Advanced Stats and the Colorado Avalanche


The Colorado Avalanche and advanced stats may as well be peas and carrots as we enter this season. Or if vegetables aren’t your thing, think of the Avs like a juicy peach, ready to be picked by the hungry pundits of the advanced stats community. Buckle up Avs fans, because for better or worse, advanced stats analysis is now a part of the game, and you’re going to hear about it. But that’s okay!

The Avs aren’t very acclaimed in the advanced stats community. The Avalanche will regress, they say. The Avs are lucky, they say. We have the pie charts to prove it, they say. In the words of Taylor Swift, “the haters gonna hate, hate, hate, hate, hate.”

My goal is to give my perspective on the world of advanced stats, and how they relate to the Avalanche. Ultimately, this all boils down to science. The science of hockey.

1. Advanced stats aren’t really advanced at all

The funniest thing about advanced stats, is there is absolutely nothing advanced about them. Seriously, don’t be scared. They don’t bite. To me, we might as well just call advanced stats… stats, because that’s all they are. A collection of data used to try and better understand the game of hockey.

The goal of most advanced stats is to try and quantify possession. The theory is, the team that has the puck more often is the better team, and thus, more likely to win. Corsi and Fenwick are the king and queen of the advanced stats world, yet despite their royalty, they are simpletons at heart. Corsi measures all shots attempted at net. Yuppp, that’s it. Addition. Super advanced stuff guys. Fenwick is no different, except blocked shots are excluded. Subtraction. Alright, now we’re flexing our mathematical muscles.

2. The science of hockey is very advanced

Remember the last time you went to the arena, and watched people flip coins for two and a half hours. It was awesome, right! Wrong. Hockey is a complex game with a lot of moving parts. That’s why we love it. Old-school folks will say you need grit, toughness, and sacrifice to win. While hockey purists will say you need skating, passing, speed, and creativity. The point is, there are many elements that influence the outcome of a hockey game.

It is a fools errand to try and predict success unequivocally with simple stats such as Fenwick and Corsi. Hockey is such a fluid sport, and as mentioned above, many factors contribute to the result of each game.

The beauty of advanced stats isn’t their raw form, but rather the fact that they represent an opportunity for a more advanced way of analyzing and thinking about the game. Statistical predictors work well for things such as flipping coins, because it is a very simple event, with two distinct outcomes. For hockey… not so much. Especially when you only use a few stats, and ignore the entire picture. It’s not impossible to come very close to explaining complex systems with numbers, but when it comes to hockey, we have a long way to go.

My opinion, is that we should be taking a more hybridized look at things. Science is about data and observation. If it works in the real world, a scientific approach should work for hockey.

3. The house/home analogy

Think of advanced stats like the foundation, framework, electrical, and plumbing of your home. You may not really care that those things are there, and that’s no big deal! I myself enjoy watching hockey without really caring or thinking about the behind the scenes stats during the game itself. That doesn’t change the fact they are there though, and that those really are the keys to having a house.

The beauty of advanced stats isn’t their raw form, but rather the fact that they represent an opportunity for a more advanced way of analyzing and thinking about the game.

That being said, nobody wants to live in a barren house. You need furniture, paint, an entertainment system, and those tacky nick-knacks before you can truly call it a home. This is where the eye test comes into play in hockey. Numbers never paint the entire story, which is where you, the viewer, come into play. In science observation is important, because you confirm your experimental data, and also can pick up on things that the numbers can’t. Things such as the shooting accuracy and power, playmaking ability, or a knack for pulling off suave dekes that each player possesses.

4. If you don’t “believe” in advanced stats, you’re wrong

Numbers never lie. I wan’t everyone to know that. Data always explains something, albeit, it is up to the savvy of the person interpreting the data to make sense of what that data is telling you.

Advanced stats do tell us useful information. Possession data is a very valuable tool, and in reality, it usually does a fairly good job at separating the good teams from the bad teams. It isn’t a coincidence that teams like the LA Kings and Chicago Blackhawks were at the top of the Corsi and Fenwick charts last season.

Teams should embrace the advanced stats movement that is coming across the league. Why wouldn’t you? Knowledge is power, and coaches and general managers should try to find every advantage they can. Any five year old can tell you, it’s easier to score if you have the puck than it is if you don’t. Just because the knowledge isn’t perfect, and doesn’t tell you everything, doesn’t mean you should ignore it entirely.

5. If you subscribe entirely to advanced stats, you’re wrong

My biggest pet peeve in all of this, is people with the following mindset. Teams who have bad possession numbers are bad at hockey… if they win, they are lucky. Eventually they have to regress to our model. This is flat out wrong. Is it possible that this is true, sure, but there are many other possible explanations. Again, let’s have a scientific mindset about this, rather than forcing every team into the cookie cutter advanced stats model we have created.

First of all, advanced stats are in their infancy. They aren’t at all cumulative or representative enough of the game itself to be used as a fool-proof predictor of success. Remember point number two? The science of hockey is complicated.

In my engineering classes, we often attempt to create models for complex real world situations. It allows us to get  a reasonable estimate or idea for what is happening. The more complex the model, the more accurately we can describe things, in general. Advanced stats are the same exact thing. Trying to create a model to predict what is going to happen during a hockey game. As the stats get more complex, the more accurate those predictions are likely to become.

