Splitting Up Duncan Keith and Brent Seabrook

Outside of Patrick Kane's impressive move to the middle, perhaps the biggest story in Chicago's impressive 7-2-2 start has been Joel Quenneville's decision to split up Duncan Keith and Brent Seabrook. The natural reaction to a coaching move so major is curiosity - why would the Blackhawks decide to split up one of the game's best pairs?

Depth

The most obvious explanation is that the move is Joel Quenneville's way of mitigating the loss of Brian Campbell. On a simple level, separating Keith and Seabrook ensures the Blackhawks will play the vast majority of their even strength minutes with at least one of their superstars on the ice, especially given Nick Leddy's relative inexperience in playing top-4 minutes.

Development

It's pretty obvious that the organization is high on Leddy (Scotty Bowman compared Leddy to Phil Housley), and I can't help but think that their confidence in his ability to eventually play top-4 minutes helped to ease the blow of trading Brian Campbell, as evidenced by the team choosing not to sign or trade for any top-4 defensemen (I believe Montador was primarily signed to solidify the bottom pair, though he obviously has shown the ability to do well in a heavier role).

I think the surprise comes not from Leddy's presence in the top-4, but mostly from who is primary defense partner has been.

Strategy/Usage

I'll note from the outset that using QualComp or any variant thereof is useless at this point in the season. There is just too much variance in strength of schedule to draw inferences from those numbers. What I will use instead is PBP data (h/t Jared).

The first section of data only focuses on Zone Starts and their Corsi numbers based off of where they started a shift.

Keith-Leddy	% of TOI	Corsi Rate
Ozone Faceoff	13.6%	33.277
All neutral	79%	3.127
Dzone Faceoff	7.4%	-38.889

Seabs-Hjalm	% of TOI	Corsi Rate
Ozone Faceoff	13.3%	47.872
All neutral	74.6%	12.54
Dzone Faceoff	12.1%	-28.18

From here we can see that the Seabrook/Hjalmarsson line is much more likely to take a defensive zone draw. We can also see that the Seabrook/Hjalmarsson pairing has performed much better territorially, no matter the situation.

As I said above, I am not using QualComp or any variant of QualComp to adjust for the toughness of the minutes. Instead, I'll use forward pairing as a proxy, as the roles in which Chicago forwards are used are pretty rigid. As we can see below, the Keith/Leddy pairing is most often used alongside Jonathan Toews and Patrick Sharp in any situation. As for Seabrook/Hjalmarsson, the forward they play with most often is David Bolland.

Ozone	Keith-Leddy	Seabs-Hjalm
Toews	41.8%	30.6%
Kane	37.9%	21.9%
Sharp	44.8%	22.5%
Hossa	31.8%	12.8%
Bolland	12.6%	39.4%
None of above	5.5%	14.1%

Neutral	Keith-Leddy	Seabs-Hjalm
Toews	32.7%	19.6%
Kane	38.2%	26%
Sharp	33%	15.3%
Hossa	33.6%	23.1%
Bolland	14.1%	44.6%
None of above	16.4%	8.8%

Dzone	Keith-Leddy	Seabs-Hjalm
Toews	33.3%	10.3%
Kane	32.3%	21.2%
Sharp	47.1%	4.6%
Hossa	23.6%	21.1%
Bolland	21.9%	62.7%
None of above	9.1%	6.8%

As I briefly mentioned above, I believe we can validly infer that the Seabrook/Hjalmarsson pairing has played tougher minutes, mainly because of how much more likely they were to play with David Bolland, whose role for the Blackhawks is well-defined as a shutdown Center. If you guys feel this is an unreasonable assumption, let me know.

Results

Here are the results of the three centers, along with the rest of the ice time, with the Keith/Leddy and Seabrook/Hjalmarsson pairings. As you can see, the Seabrook/Hjalmarsson pairing has gotten better Corsi results with each of the top three lines in the small sample we have.

Forward	Pairing	Corsi/60	Time
Toews	KL	6.063	49.5
Toews	SH	21.352	28.1
Kane	KL	7.643	55
Kane	SH	22.222	35.1
Bolland	KL	2.841	21.1
Bolland	SH	10.154	65
None	KL	-2.857	21
None	SH	-4.604	13

Conclusions/Recommendations

To be honest there are numerous explanations for why the Keith/Leddy pairing hasn't performed as well as the Seabrook/Hjalmarsson pairing. The first is that Seabrook/Hjalmarsson have played together more (in previous seasons) than Keith/Leddy have and the disparity is largely driven by a lack of familiarity. The second is that Nick Leddy isn't as good (yet) as Keith, Seabrook, or Hjalmarsson - it is possible that Leddy is dragging Keith down a bit. Finally, this could merely be variance.

As for my recommendation, I honestly see no issue with keeping these pairings together. As I noted above, all 4 players are off to solid starts, and while there is no doubting the chemistry and effectiveness in a pairing of Duncan Keith and Brent Seabrook, the fact that the Keith/Leddy pairing has done as well as it has speaks volumes to both of those players. The eye test leads me to believe reason #1 above is the best explanation for why there has been a disparity in their possession totals. I believe that as the Keith/Leddy pair grows and each player becomes more comfortable with each other, the net result for the Hawks will be positive, couple that with the long-term developmental benefit of pairing Keith and Leddy, and I see no issue with continuing this pair.