Possession stats like Corsi are a game of averages. And like most attempts to average things, they're right a statistically significant amount of times, but also wrong quite a lot. If arbitrary stats like shot attempts were how we actually decided who gets more points at the end of the game, it would be easier for me to put a lot of stock in advanced stats. What matters is pucks put in the net. Averages like Corsi are an attempt to try and make sense of large amounts of data, but they're pretty bad at making individual predictions. They can help illuminate patterns on a large scale, but they don't have a lot of predictive power when it comes to games, because hockey games can sometimes be won on things like lucky bounces, off games by goaltenders, locker room mentality, or just superstars on a tear that day.
That graph looks scary, but it would scare me a lot more if the Avs were actually the worst team in the league in terms of points, and that is very objectively not the case, you don't need fancy stats to tell you that. We could be taking more low-percentage shots in order to pad those numbers, but would that really achieve the actual goal of putting more pucks in the net than the other team? The eye test says something else. If we had lost the game every time we lost Corsi this season, we would literally have less than ten wins.
Roy's system is anti-Corsi as a stat, but when you look at how Corsi is actually notated in a lot of places as "Chances For", then maybe he DOES respect that aspect of it. His system seems to be to limit the team to lots of low-quality shots. Is a shot from the half boards that gets blocked REALLY a "chance for"? Corsi says so, and that's where my major beef with it is and why I get where Roy is coming from on this. What interests me is the admittedly subjective stat of "Scoring Chances", which doesn't really get recorded on a large scale. Those are the high-danger chances where the individual players matter. Did the shooter miss? Did the goalie make a huge save? That's where hockey games are won and lost, IMO. Does your D give up too many of those big chances? That's what sunk us in Minnesota last week. Do your shooters miss critical chances? That's why we lost the last game in Vancouver.
I think that's where Roy's philosophy lies. Generate as many of those big chances off the rush as you can, and limit the other team to bad ones as much as possible. It relies on your forwards to not choke, and your defenders to keep the puck to the outside as much as possible in the D zone, something that the Avs are actually pretty good at. What sinks us is that our weak defenders like Holden, Guenin, and Bodnarchuck don't make good decisions in one-on-one battles defending those high-percentage chances, leaving the goalie out to dry. How many of those massive ****ups this year have led to brutal goals against? Which is why I think once our defensive corps matures, loses the dead weight and maybe switches up the pairings to give the weak defenders better matchups and cover when they screw up, we'll be a much better team even if Corsi says we still suck, because we'll be better able to implement Roy's actual system with better players. How many 2-on-1s has Beauch broken up this year just from solid defense? If we had more of those kinds of players rather than the loser squad we'd be in much better shape once you look at the game this way. I'd also guess that that's why Zadorov has a short leash, because he does a lot of things right but also makes those boneheaded mistakes sometimes that lead to scoring chances against, which is exactly what Roy wants to avoid.
I'd hazard a guess that Roy's experience as a goalie and his intimate familiarity with scoring chances and how much those matter are informing his philosophy that low-danger shots don't matter as much as high-danger ones. Now, do I have a problem with some of the other technical things that Roy does with his systems? You bet your ass I do. Do I think the Avs would be a better team under a better coach? Probably. But I actually agree with his underlying philosophy that Corsi is a flawed stat that doesn't tell the whole story.