Reunite Ham and Juice thread.

[Apples]

I just want Tanev back in the lineup when the playoffs starts. He plays great with anyone and makes up for his partner mistakes

Ham Juice

Garrison Edler

Ballard Tanev

[GoaltenderInterference]

Bieksa needs Hamhuis more than Hamhuis needs Bieksa.

[Drouin]

Garrison-Hamhuis

Edler-Tanev

Ballard-Bieska

Garrison has played great with hamhuis.

Edler and tanev can be great defensive players at their top.

B-B Bieska has played worse than Edler, so Bieska can get back to old juice.

[needtogetswole]

why would we try something when nothing is broken at the back?

Edlers finally coming around

Ballard and Tanev been solid

only player I'm worried about right now is Bieksa but you know he'll come around just has had a few bad games...

our team is fine the way it is.

only thing I have a problem with is EBBETT - god damit i hate him.

[Edlerberry]

I too would like to see Eddie and Garrison back together. Edlers slowly pulling his head out from his backside, and Garrison has his flow goin good these days.

[Bananas]

Ham and juice walk into a bar and the bartender says "sorry we dont serve food here"

[Everybody Hates Raymond]

Anyone suggesting Garrison gets moved to anywhere but the first pairing is insane. Garrison is our best defenseman. If he keeps improving he could turn into that legitimate number one we've been looking for. God damn beast of a player

He's nearly PPG in the playoffs as well (although one NHL series is an incredibly small sample size, he's not at fault for that)

[SID.IS.SID.ME.IS.ME]

Reunite Ham-Juice for the playoffs? It's an interesting idea and there's certainly some good history there. Recently, I wrote a stats-heavy post discussing the D pairings, most notably the reasons why Edler-Bieksa and Hamhuis-Garrison made sense (and why Hamhuis-Bieksa didn't make sense).

In that case, I mostly considered GF%. GF% is the percentage of total 5v5 goals scored during while a player/pairing/line/team is on the ice that are goals-for (ie: player X is on the ice for 6 goals-for and 4 goals-against, therefore player X has a GF% of 6/(6+4) = 6/10 = 60%).

GF% is great for the regular season because it basically tells you what combinations and pairings put a team in the best position to win games. It's an excellent stat for choosing lineups.

However, we've all seen how the game changes once the playoffs roll around. Great regular season stats don't necessarily translate into great playoff performances. The funny thing is, there actually is a set of regular season stats that translate well as predictors of a team's overall playoff performance. Generally speaking, these are advanced stats differentials that are calculated from a team's numbers at 5v5 in "CLOSE" situations.

What is 5v5 CLOSE? Basically, these are the even strength minutes of games where the score is close. More specifically, the EV (5v5) minutes during the first and second periods when the score is either tied or within one goal and the third period minutes where the score is tied (this is the formula the stats geeks have chosen).

Recently, there have been a few articles that have looked at team 5v5 CLOSE stats and determined that the teams with the best GF% CLOSE and FF% (Fenwick-for percentage) CLOSE are the teams that tend to generally perform better in the playoffs and are almost always the teams that end up winning the Stanley Cup. Here's a very good article (the best one of the recent bunch) on this phenomenon:

http://hockeyanalysi...layoff-success/

Over the past 5 seasons, the teams that have put-up GF% (Close) over 55 and FF% (Close) over 53, during the regular season, have gone on to the greatest success in the playoffs. All five Cup-winners had GF% over 55 during the regular season. 3 of 5 had FF% over 53 (and the other two had FF% between 50 and 53). Only one team with a GF% over 55 (and zero teams with FF% over 53) has missed the playoffs during this period.

These numbers don't lie. If a team puts up excellent GF% and FF% during 5v5 CLOSE in the regular season, that team has a vastly superior chance of doing well in the playoffs (compared to the teams that don't put-up the numbers). Here's some good news: the Canucks are one of the few NHL teams that has excellent GF% and FF% during 5v5 CLOSE (Vancouver is currently #3 overall in GF% at 59.6 and #7 overall in FF% 53.2: http://stats.hockeya...CT&sortdir=DESC).

Given that the standard for team stats is >55 GF% (Close) and >53 FF% (Close), I figured that this would also be a good way to determine the best defensive pairings for the playoffs. Basically, I'm theorizing that the Canucks should choose a defensive lineup that pairs defensemen who put-up the best GF% and FF% when they play together.

