Is consistency the unrecognized analytics factor?

I’ve been thinking a lot about this Senators’ goalie situation and the fact that we just absolutely MUST trade a goaltender and get good value back otherwise everyone’s going to write 200 blog posts about how we didn’t. I was also thinking about how Chicago is probably going to have to trade another 3-4 good players to stay under the cap next year, and Cory Crawford, he of the lucrative new deal could be on the block.

If I were a GM in either situation, what would I do? Could Chicago really go into next year with Scott Darling as its starting goaltender? Would Ottawa go into next year with an injury-prone older starter and Wildcard Hammond? How about Lehner, who has all of the potential in the world but is in the middle of his own personal time travel journey?

This led me to think about how much GMs pay for potential and projections, and how little they pay for consistency. Here’s a theory:

In an analytics-driven, salary-cap era, a player’s value is expressed in three ways: 1) The degree to which he drives possession. 2) His statline relative to his peers. And 3) How much money he makes.

Each of these three factors are dependent on one another. It’s why expensive Alex Ovechkin can be one of the best players in the world and still have people question his value, and why scoreless Erik Condra can be cheap and drive possession but be allowed to walk for nothing.

(I guess there’s a fourth factor in here encompassing all of those other intangible qualities, like sticktoittiveness and likeability and being a good dude.)

One factor that I think might be missing in all of this is if a player performs consistently. I know, you must be thinking “this is obvious. If a player performs well consistently, then a GM knows it and factors it in.” But I’m talking about if a player performs only averagely, but consistently. I’d like to argue that that consistency provides a level of value over and above the player’s statline and possession metrics.

Because a player is consistent, it means that a GM can plan around him. He can better understand the gaps on a team in a coming year and spend his meager resources, in terms of picks and cap space, to plug those gaps, confident in the knowledge that they won’t produce yet more gaps.

This is the place in the blog post where I would love to include a spreadsheet of the most consistent players, but frankly I don’t have the data. War on Ice will give you cumulative totals, but I don’t have the patience to download stats season-by-season and run the spreadsheets to describe variance. What I’d love to see, though, is which players above a certain threshold for possession (so we weed out the players who are consistently bad) display the lowest variance from season to season. Basically what you end up with are players about whom you know what to expect. And that, in itself, has value.

I know I’m mixing up skaters and goalies here, and I don’t know if Chicago will trade Cory Crawford, especially if he helps them win another Cup. But I do know that if Chicago can find a cheaper option with less variance in his statline, they can reliably pour the savings elsewhere to compensate. The same holds true for Ottawa. Lehner might have a higher ceiling, but also a higher variance than, say, Craig Anderson. And knowing that variance has value.

I hope you enjoyed this summertime blog post. Stay safe everyone, and wear sunscreen.

3 thoughts on “Is consistency the unrecognized analytics factor?

  1. I wonder if eventually we could analyze expected corsi situation by situation. Player X on the ice with his 4 bros against 5 bros on the other team. What would that player’s dCorsi be for that shift? If there’s minimal variation shift-to-shift in their dCorsi, could we say that that player is consistent?

    • That’s interesting. What would also be interesting is measuring variance irrespective of situation. If someone produces consistently regardless of line mates and opponents, I’d have to think that was even more valuable than consistency in certain situations.

      • I think if someone is producing the same regardless of their linemates, you would see that in the WOWYs. @ineffectivemath had a quick twitter convo with Methot because strangely there’s not a huge variability in Methot’s on-ice performance with different teammates. Not huge looked pretty huge to me, but it’s all relative.

        If someone is the same with Neil as they are with Turris, they’re either really good or really bad. There’s some consistency that you don’t want

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s