Digging Into Jarrett Stoll's Possession Numbers

Another benefit of the Stoll signing is giving Rangers fans a new
#26 to scream at in the absence of the retired Martin St. Louis

        I like Jarrett Stoll. I think it was a smart signing to get somebody who can really help this team on a cheap contract. He's good on faceoffs, a solid forechecker, good defensively, can kill penalties, and is a righty with a decent shot, all of which are needs for the Rangers, particularly another penalty killer after dealing Carl Hagelin. Moreover, I highlighted in my piece on Viktor Stalberg [read it here] why the Rangers need a possession driver, and Stoll at first glance seems to fit this need. Over the last 4 seasons, the Kings take 53.9% of all shot attempts (corsi) when Stoll is on the ice, meaning he sees a lot of the puck. But as with any stat, corsi alone doesn't tell the whole story. Corsi is measured like plus/minus, meaning it isn't really stat which measures anything an individual player does, but rather, it measures what happens when that player is on the ice. Therefore, it is highly dependent on the performance of your team as a whole. For a player on a great team, the team taking 51% of the shot attempts when he's on the ice might actually be very weak. Conversely, for a player on a very below average team, 47% might actually be a solid mark. How can we tell?
       Enter CF%Rel, or corsi-for percentage relative. This is a stat which measures possession metrics relative to one's own team to determine if a player's stats are inflated or dragged down by a very good or very poor performing team. It's actually a fairly simple stat. A player's corsi for percentage is the percentage of total shot attempts taken by his team out of all the shot attempts taken when that player is on the ice. For example, if a player is on the ice for eight shot attempts, and four are taken by his team, he has a CF% of 50. CF%Rel is the difference between a player's CF% and a team's CF% when he is not on the ice. So if that 50% player is on a team which is at 55% without him, his CF%rel is -5.
         Like I said, Stoll's raw CF% of 53.9 over the last four seasons looks pretty good, but keep in mind he plays for a great possession team. That appears to be inflating his numbers, because his CF%Rel numbers over the last four seasons are 0.67, -1.17, -2.80, and a very alarming -5.78 last year. This would indicate that Stoll is a positive possession player because he plays for a great possession team, but he's not the one driving possession for them, which is why many corsi proponents dislike the Stoll signing.
       Being I am someone who uses corsi as a player evaluation tool, why do I still like the Stoll signing? Context is always king, and although CF%Rel attempts to apply some context, it's still not a perfect stat. During the same four season time-frame, the Kings have consistently been a 55 or 56% possession team, the best in the league. I believe that when players play for elite possession teams, it becomes extremely difficult for them to impact that team's possession positively. In other words, the Kings are so good at this statistic, that Stoll can't help them.
       Here's a silly made up scenario which I think will help visualize the flaw in impact-based stats such as CF%Rel. Shea Weber has probably the hardest shot in hockey. But if I were up against Shea Weber and you gave me a puck to shoot, and you gave Weber and 20-pound weight to shoot, I guarantee you I'll beat him on a radar gun. And that's the issue with impact stats. Some players are impacting poor teams, some are impacting average teams, and others, like Stoll, are impacting elite teams and have trouble making their impact felt.
      Don't get me wrong, I'm not saying Jarrett Stoll is a great possession player. A truly great possession player will positively impact their team regardless. But I do think he's a much better one than CF%Rel gives him credit for, and I believe he will positively impact an average possession team like the Rangers.