WOWYs ("with or without you"--stats for when players play together and apart) from Stats.HockeyAnalysis.com track the GF% numbers for all NHL players but don't tract FF% (WOWY) for individual players. However, they do track CF% (Corsi-for percentage), which should be very close to FF% (please note: Corsi=shots+missed shots+blocked shots; Fenwick=shots+missed shots). With that all said, here's what would seem to be the best choices for defensive pairings, based on statistical evidence from the 2013 regular season (all stats from 5v5 Close situations):

Edler-Garrison (62.5 GF%, 55.2 CF%)

Hamhuis-Tanev (100 GF%, 53.3 CF%)

Alberts-Bieksa (66.7 GF%, 54.0 CF%)

The above players all provide stats (when paired together) that are better than the Cup-winning benchmarks of 55 GF% and 53 CF% (substituted for FF%). Here are the stats for some other possibilities (including some of the "popular" options):

Edler-Bieksa (100 GF%, 53.3 CF%)

Hamhuis-Garrison (50.0 GF%, 55.3 CF%)

Ballard-Tanev (50.0 GF%, 53.8 CF%)

Hamhuis-Bieksa (60.0 GF%, 49.6 CF%)

Edler-Tanev (25.0 GF%, 59.6 CF%)

Ballard-Garrison (33.3 GF%, 64.0 CF%)

A few things worth noting:

The Canucks' current top-four (Edler-Bieksa and Hamhuis-Garrison), while putting up the best 5v5 stats in all situations (total EV TOI), doesn't put-up the best stats during close situations. Edler-Bieksa, to the surprise of many, put-up excellent numbers across the board and in all situations but that pairing forces Garrison into a suboptimal role with Hamhuis, Alberts, or Ballard (in terms of 5v5 Close and its predictive value for playoff performance). That's why Edler-Garrison, for the playoffs, becomes preferable over Edler-Bieksa.

Hamhuis-Garrison has been very strong through the regular season, putting up good overall numbers in all situations but, in terms of 5v5 Close and playoff performance prediction, we see that Hamhuis can get a significant boost from playing alongside Tanev instead of Garrison.

The thread itself was about the Hamhuis-Bieksa (Ham-Juice) pairing and, while that pairing does put-up good GF% (60.0), their CF% (49.6) isn't all that impressive. Better off, again likely to the surprise of many, to pair Alberts-Bieksa (66.7 GF%, 54.0 CF%).

I've ranted significantly in the past about Edler-Tanev as a pairing. They're just terrible together. Their overall stats together are horrendous and there are probably no two other players on the Canucks (who have played significant minutes together) who have a worse effect on each other's performance. GF% tells the story clearest with the pairing putting up a miserable 25.0 (in 5v5 close situtations). Play them away from each other (ie: paired with anyone else) and Edler's GF% (Close) jumps to 60.9 and Tanev's to 64.3. Their CF% (Corsi) might create something of a distraction, in that a 59.6 CF% is quite high. However, this becomes far less significant when you consider that both Edler and Tanev put-up overall CF% numbers (ie: their average with all partners) that are in the mid-to-high 50's. From what we've seen over the 116:30 (and 75:20 in close situations) that these two have played together this season, Edler-Tanev probably shouldn't play alongside each other again until next season's training camp (and maybe not even then) and they certainly shouldn't be considered as a pairing for this postseason.

And finally, a pairing that I've wanted to see tried together during the regular season is Ballard-Bieksa (they have a 100 GF% in their 2013 EV TOI). However, their CF% (Close) is only 40.0 and they have only played a TOI (together) of 14:33 (of 5v5 Close). Alberts and Bieksa have played more than twice the TOI together this season and have excellent stats for both GF% and CF%. Even though I'd like to see KB4-KB3, I have to pick AA-Juice for the final playoff pairing.

For those interested, here are links to the 2013 WOWY stats (GF%, CF%, and much more) for each of the top seven Canucks' defensemen (for 5v5 CLOSE situations & with zone starts adjusted):

Edler: http://stats.hockeya...13&sit=5v5close

Bieksa: http://stats.hockeya...13&sit=5v5close

Hamhuis: http://stats.hockeya...13&sit=5v5close

Garrison: http://stats.hockeya...13&sit=5v5close

Ballard: http://stats.hockeya...13&sit=5v5close

Tanev: http://stats.hockeya...13&sit=5v5close

Alberts: http://stats.hockeya...13&sit=5v5close

For an extra bonus, here are the GF% and CF% stats (from 5v5 close situations) for each of the top seven defensemen, compared for whether they're playing in front of Cory Schneider or Roberto Luongo. I wanted to check this myself, just out of curiosity, so I figured I'd share. I'll put the results in a spoiler, as this post is already quite lengthy:

Edler: with Schneider (43.8 GF%, 56.5 CF%); with Luongo (72.7 GF%, 57.1 CF%)

Bieksa: with Schneider (66.7 GF%, 49.3 CF%); with Luongo (80.0 GF%, 53.5 CF%)

Hamhuis: with Schneider (66.7 GF%, 50.8 CF%); with Luongo (30.0 GF%, 61.8 CF%)

Garrison: with Schneider (60.0 GF%, 51.2 CF%); with Luongo (58.3 GF%, 58.9 CF%)

Ballard: with Schneider (50.0 GF%, 54.7 CF%); with Luongo (60.0 GF%, 53.9 CF%)

Tanev: with Schneider (45.5 GF%, 54.7 CF%); with Luongo (71.4 GF%, 52.9 CF%)

Alberts: with Schneider (40.0 GF%, 58.3 CF%); with Luongo (66.7 GF%, 42.2 CF%)

I've posted previously on how the Canucks players seem to play differently in front of each of the two goaltenders, with some players simply "better" with Schneider and others "better" with Luongo. Some people have a problem with this idea, believing that the goaltender shouldn't actually effect the performance of the players in front of him, and these people tend to discount numbers like those above. I tend to see them as indicators of the complex relationships (both on and off ice) between players.