The thing I can’t stress enough about statistics and modeling, is even if the model works 99% of the time, it’s still not accurate if it doesn’t account for that additional 1%. It’s not the 1% that is lying, it’s the model.  Even if advanced stats do a reasonable job at separating the good teams from the bad, you are kidding yourself if you treat them as a foolproof criteria for evaluating the quality of a team.

6. The human element

The other element of sports, is what makes them sports. Human nature. Because of the element of humanity, stats will always be somewhat limited in what they can tell you about sports, especially when it comes to predicting winning and losing.

I really do believe that people are capable of things that can’t be explained by numbers or stats. You always knew Michael Jordan was going to make that final shot, Joe Sakic would score a goal when his team needed it most, and that Tony Romo is always going to flub things up when the game is on the line. There are no statistics that can explain this. The only explanation is that some guys have it, and some don’t. In the greatest moments, some rise above, and are able to literally will their team to victory. Others simply can’t handle the magnitude of those moments.

Confidence, belief, and momentum go a long way in sports. Players and teams who have that passion, and burning desire to win – those are the players and teams who achieve greatness. Those are the players and teams that defy the odds. You have to have the talent, but to an extent, I’d take a team of players with that belief and will to win over a bunch of robots with favorable possession numbers any day.

7. How all of this relates to the Avalanche

Mar 2, 2014; Denver, CO, USA; Colorado Avalanche center Matt Duchene (9) attempts a shot on Tampa Bay Lightning goalie Ben Bishop (30) in the first period at the Pepsi Center. Mandatory Credit: Ron Chenoy-USA TODAY Sports

The Avalanche were a bad possession team last season, there’s really no way around it. Like I said, numbers don’t like. However, the Avalanche were an outstanding hockey team last year. They were the fourth highest scoring team in the league, have exciting dynamic play-makers at the forward position, and got stellar goaltending from Semyon Varlamov.

Despite this, the advanced stats community isn’t convinced the Avs are anything more than a below average team, that rode unsustainable good fortune to the Central Division crown last season. They look at things through the possession metric goggles, that prevent them realizing, the Avs may very well be a very good hockey team, despite those poor possession numbers.

8. Why advanced stats matter less to the Avs than most other hockey teams

I’ll start this with a caveat. In some ways, I do think the Avs defied the odds, and hope they do improve their possession numbers. The margin for error gets slimmer when you lose the possession battle. Just ask our five year old friend from earlier. Easier to score when you have the puck.

However, looking at the way the Avalanche play, they don’t need to possess the puck as much as most teams to be very successful.

Nathan MacKinnon and Matt Duchene are two of the fastest players in the league. At top speed it takes them mere seconds to travel the length of the ice, and create an explosive scoring chance. For Patrick Roy and the Avs, their system is all about this, winning the scoring chance battle.

You look at a team like the Bruins. They have a lot of really good hockey players, but their style is much more bruising, and less explosive. They need to have the puck, and cycle it around to wear the other team down. This is how they take over games, and start putting pressure on their opposition. The Avs really don’t play that style, because they don’t need to.

More from Avalanche News

When you think about it, shots do indirectly measure possession, because you need to have had the puck to shoot it. In reality though, a shot is a decision a player makes. He basically is saying, at this point, I feel like I am better off giving up possession of the puck, and inducing a potential element of randomness to the game, because I feel the chances I have to score or create a scoring chance are worth it.

Quality shots are much more valuable when it comes to winning than just shots. If you needed a goal, would you rather have Matt Duchene on a breakaway for one shot, or give Cody McLeod ten shots from the blue-line. Easy. You take Duchene. This is one of the weaknesses of advanced stats in their current form. They don’t asses which teams are generating the more quality scoring chances, and which team has players who are more skilled and dangerous when they do get those chances.

The Avs themselves have said their defensive philosophy is all about limiting the juicy shots. They are perfectly content to give up shots if they are on the perimeter. They have full confidence that Varly will stop those shots. Which leads us into another statement I often hear.

Well, if the Avs want to succeed, Varly is going to have to do well for them again. Duh! Every team needs their goalie to do well, and it just might be possible that Varlamov is one of the elite goalies in the league at this point. A 0.927 save percentage isn’t ludicrous, despite the large quantity of shots Varlamov faced last year, especially if the Avs method of giving up outside shots in an effort to limit higher quality scoring chances is working. I don’t see a reason why Varly can’t post a save percentage near 0.930 again this season. It’s possible he has a bad year, but I wouldn’t bet on it.

The Avs having poor Corsi and Fenwick last season doesn’t concern me at all. I feel like they have an elite goaltender, and scoring threats who can create a lot of instant offense. Plus, as guys like Duchene, MacKinnon, and O’Reilly enter their prime years, and the youthful MacKinnon gets a bit more experience, they will learn how to control the puck better and really take over games. The fact that the Avs are able to win without the supporting numbers that make the advanced stats community sleep well at night should be a scary thought. What if the Avs can combine their explosiveness and raw talent with the ability to really take over games and hold on to the puck. Goodnight NHL.

The Avs believe in themselves. They believe in Patrick Roy, and they believe in the system. I don’t need stats to get the sense that this team has a winning vibe. Roy’s fire and will to win has been infused throughout this roster. And that’s good enough for me.