they will become more reliable as more numbers are tracked and more formulas are used.
10 years ago, few teams had an analytics expert on staff. Now teams have analytic departments.
I think it is an evolution. It will never be a perfect science, for many of the reasons that you cite, but it will give teams substantial advantages in helping them build their teams and acquire the right personnel.
In fact, I also think over time that it will help teams make better choices in terms of coaching as well. As Flames' fans, we saw how blatant Brent Sutter's system was a poor fit for the team at the time.
I think it is already starting to show where teams struggle with on the ice, and will help not only bring players in that can help with those areas of weaknesses, but also an assistant coach or a head coach.
There will always be a lot of noise because of how hockey's events happen and how they are related to players. For instance, Bennett had a strong chance the other night against Winnipeg, and the goalie made the save. The puck gets instantly turned up the ice for a quick goal against - Bennett was not at fault, but I am sure he got a -1 and all the analytics point to him being part of the problem on the ice at the time. As analytics continue to evolve, it will remove some of that noise and do a better job at pin-pointing statistically what is happening on the ice exactly.
I am not a big fan of a lot of the stats that we as fans are privy too and seem to use a lot, just because it does contain too much noise, but I do think it is getting there.
I think the goalie formula where Markstrom has been used as the example of seems like an easier model to make predictions on, and though there are probably a tonne of reliability questions around it, I am sure it is likely to be something that more teams take a harder look at and improve upon.