On the ice, it's especially easy to see how a defenseman might play better with one goalie than with another. There is a lot of interaction between the D and the goaltender. How a defenseman pressures shooters, blocks/tips shots, screens his own net, etc. is going to effect the goalie behind him. How well a goalie copes with screens and redirects is going to effect the defensemen in front of him. Same can be said for how a D clears the crease and whether he plays the body or the stick. This can effect the goalie and how the goalie copes in these situations can effect the D. There are countless other on-ice situations where one goaltender can be found to be a better "fit" than the other goaltender, depending on the variety of factors that make up the complex playing style of each defendor playing in front of the goalies. So yes, I certainly believe that the goalie in the net can effect how well a defenseman plays the game and therefore, that there are defensemen on this team that are better with Schneider and others who are better with Luongo.

And that's not even getting into the mental side of things. Players might simply prefer one goalie over the other due to personality traits, loyalty, friendship, etc. Players might perceive one goalie to be better than the other. They might have greater confidence with one or the other. There are too many factors to list.

When we are given stats that show variances like that of Edler (GF% goes from 43.8 with Schneids to 72.7 with Lu), Hamhuis (GF% from 66.7 to 30.0), or Tanev (GF% from 45.5 to 71.4), we can clearly see that some members of this team seem to change into different players depending on who is in the net. Given this factor, there might be one more thing to consider before picking pairings for the playoffs: who is starting in net?

Most of us are assuming that Schneider gets the start, so the stats in front of CS become the important ones. Edler, Tanev, and Alberts are the weakest defensemen, in terms of GF%, when Schneider is in the net. If those numbers hold up as indicators of playoff performance, then the weaker-with-Schneids defensemen will need to be paired with guys who perform extra well in front of Schneider (like Bieksa, Hamhuis, and Garrison).

There's also some Corsi differences (Bieksa, Hamhuis, and Garrison are slightly lower in CF%, when Schneider plays), but these are less significant and most of the CF% are good, regardless of who plays goal (only Bieksa with Schneider and Alberts with Luongo yield a CF% under 50).

Interestingly, the general trend we see is better numbers with Luongo than with Schneider. Overall, the team has a GF% of 56.2 and a CF% of 53.1 when Cory Schneider plays (in 5v5 close situations). With Luongo, the overall team numbers are a GF% of 60.7 and a CF% of 56.0. Both goalies have great team numbers (Stanley Cup contendor quality) but Luongo's are better.

If the team goes with Luongo, the defenseman of concern is Dan Hamhuis, who has an awful GF% (close) of 30.0. If the stats hold up as indicators, Hamhuis will need to be "protected" by pairing him with a stronger-with-Lu player (Bieksa or Tanev would be the best choices here).

So, if we are to consider the goaltenders, in determining the best defensive pairings, the pairings I listed earlier in the post offer the best balance, for all conditions:

Edler-Garrison

Hamhuis-Tanev

Alberts-Bieksa

Garrison, Hamhuis, and Bieksa are the strongest (statistically) defensemen with Schneider in the net. They are spread through the pairings, one on each, to balance out any weaknesses. Hamhuis, who is statistically weak with Luongo, is balanced with Tanev, who is strong with Lu. These pairings not only offer the best top-to-bottom options, based on the WOWYs between the various defensemen, but they offer balance in each pairing that compensates for statistical differences in the defensemen's quality of play dependent on the goaltending. Basically, in terms of 5v5 Close GF% and CF% (again, used here as a best replacement stat for FF%, as Fenwick-for wasn't available for individual players), those pairings should give the Canucks the best predictive value for playoff success (assuming that the strong statistical evidence based on team stats can also translate into predictive value based on individual player stats).

[canuckelhead70]

I think KB needs a game off, call it a maintenace day or benching, he has be awful the last few games. If this was Ballard, Ballard would probably have been sent down to Chicago.

[DownUndaCanuck]

Bieksa and Edler together is just asking for trouble, especially in the playoffs when speed will kill both of them. Both get caught flat-footed far too often and both are guilty of turnovers.

Garrison is surprisingly quick on his feet and rarely gives up turnovers.

Hamhuis sits back a fair bit and rarely gets beaten around by speed (as is Tanev).

Makes sense that you pair one of our good D-men with our poor, offensive ones, but right now Garrison and Hamhuis are playing well together. In the playoffs though, I'd like to see this for Game 1:

Edler - Garrison (both can play physically, both bring offence and Garrison can mop up Edler's mistakes)

Bieksa - Hamhuis (as above, but Bieksa is the physical one)

Alberts - Tanev/Barker (need at least one physical beast and Alberts is that).

[avelanch]

+

Anyone miss this D pairing?

The time is now to reunite them, Garrisons been a beast defensively lately just check his +/-

He could cover Edler's mistake and hes used to playing on the right side now...

Soo I think the best shutdown line from 10/11 should reunite, especially come playoff time.

