Defense pairs

[Buppy 1,720]

With all the talk about Kindl, and Smith being scratched for E, made me think we could use a thread for our defense pairs.

To kick things off I figured I'd look up some stats. http://www.puckalytics.com/superwowy.html My new favorite toy.

Not counting Kindl, we've had 8 different pairings together for significant time, and a handful of other combos for at least a few games. Stats for the top 8 pairs below:

Pairing	            5v5 Minutes	GF/60	GA/60	GF%	CF/60	CA/60	CF%	OvDZS%	Goals +/-
Ericsson + Dekeyser	347	2.42	1.38	63.68%	43.23	51.18	45.79%	47.10%	+6
Kronwall + Green      	336	1.61	2.50	39.17%	51.06	47.85	51.62%	62.50%	-5
Smith + Green	        285	2.10	1.89	52.63%	62.07	42.92	59.12%	74.80%	+1
Dekeyser + Quincey	273	1.32	2.42	35.29%	44.51	51.56	46.33%	49.20%	-5
Kronwall + Ericsson	230	1.83	1.30	58.47%	46.93	57.36	45.00%	46.60%	+2
Smith + Marchenko	224	2.94	2.67	52.41%	52.41	43.58	54.60%	52.80%	+1
Dekeyser + Marchenko	147	2.04	1.63	55.59%	44.50	50.63	46.78%	33.30%	+1
Ericsson + Marchenko 	124	2.42	2.42	50.00%	52.30	48.43	51.92%	46.50%	E

Kronwall + Green was notably bad.

E + Deke have been the best pairing, results-wise. The poor possession metrics suggest that might be anomalous, but could also be limiting opponents to low quality chances. Both E and Dekeyser see the same thing with Kronwall and Marchenko respectively.

Smith + Green was very sheltered, but had great possession numbers. Goal numbers both for and against not so great relative to the shots. Could also be anomalous.

Deke + Q has been bad, though they were good last year.

Kronwall + E better than last year, but still poor offense and possession numbers.

Marchenko with both Smith and E has GA numbers higher than they should be, could be an anomaly or too many high-quality chances against.

Deke + Marchenko has been very good for being so heavily used in the defensive zone.

Quincey is maybe under-represented since he's only played about 1/2 as much as the others.

But overall, E+Deke and Smith+Green have both been good. Q looks expendable to me, especially if Kronwall and March work together.

Discuss

Edited February 28, 2016 by Buppy

[matt198913 932]

Translation Smith should be playing. Ericsson is a product of playing with Dekeyser and we should trade Wuincey before he completely kills our souls

[kliq 3,755]

Not to dismiss this as it definitely has a place, but the one thing I don't like about stats like this are they don't account for quality of opposition (unless I am missing something).

For example, a pair like Kronwall/E (when they were the first pairing) is likely going up every night against the best of the best, while a pairing like E and Marchenko are likely going up against lesser competition.

Essentially Marchenko/E have an easier shot of performing better together, so IF they had the better numbers, it doesn't mean they are necessarily the better pairing.

[kickazz 5,459]

I need to see quality of competition to be honest because this could easily be misinterpreted otherwise.

Edited February 28, 2016 by kickazz

[Jesusberg 1,256]

I see where the underlying numbers have their place, but as others have alluded to, QoC isn't always factored in.

Sometimes I get tired of seeing people (not you guys, the bloggers, etc.) preaching on about how great "Player X" is and that it's just a matter of time. If his underlying numbers are based on third pairing duties, they might not necessarily translate if he were given a larger role. There are just a lot of factors that people leave out when discussing analytics (Again, not directed toward you, Buppy. Just a general comment about how people sometimes lean on them too much, IMO)

Edited February 28, 2016 by Jesusberg

[Buppy 1,720]

Prevailing opinion on QoC seems to be that it balances out over time. For all that you try to get favorable matchups for your guys, the opposing coach does the same. Not that it would end up exactly equal, but that it wouldn't have as much impact as people think, at least for players that play a lot of minutes. I put together some numbers for quality of both teammates and competition so you can see the differences aren't huge. Someone should probably do some recursive analysis, but that's far more work than I want to do unless someone's paying me.

Player Name	TOIC	OGF60	OGA60	OGF%	CCF60	CCA60	CorC%
DEKEYSER	17.79	2.19	2.11	51	54.46	54.11	50.16
ERICSSON	17.73	2.17	2.13	50.4	54.59	53.8	50.37
GREEN	        17.18	2.06	2.1	49.5	53.57	53.54	50.02
KRONWALL	17.51	2.1	2.14	49.5	54.11	53.71	50.18
MARCHENKO	17.27	2.09	2.08	50.1	53.53	53.78	49.88
QUINCEY	        17.73	2.18	2.16	50.3	54.44	55.55	49.5
SMITH	        17.09	2.04	2.08	49.5	53.11	53.6	49.77
							
Player Name	TOIT	TMGF60	TMGA60	TMGF%	TCF60	TCA60	CorT%
DEKEYSER	17.02	2.13	1.99	51.7	50.97	48.63	51.18
ERICSSON	16.79	1.92	2.04	48.5	50.41	48.84	50.79
GREEN	        16.7	2.12	1.78	54.4	52.03	47.52	52.27
KRONWALL	16.9	2.15	1.9	53.1	51.61	48.62	51.49
MARCHENKO	16.53	2.02	1.87	52	51.85	47.83	52.01
QUINCEY	        17.3	2.07	1.85	52.8	51.7	49.02	51.33
SMITH	        16.36	1.78	1.91	48.3	52.43	48.1	52.16

TOI is time on ice per 60 minutes 5v5 time only, not per game. Difference of less than a minute top to bottom. Fairly small range across all the different stats. Players that face better competition usually also play with better teammates. I believe it's a factor worth considering, if someone ever puts more work into it, but I don't think it would make any huge difference.

[Ekmanc 586]

I don't know anything about this but how would it balance out over time? the top pair will always play against the top line of the other team, or as much as possible anyway. It's not like both teams just roll all 4lines equally which would have to be the case for it to even out.

A first dpair will always have a tougher matchup than a third pair will.

[Buppy 1,720]

I don't know anything about this but how would it balance out over time? the top pair will always play against the top line of the other team, or as much as possible anyway. It's not like both teams just roll all 4lines equally which would have to be the case for it to even out.

A first dpair will always have a tougher matchup than a third pair will.

It balances because over time, there will be shifts where the opposing team gets their top players out against your bottom players, or vice-versa. Injuries/line-juggling move a player up/down in the lineup.

And it's not like teams have only one good player or line. A top player may have a lot more time against a specific player (like Dekeyser may play more minutes against Crosby), but if you look more of an overall picture (like Dekeyser against "good offensive players") it will be more even.

Yes, if you compare a player with a coach making a concerted effort to shelter him to a player with a coach giving them all the toughest assignments, there will be a difference (though I'd bet less than you'd think). But that's the absolute extreme.

http://war-on-ice.com/series.html will show head-to-head matchups for all games between two teams in a given season. You can use that to look at who a player played with and against to get an idea. For example:

On the actual site, you can mouse-over the squares to see actual minutes.

For the Wings, Kronwall, Ericsson, Dekeyser, and Quincey are all used about the same. Smith, Green, and Marchenko tend to be more sheltered. I'm not going to go through every bit of data for all the hundreds of players in the league, but I have to believe that if there was a big difference, it would show up in the stats.

I suspect that no matter what, people who don't like what the data says are going to dismiss it anyway.