In my analysis of the Andrew Ference signing (now including more NMC!), I wrote:
Last season, he was a 4/5 defenseman, playing even up zonestarts and had the worst relative corsi on the defensive corps. Two seasons ago, he was a 4/5 defenseman, playing even up zonestarts and had the worst relative corsi on the defensive corps. Three seasons ago, he was a 5/6 defenseman, playing even up zonestarts and was 4/6 in relative corsi on the defensive corps. Stop me if you’ve heard this before: four seasons ago he was a 5/6 defenseman, playing even up zonestarts and was 5/7 in relative corsi on the defensive corps.
Ference is an okay bottom-pairing defensemen in the Eastern Conference, and when he plays higher than that, he’s not good.
This caught Tyler Dellow's attention, who responded:
Here’s the thing about hockey data: it’s not an end in and of itself. It’s a reasonably objective record of what took place. I think that it’s better than relying on observation and recollection alone because people are not particularly well evolved to consume vast quantities of data by watching and then make good decisions. That being said, there is no number you can just look at and say "This. This says it all." You’ve got to question the context.
He goes on to make (a seemingly reasonable on the surface) argument that Andrew Ference makes players worse because Zdeno Chara makes them so much better.
Speaking of context:
There's no doubt that Chara's a great player, and saying that players did better with Chara than with Ference might not be a slight on Ference. The easiest way to the bottom of this is removing Chara's minutes from the analysis altogether. With the caveat of sample size considerations, the right most column, under the black delta symbol is Chara-free Corsi Rel, comparing how the player did when paired with Ference to how he did when paired with someone else altogether.
|w/o Ference or Chara||With Ference||▲|
Consider either the individual seasons or the players overall and two-thirds of the data is negative. Within the non-Chara tier of Boston defensemen, and we're considering the same five here in each season, Ference has been a below-average performer. As a team, the Bruins finished 29th, 15th, and 15th in even strength shots against in these three years, so "below average among defensemen other than Chara" is not exactly a ringing endorsement.
"The Chara Bump" may be a real phenomenon and Ference's team-worst Corsi Rel may underrate him a bit. But even when Chara's impact is accounted for in the analysis, we don't come away with an optimistic view of Ference's results.