[TOMapleLaughs]

[Ohnoeszz]

What is the sample size for Edler/Tanev together? I can't recall them being together for any meaningful amount of time that would make their stats together relevant... could be wrong though....

[Gustavo Fring]

question, why did you use orange juice? his nicname comes from his love of pineapple juice:

[higgyfan]

Reunite Ham-Juice for the playoffs? It's an interesting idea and there's certainly some good history there. Recently, I wrote a stats-heavy post discussing the D pairings, most notably the reasons why Edler-Bieksa and Hamhuis-Garrison made sense (and why Hamhuis-Bieksa didn't make sense).

In that case, I mostly considered GF%. GF% is the percentage of total 5v5 goals scored during while a player/pairing/line/team is on the ice that are goals-for (ie: player X is on the ice for 6 goals-for and 4 goals-against, therefore player X has a GF% of 6/(6+4) = 6/10 = 60%).

GF% is great for the regular season because it basically tells you what combinations and pairings put a team in the best position to win games. It's an excellent stat for choosing lineups.

However, we've all seen how the game changes once the playoffs roll around. Great regular season stats don't necessarily translate into great playoff performances. The funny thing is, there actually is a set of regular season stats that translate well as predictors of a team's overall playoff performance. Generally speaking, these are advanced stats differentials that are calculated from a team's numbers at 5v5 in "CLOSE" situations.

What is 5v5 CLOSE? Basically, these are the even strength minutes of games where the score is close. More specifically, the EV (5v5) minutes during the first and second periods when the score is either tied or within one goal and the third period minutes where the score is tied (this is the formula the stats geeks have chosen).

Recently, there have been a few articles that have looked at team 5v5 CLOSE stats and determined that the teams with the best GF% CLOSE and FF% (Fenwick-for percentage) CLOSE are the teams that tend to generally perform better in the playoffs and are almost always the teams that end up winning the Stanley Cup. Here's a very good article (the best one of the recent bunch) on this phenomenon:

http://hockeyanalysi...layoff-success/

Over the past 5 seasons, the teams that have put-up GF% (Close) over 55 and FF% (Close) over 53, during the regular season, have gone on to the greatest success in the playoffs. All five Cup-winners had GF% over 55 during the regular season. 3 of 5 had FF% over 53 (and the other two had FF% between 50 and 53). Only one team with a GF% over 55 (and zero teams with FF% over 53) has missed the playoffs during this period.

These numbers don't lie. If a team puts up excellent GF% and FF% during 5v5 CLOSE in the regular season, that team has a vastly superior chance of doing well in the playoffs (compared to the teams that don't put-up the numbers). Here's some good news: the Canucks are one of the few NHL teams that has excellent GF% and FF% during 5v5 CLOSE (Vancouver is currently #3 overall in GF% at 59.6 and #7 overall in FF% 53.2: http://stats.hockeya...CT&sortdir=DESC).

Given that the standard for team stats is >55 GF% (Close) and >53 FF% (Close), I figured that this would also be a good way to determine the best defensive pairings for the playoffs. Basically, I'm theorizing that the Canucks should choose a defensive lineup that pairs defensemen who put-up the best GF% and FF% when they play together.

WOWYs ("with or without you"--stats for when players play together and apart) from Stats.HockeyAnalysis.com track the GF% numbers for all NHL players but don't tract FF% (WOWY) for individual players. However, they do track CF% (Corsi-for percentage), which should be very close to FF% (please note: Corsi=shots+missed shots+blocked shots; Fenwick=shots+missed shots). With that all said, here's what would seem to be the best choices for defensive pairings, based on statistical evidence from the 2013 regular season (all stats from 5v5 Close situations):

Edler-Garrison (62.5 GF%, 55.2 CF%)

Hamhuis-Tanev (100 GF%, 53.3 CF%)

Alberts-Bieksa (66.7 GF%, 54.0 CF%)

The above players all provide stats (when paired together) that are better than the Cup-winning benchmarks of 55 GF% and 53 CF% (substituted for FF%). Here are the stats for some other possibilities (including some of the "popular" options):

Edler-Bieksa (100 GF%, 53.3 CF%)

Hamhuis-Garrison (50.0 GF%, 55.3 CF%)

Ballard-Tanev (50.0 GF%, 53.8 CF%)

Hamhuis-Bieksa (60.0 GF%, 49.6 CF%)

Edler-Tanev (25.0 GF%, 59.6 CF%)

Ballard-Garrison (33.3 GF%, 64.0 CF%)

A few things worth noting:

The Canucks' current top-four (Edler-Bieksa and Hamhuis-Garrison), while putting up the best 5v5 stats in all situations (total EV TOI), doesn't put-up the best stats during close situations. Edler-Bieksa, to the surprise of many, put-up excellent numbers across the board and in all situations but that pairing forces Garrison into a suboptimal role with Hamhuis, Alberts, or Ballard (in terms of 5v5 Close and its predictive value for playoff performance). That's why Edler-Garrison, for the playoffs, becomes preferable over Edler-Bieksa.

