clock menu more-arrow no yes mobile

Filed under:

Stat Nerd Culmination: The Curious Case of Cam Fowler and the Role of Advanced Stats

The moment you've all been waiting for.... Daniel's Advanced Stats series concludes with his analysis of Cam Fowler.


With the numbers laid out, It's time to look at how Cam Fowler lined up in all the particular categories so we can take a larger look at how advanced stats help us understand the game.


Fowler had a 3.03 corsi and a 9.2 rel corsi. That means that every 60 minutes he played we directed three more shots at the net than were directed at ours. It also means that when he was on the ice the Ducks were directing 9 more shots on net as opposed to when he was off it.

High relative corsi is good, because it means that the team does better when that player is on the ice. Fowler was third on the team in corsi and rel corsi and was tops among our defenseman. He was also 14th in the league in rel corsi among defensemen who played over half a season. What does that mean?

It means only 13 other defenseman in the league were as valuable to their team in terms of pushing play the other way. That's pretty impressive for a guy who's corsi numbers just a year before were -17.42, and -8.4 respectively. The turnaround alone is noteworthy.


Last season, Fowler's OZ start % was 50.5. That means he was pretty much split down the middle in terms of where he started his shifts. To me, that means that his corsi numbers aren't strongly influenced by getting to start in advantageous situations, because he started in the offensive zone about as often as he started in the defensive zone. That was the third highest number among our six defensemen. It might also be considered that his coaches were comfortable playing him both offensive and defensive situations.


This might be the area where Cam needs the most improvement. The only worse point producers in ES were Beauchemin, Lydman and Pelley. Not exactly the scoring company for a top defenseman. Honestly, there's nothing pretty about it.


Fowler had a pretty low PDO last year of 962, and his 976 the year before isn't much sexier. However, that's also what is so encouraging. Cam's on ice shooting% of 7.40 was second worst among defensemen by a statistically insignificant margin. Beauchemin was last with a 7.39. Cam was also worst among all skaters with a 88.8 on ice save %.

What does this all mean? Cam was out there for an amount of offensive and defensive failures that belie his positive corsi numbers. In other words, Cam had a lot of pucks go in on him that probably shouldn't have and didn't get a lot of pucks to go in that should have. His horrible plus/minus stat is more likely the result of unlucky circumstances than horrible play. Unfortunately, plus/minus is a stat that Fowler's reputation is primarily centered around.


Cam was third among defenders in pure QoC behind Lydman and Beauchemin, but he was second in terms of rel QoC, meaning that he was most often used by the team to face the toughest competition. He finished 58th in the league in rel QoC. If we operate under the premise that top 60 represents every team's top pairing, than Fowler was bout where he should be.

Among those top 58, only 18 finished with corsi numbers higher than his. Only 18 other defenders were better at moving the puck forward while facing simiarly tough competition. That's nice, but not as impressive as being fourth in relative corsi among those defenders, meaning only three other of those defenders were more valuable to their respective teams in terms of moving play forward. There's certainly an argument to be made that as you go down the list in rel QoC those minutes get easier. However, I'm still convinced that this is an admirable showing.


Having looked at all of these statistics as they relate to Cam, it's time to start talking about the Qualitative versus Quantitative conundrum. I think if you looked at Cam from a purely Qualitative standpoint, meaning just by watching his games and not checking his numbers, you would probably think of him like a Jack Johnson. He might be seen as a guy whose mistakes don't out-weigh his offensive contributions, a PP quarterback who should be relegated to third line minutes in even strength situations to limit the damage he can do to the team defensively. You would see glaring turnovers, misreads, and Cam getting overpowered by the occasional big body up front.

If you look at Cam quantitatively, you would see a player who drew some of the toughest assignments on his team and still managed to drive play forward so that his team was taking more shots on goal than it faced while he was on the ice. I'd see a player who was a little unlucky in terms of goals for and against while he was on the ice. I'd see a player who was the 15th most valuable defensemen in the league last year in terms of moving play forward for his team. I'd see a player who struggled offensively and would be likely to see an offensive uptick.

I believe both of these players are Cam Fowler. One shows the intricacies and nuances of his game, particularly where he struggles, and one shows the overall positive effect he had on the team that can only be seen by stepping back and looking at the big picture. Both viewpoints are necessary. I believe it's seeing the forest and the trees. Quantitative analysis in hockey allows us to look at the forest, to see the bigger picture. Qualitative analysis allows us to discuss particular nuances as they happen and dissect an individual player rather than his role and overall effect on the team.

The major problem is that people misunderstand what sports statistics actually tell us. They measure effect; they do not tell us cause. However, measuring effect is not useless. Corsi cannot tell us the quality of shots given up, because the quality of a shot is a biased interpretation based on the person watching. I've seen Teemu Selanne bury a ton of shots taken from horrible angles, especially on the PP, and I've seen Bobby Ryan pull up between the circles and shoot so high it goes around the boards and back up ice as a turnover. However, no matter how poor the quality of shot, if you spend enough time in your zone the puck will inevitably go in. Also, depending on who you read, there's an argument out there that scoring chances, i.e. the quality of a shot, is in fact not relevant to predicting scoring, as shot metrics such as corsi and fenwick are equal if not better predictors.

There will always be anomalies; this is sports. However, understanding these trends gives us larger predictability for how players and teams will perform. PDO is a perfect example. Since it has been proven that PDO regresses to 1000, we can monitor our team shooting and save percentage to see if we are operating at a sustainable level. Yes, good teams will not regress as far as others, but some stats are not sustainable. The Ducks are shooting at 13% right now, and that is contributing to a large plus/minus in terms of 5-on-5 play. We have a team save percentage of 92.8 that might be closer to a true value. This gives us a PDO of 1059, tops in the league as of Feb 11. At the same time, the Ducks are 28th in shots taken.

PDO tells us that the combined number of 1059 is going to regress towards 1000, so where will the Ducks feel the pinch. My best guess would be shooting percentage as that seems the most outlandishly high. The league average in shooting percentage is currently 8.1%. Meanwhile, the league average save percentage is 91.8%. That means we are closer to league save percentage average and farther from league shooting percentage average. Therefore, our shooting percentage is the most likely to fall down. That means that no matter who is flinging the puck on net, unless the overall volume of shots we are taking drastically increases, we will see a drop in the 3.25 goals per game we are posting right now.

If you've ever taken an introductory comm course, your instructor should have talked to you about the process of perception. If he/she didn't, slap that person the next time you see him/her. I won't bore you with the details; I will only say that the first step in the process is selection. We choose to focus on certain phenomena for various reasons, but we choose nonetheless. We've all emphasized good traits of players we liked and downplayed traits we don't through the course of watching a game. Hockey is a fast moving sport and it's impossible to absorb everything that's happening 100% of the time. Statistics allow us to augment that viewpoint and enhance our overall understanding of the game. It allows us to fill the gaps on things we are missing. It helps us to look at a great start and identify the places we need to improve, or to recognize that a players few mistakes might not out-weigh his overall contribution to the team.

I encourage all of you to pursue a deeper understanding of advanced hockey statistics in order to further augment your own perceptions and advance your overall understanding of the sport and how it operates. It is impossible to look at the game today and think that observation alone will tell us everything we need to know about a player. That's an arrogance even I won't approach, and I'm arrogant.