To Frame the Unframeable Team (To Scrutinize the Inscrutable Squad)

I don’t even really know where to begin with this article. It originally started as me trying to find some rhyme or reason to our Weird Shots Problem (we only win when we lose), morphed into looking at the defensive metrics of the forwards to see if that would correlate to any metric ever, and then, finally, on the verge of tears, I had to look at myself in the mirror and go “Carolyn, this isn’t normal.”

But what really does normal even mean for a Stars fan? Isn’t sitting on the couch, clutching your mouth while staring at the tv during the waning minutes of the 3rd period praying that your team stops shooting long enough to win…isn’t that normal? Isn’t that what most hockey fans are doing this season?

Well, the good news is there are ways to measure “normal” with analytics. It’s called a Standard Deviation. As I briefly mentioned in the Armchair GM article, a standard deviation is the allowable “wiggle room” before a statistic can be significant. Here’s the official definition of the term per Wikipedia.

Usually, large enough data sets will have the majority of data points fall within 1 deviation (up or down) of the mean. If you ever had a test graded on a curve in the US, then you should be familiar with the concept – the majority of students will get Bs & Cs and only the very top and very bottom students get As & Ds, regardless of what % of the questions you answered correctly. Knowing that most data falls within 1 deviation, things that fall outside of that range are what is considered “significant”.*

So with that in mind, I started to take a look at the most confounding numbers the Stars’ have put up this season: Corsi For & Corsi Against. I decided to look at the actual event counts instead of percents because my starting hypothesis is that the Stars have been wildly streaky with shots against, and more consistent with shots for, so separating the two can prove that.

Deviations 5v5 Season

How to read this chart:

This is every team’s standard deviation for CF & CA events 5v5 (score adjusted). The smaller the number here, the more consistent that team is, though this isn’t an indicator of a good team or a bad team, simply one that does the same thing every night. The teams in purple are the closest to the Stars in points, because I was wondering if inconsistency was a trait of middling teams.

Why look at all these Std Devs? Because I want to know if our fluctuations are normal, and the only other way to do that is to look at other teams’ fluctuations.

As you can see, Buffalo is the most consistent team on offense – typically they will have within 6.67 CF events every night. Having just watched them play the Stars, we know they’re not a good offensive team, but now we also know their fans should never expect them to be better than they are now. At least not until next season (McEichel and Kane on the same line? That could be interesting).

On the reverse, Colorado is the least consistent defensive team. That means that their fans never know what to expect from the them – it could be on one night and off the next. This also supports the “Goaltending is propping this team up” theory, because they clearly don’t have a defense that is capable of shutting down the opposition night after night, like Nashville.

Where is Dallas? Well, on the CF side, we’re right in line with the rest of the NHL, but we’re 2nd worst in CA. Similar to the Avs, fans are snakebitten about our defensive capabilities, and this proves we have good reason to be.

It feels like, though, that things have gotten even shakier since the beginning of the year, especially with our recent Don’t Shoot To Win plan.

Devs 5v5 Scoring Jan-Feb

To take this one step further, I’ve pulled the Scoring Chances For and Against for all games since Jan 1st. Scoring Chances take into account actual probability of scoring on the shots taken, so tend to be a better predictor of winning than Corsi. Again, the Stars are outside of 1 deviation (if only barely on the For column).

It’s bad, but it could be worse: 5v5, only Toronto fans are being jerked around more with how inconsistent their team is playing on both sides of the ice.

But the fans all know the most nerve wracking part of our game since Jan 1st? That’s our special teams. So here’s one last chart (with less data, sorry**).

Devs All Scoring Jan-Feb

Well. Ok then.

When you include all situations (5v5, PK, PP, etc) with score adjusted data, the Stars become the least predictable team in both categories.

There you have it Stars fans. We love the most stressful team in the NHL – because you never know what they’re going to do.

Buffalo fans know their team is terrible. Florida fans know their team is midpack. Chicago fans know their team will be top tier. They’re pretty even keel teams.

But Stars fans? We’re taking a gamble every time we turn on the TV.

*This is really a simplified understanding, but works for the methodology used here.

**This data was chosen fairly randomly – the teams closest in points, plus the next few teams alphabetically until I hit 15 teams. Why not do all 30? Because my carpal tunnel started acting up. Just being honest.


Leave a Reply

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

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

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s