Hamhuis-Garrison has been very strong through the regular season, putting up good overall numbers in all situations but, in terms of 5v5 Close and playoff performance prediction, we see that Hamhuis can get a significant boost from playing alongside Tanev instead of Garrison.

The thread itself was about the Hamhuis-Bieksa (Ham-Juice) pairing and, while that pairing does put-up good GF% (60.0), their CF% (49.6) isn't all that impressive. Better off, again likely to the surprise of many, to pair Alberts-Bieksa (66.7 GF%, 54.0 CF%).

I've ranted significantly in the past about Edler-Tanev as a pairing. They're just terrible together. Their overall stats together are horrendous and there are probably no two other players on the Canucks (who have played significant minutes together) who have a worse effect on each other's performance. GF% tells the story clearest with the pairing putting up a miserable 25.0 (in 5v5 close situtations). Play them away from each other (ie: paired with anyone else) and Edler's GF% (Close) jumps to 60.9 and Tanev's to 64.3. Their CF% (Corsi) might create something of a distraction, in that a 59.6 CF% is quite high. However, this becomes far less significant when you consider that both Edler and Tanev put-up overall CF% numbers (ie: their average with all partners) that are in the mid-to-high 50's. From what we've seen over the 116:30 (and 75:20 in close situations) that these two have played together this season, Edler-Tanev probably shouldn't play alongside each other again until next season's training camp (and maybe not even then) and they certainly shouldn't be considered as a pairing for this postseason.

And finally, a pairing that I've wanted to see tried together during the regular season is Ballard-Bieksa (they have a 100 GF% in their 2013 EV TOI). However, their CF% (Close) is only 40.0 and they have only played a TOI (together) of 14:33 (of 5v5 Close). Alberts and Bieksa have played more than twice the TOI together this season and have excellent stats for both GF% and CF%. Even though I'd like to see KB4-KB3, I have to pick AA-Juice for the final playoff pairing.

For those interested, here are links to the 2013 WOWY stats (GF%, CF%, and much more) for each of the top seven Canucks' defensemen (for 5v5 CLOSE situations & with zone starts adjusted):

Edler: http://stats.hockeya...13&sit=5v5close

Bieksa: http://stats.hockeya...13&sit=5v5close

Hamhuis: http://stats.hockeya...13&sit=5v5close

Garrison: http://stats.hockeya...13&sit=5v5close

Ballard: http://stats.hockeya...13&sit=5v5close

Tanev: http://stats.hockeya...13&sit=5v5close

Alberts: http://stats.hockeya...13&sit=5v5close

For an extra bonus, here are the GF% and CF% stats (from 5v5 close situations) for each of the top seven defensemen, compared for whether they're playing in front of Cory Schneider or Roberto Luongo. I wanted to check this myself, just out of curiosity, so I figured I'd share. I'll put the results in a spoiler, as this post is already quite lengthy:

Edler: with Schneider (43.8 GF%, 56.5 CF%); with Luongo (72.7 GF%, 57.1 CF%)

Bieksa: with Schneider (66.7 GF%, 49.3 CF%); with Luongo (80.0 GF%, 53.5 CF%)

Hamhuis: with Schneider (66.7 GF%, 50.8 CF%); with Luongo (30.0 GF%, 61.8 CF%)

Garrison: with Schneider (60.0 GF%, 51.2 CF%); with Luongo (58.3 GF%, 58.9 CF%)

Ballard: with Schneider (50.0 GF%, 54.7 CF%); with Luongo (60.0 GF%, 53.9 CF%)

Tanev: with Schneider (45.5 GF%, 54.7 CF%); with Luongo (71.4 GF%, 52.9 CF%)

Alberts: with Schneider (40.0 GF%, 58.3 CF%); with Luongo (66.7 GF%, 42.2 CF%)

I've posted previously on how the Canucks players seem to play differently in front of each of the two goaltenders, with some players simply "better" with Schneider and others "better" with Luongo. Some people have a problem with this idea, believing that the goaltender shouldn't actually effect the performance of the players in front of him, and these people tend to discount numbers like those above. I tend to see them as indicators of the complex relationships (both on and off ice) between players.

On the ice, it's especially easy to see how a defenseman might play better with one goalie than with another. There is a lot of interaction between the D and the goaltender. How a defenseman pressures shooters, blocks/tips shots, screens his own net, etc. is going to effect the goalie behind him. How well a goalie copes with screens and redirects is going to effect the defensemen in front of him. Same can be said for how a D clears the crease and whether he plays the body or the stick. This can effect the goalie and how the goalie copes in these situations can effect the D. There are countless other on-ice situations where one goaltender can be found to be a better "fit" than the other goaltender, depending on the variety of factors that make up the complex playing style of each defendor playing in front of the goalies. So yes, I certainly believe that the goalie in the net can effect how well a defenseman plays the game and therefore, that there are defensemen on this team that are better with Schneider and others who are better with Luongo.

And that's not even getting into the mental side of things. Players might simply prefer one goalie over the other due to personality traits, loyalty, friendship, etc. Players might perceive one goalie to be better than the other. They might have greater confidence with one or the other. There are too many factors to list.

When we are given stats that show variances like that of Edler (GF% goes from 43.8 with Schneids to 72.7 with Lu), Hamhuis (GF% from 66.7 to 30.0), or Tanev (GF% from 45.5 to 71.4), we can clearly see that some members of this team seem to change into different players depending on who is in the net. Given this factor, there might be one more thing to consider before picking pairings for the playoffs: who is starting in net?

Most of us are assuming that Schneider gets the start, so the stats in front of CS become the important ones. Edler, Tanev, and Alberts are the weakest defensemen, in terms of GF%, when Schneider is in the net. If those numbers hold up as indicators of playoff performance, then the weaker-with-Schneids defensemen will need to be paired with guys who perform extra well in front of Schneider (like Bieksa, Hamhuis, and Garrison).

There's also some Corsi differences (Bieksa, Hamhuis, and Garrison are slightly lower in CF%, when Schneider plays), but these are less significant and most of the CF% are good, regardless of who plays goal (only Bieksa with Schneider and Alberts with Luongo yield a CF% under 50).

Interestingly, the general trend we see is better numbers with Luongo than with Schneider. Overall, the team has a GF% of 56.2 and a CF% of 53.1 when Cory Schneider plays (in 5v5 close situations). With Luongo, the overall team numbers are a GF% of 60.7 and a CF% of 56.0. Both goalies have great team numbers (Stanley Cup contendor quality) but Luongo's are better.

If the team goes with Luongo, the defenseman of concern is Dan Hamhuis, who has an awful GF% (close) of 30.0. If the stats hold up as indicators, Hamhuis will need to be "protected" by pairing him with a stronger-with-Lu player (Bieksa or Tanev would be the best choices here).

So, if we are to consider the goaltenders, in determining the best defensive pairings, the pairings I listed earlier in the post offer the best balance, for all conditions:

Edler-Garrison

Hamhuis-Tanev

Alberts-Bieksa

Garrison, Hamhuis, and Bieksa are the strongest (statistically) defensemen with Schneider in the net. They are spread through the pairings, one on each, to balance out any weaknesses. Hamhuis, who is statistically weak with Luongo, is balanced with Tanev, who is strong with Lu. These pairings not only offer the best top-to-bottom options, based on the WOWYs between the various defensemen, but they offer balance in each pairing that compensates for statistical differences in the defensemen's quality of play dependent on the goaltending. Basically, in terms of 5v5 Close GF% and CF% (again, used here as a best replacement stat for FF%, as Fenwick-for wasn't available for individual players), those pairings should give the Canucks the best predictive value for playoff success (assuming that the strong statistical evidence based on team stats can also translate into predictive value based on individual player stats).

[Gustavo Fring]

Reunite Ham-Juice for the playoffs? It's an interesting idea and there's certainly some good history there. Recently, I wrote a stats-heavy post discussing the D pairings, most notably the reasons why Edler-Bieksa and Hamhuis-Garrison made sense (and why Hamhuis-Bieksa didn't make sense).

In that case, I mostly considered GF%. GF% is the percentage of total 5v5 goals scored during while a player/pairing/line/team is on the ice that are goals-for (ie: player X is on the ice for 6 goals-for and 4 goals-against, therefore player X has a GF% of 6/(6+4) = 6/10 = 60%).

GF% is great for the regular season because it basically tells you what combinations and pairings put a team in the best position to win games. It's an excellent stat for choosing lineups.

However, we've all seen how the game changes once the playoffs roll around. Great regular season stats don't necessarily translate into great playoff performances. The funny thing is, there actually is a set of regular season stats that translate well as predictors of a team's overall playoff performance. Generally speaking, these are advanced stats differentials that are calculated from a team's numbers at 5v5 in "CLOSE" situations.

What is 5v5 CLOSE? Basically, these are the even strength minutes of games where the score is close. More specifically, the EV (5v5) minutes during the first and second periods when the score is either tied or within one goal and the third period minutes where the score is tied (this is the formula the stats geeks have chosen).

Recently, there have been a few articles that have looked at team 5v5 CLOSE stats and determined that the teams with the best GF% CLOSE and FF% (Fenwick-for percentage) CLOSE are the teams that tend to generally perform better in the playoffs and are almost always the teams that end up winning the Stanley Cup. Here's a very good article (the best one of the recent bunch) on this phenomenon:

http://hockeyanalysi...layoff-success/

Over the past 5 seasons, the teams that have put-up GF% (Close) over 55 and FF% (Close) over 53, during the regular season, have gone on to the greatest success in the playoffs. All five Cup-winners had GF% over 55 during the regular season. 3 of 5 had FF% over 53 (and the other two had FF% between 50 and 53). Only one team with a GF% over 55 (and zero teams with FF% over 53) has missed the playoffs during this period.

These numbers don't lie. If a team puts up excellent GF% and FF% during 5v5 CLOSE in the regular season, that team has a vastly superior chance of doing well in the playoffs (compared to the teams that don't put-up the numbers). Here's some good news: the Canucks are one of the few NHL teams that has excellent GF% and FF% during 5v5 CLOSE (Vancouver is currently #3 overall in GF% at 59.6 and #7 overall in FF% 53.2: http://stats.hockeya...CT&sortdir=DESC).

Given that the standard for team stats is >55 GF% (Close) and >53 FF% (Close), I figured that this would also be a good way to determine the best defensive pairings for the playoffs. Basically, I'm theorizing that the Canucks should choose a defensive lineup that pairs defensemen who put-up the best GF% and FF% when they play together.

WOWYs ("with or without you"--stats for when players play together and apart) from Stats.HockeyAnalysis.com track the GF% numbers for all NHL players but don't tract FF% (WOWY) for individual players. However, they do track CF% (Corsi-for percentage), which should be very close to FF% (please note: Corsi=shots+missed shots+blocked shots; Fenwick=shots+missed shots). With that all said, here's what would seem to be the best choices for defensive pairings, based on statistical evidence from the 2013 regular season (all stats from 5v5 Close situations):

Edler-Garrison (62.5 GF%, 55.2 CF%)

Hamhuis-Tanev (100 GF%, 53.3 CF%)

Alberts-Bieksa (66.7 GF%, 54.0 CF%)

The above players all provide stats (when paired together) that are better than the Cup-winning benchmarks of 55 GF% and 53 CF% (substituted for FF%). Here are the stats for some other possibilities (including some of the "popular" options):

Edler-Bieksa (100 GF%, 53.3 CF%)

Hamhuis-Garrison (50.0 GF%, 55.3 CF%)

Ballard-Tanev (50.0 GF%, 53.8 CF%)

Hamhuis-Bieksa (60.0 GF%, 49.6 CF%)

Edler-Tanev (25.0 GF%, 59.6 CF%)

Ballard-Garrison (33.3 GF%, 64.0 CF%)

A few things worth noting:

The Canucks' current top-four (Edler-Bieksa and Hamhuis-Garrison), while putting up the best 5v5 stats in all situations (total EV TOI), doesn't put-up the best stats during close situations. Edler-Bieksa, to the surprise of many, put-up excellent numbers across the board and in all situations but that pairing forces Garrison into a suboptimal role with Hamhuis, Alberts, or Ballard (in terms of 5v5 Close and its predictive value for playoff performance). That's why Edler-Garrison, for the playoffs, becomes preferable over Edler-Bieksa.

Hamhuis-Garrison has been very strong through the regular season, putting up good overall numbers in all situations but, in terms of 5v5 Close and playoff performance prediction, we see that Hamhuis can get a significant boost from playing alongside Tanev instead of Garrison.

The thread itself was about the Hamhuis-Bieksa (Ham-Juice) pairing and, while that pairing does put-up good GF% (60.0), their CF% (49.6) isn't all that impressive. Better off, again likely to the surprise of many, to pair Alberts-Bieksa (66.7 GF%, 54.0 CF%).

I've ranted significantly in the past about Edler-Tanev as a pairing. They're just terrible together. Their overall stats together are horrendous and there are probably no two other players on the Canucks (who have played significant minutes together) who have a worse effect on each other's performance. GF% tells the story clearest with the pairing putting up a miserable 25.0 (in 5v5 close situtations). Play them away from each other (ie: paired with anyone else) and Edler's GF% (Close) jumps to 60.9 and Tanev's to 64.3. Their CF% (Corsi) might create something of a distraction, in that a 59.6 CF% is quite high. However, this becomes far less significant when you consider that both Edler and Tanev put-up overall CF% numbers (ie: their average with all partners) that are in the mid-to-high 50's. From what we've seen over the 116:30 (and 75:20 in close situations) that these two have played together this season, Edler-Tanev probably shouldn't play alongside each other again until next season's training camp (and maybe not even then) and they certainly shouldn't be considered as a pairing for this postseason.

And finally, a pairing that I've wanted to see tried together during the regular season is Ballard-Bieksa (they have a 100 GF% in their 2013 EV TOI). However, their CF% (Close) is only 40.0 and they have only played a TOI (together) of 14:33 (of 5v5 Close). Alberts and Bieksa have played more than twice the TOI together this season and have excellent stats for both GF% and CF%. Even though I'd like to see KB4-KB3, I have to pick AA-Juice for the final playoff pairing.

For those interested, here are links to the 2013 WOWY stats (GF%, CF%, and much more) for each of the top seven Canucks' defensemen (for 5v5 CLOSE situations & with zone starts adjusted):

Edler: http://stats.hockeya...13&sit=5v5close

Bieksa: http://stats.hockeya...13&sit=5v5close

Hamhuis: http://stats.hockeya...13&sit=5v5close

Garrison: http://stats.hockeya...13&sit=5v5close

Ballard: http://stats.hockeya...13&sit=5v5close

Tanev: http://stats.hockeya...13&sit=5v5close

Alberts: http://stats.hockeya...13&sit=5v5close

For an extra bonus, here are the GF% and CF% stats (from 5v5 close situations) for each of the top seven defensemen, compared for whether they're playing in front of Cory Schneider or Roberto Luongo. I wanted to check this myself, just out of curiosity, so I figured I'd share. I'll put the results in a spoiler, as this post is already quite lengthy:

Edler: with Schneider (43.8 GF%, 56.5 CF%); with Luongo (72.7 GF%, 57.1 CF%)

Bieksa: with Schneider (66.7 GF%, 49.3 CF%); with Luongo (80.0 GF%, 53.5 CF%)

Hamhuis: with Schneider (66.7 GF%, 50.8 CF%); with Luongo (30.0 GF%, 61.8 CF%)

Garrison: with Schneider (60.0 GF%, 51.2 CF%); with Luongo (58.3 GF%, 58.9 CF%)

Ballard: with Schneider (50.0 GF%, 54.7 CF%); with Luongo (60.0 GF%, 53.9 CF%)

Tanev: with Schneider (45.5 GF%, 54.7 CF%); with Luongo (71.4 GF%, 52.9 CF%)

Alberts: with Schneider (40.0 GF%, 58.3 CF%); with Luongo (66.7 GF%, 42.2 CF%)

I've posted previously on how the Canucks players seem to play differently in front of each of the two goaltenders, with some players simply "better" with Schneider and others "better" with Luongo. Some people have a problem with this idea, believing that the goaltender shouldn't actually effect the performance of the players in front of him, and these people tend to discount numbers like those above. I tend to see them as indicators of the complex relationships (both on and off ice) between players.

On the ice, it's especially easy to see how a defenseman might play better with one goalie than with another. There is a lot of interaction between the D and the goaltender. How a defenseman pressures shooters, blocks/tips shots, screens his own net, etc. is going to effect the goalie behind him. How well a goalie copes with screens and redirects is going to effect the defensemen in front of him. Same can be said for how a D clears the crease and whether he plays the body or the stick. This can effect the goalie and how the goalie copes in these situations can effect the D. There are countless other on-ice situations where one goaltender can be found to be a better "fit" than the other goaltender, depending on the variety of factors that make up the complex playing style of each defendor playing in front of the goalies. So yes, I certainly believe that the goalie in the net can effect how well a defenseman plays the game and therefore, that there are defensemen on this team that are better with Schneider and others who are better with Luongo.

And that's not even getting into the mental side of things. Players might simply prefer one goalie over the other due to personality traits, loyalty, friendship, etc. Players might perceive one goalie to be better than the other. They might have greater confidence with one or the other. There are too many factors to list.

When we are given stats that show variances like that of Edler (GF% goes from 43.8 with Schneids to 72.7 with Lu), Hamhuis (GF% from 66.7 to 30.0), or Tanev (GF% from 45.5 to 71.4), we can clearly see that some members of this team seem to change into different players depending on who is in the net. Given this factor, there might be one more thing to consider before picking pairings for the playoffs: who is starting in net?

Most of us are assuming that Schneider gets the start, so the stats in front of CS become the important ones. Edler, Tanev, and Alberts are the weakest defensemen, in terms of GF%, when Schneider is in the net. If those numbers hold up as indicators of playoff performance, then the weaker-with-Schneids defensemen will need to be paired with guys who perform extra well in front of Schneider (like Bieksa, Hamhuis, and Garrison).

There's also some Corsi differences (Bieksa, Hamhuis, and Garrison are slightly lower in CF%, when Schneider plays), but these are less significant and most of the CF% are good, regardless of who plays goal (only Bieksa with Schneider and Alberts with Luongo yield a CF% under 50).

Interestingly, the general trend we see is better numbers with Luongo than with Schneider. Overall, the team has a GF% of 56.2 and a CF% of 53.1 when Cory Schneider plays (in 5v5 close situations). With Luongo, the overall team numbers are a GF% of 60.7 and a CF% of 56.0. Both goalies have great team numbers (Stanley Cup contendor quality) but Luongo's are better.

If the team goes with Luongo, the defenseman of concern is Dan Hamhuis, who has an awful GF% (close) of 30.0. If the stats hold up as indicators, Hamhuis will need to be "protected" by pairing him with a stronger-with-Lu player (Bieksa or Tanev would be the best choices here).

So, if we are to consider the goaltenders, in determining the best defensive pairings, the pairings I listed earlier in the post offer the best balance, for all conditions:

Edler-Garrison

Hamhuis-Tanev

Alberts-Bieksa

Garrison, Hamhuis, and Bieksa are the strongest (statistically) defensemen with Schneider in the net. They are spread through the pairings, one on each, to balance out any weaknesses. Hamhuis, who is statistically weak with Luongo, is balanced with Tanev, who is strong with Lu. These pairings not only offer the best top-to-bottom options, based on the WOWYs between the various defensemen, but they offer balance in each pairing that compensates for statistical differences in the defensemen's quality of play dependent on the goaltending. Basically, in terms of 5v5 Close GF% and CF% (again, used here as a best replacement stat for FF%, as Fenwick-for wasn't available for individual players), those pairings should give the Canucks the best predictive value for playoff success (assuming that the strong statistical evidence based on team stats can also translate into predictive value based on individual player stats).

[Gustavo Fring]

Edler needs a new d partner. His name is garrison.