rakii's line matching thread

[rakish]

The prehistory of this thread comes from me and Taro talking about the line matching during the preseason. It's hard to pay attention watching online, so I had to look at how Housley is matching lines using the NHL's html reports. Normally I would sit on this for a while, because it's difficult to tell if the numbers are right, but this is interesting for me, so I'm going to go ahead and post, and if I decide the numbers are wrong, there's worse things in life than being wrong on the internet.

 

I will cut and paste the first couple games from the New Lineup thread, which was successful, I think, in getting readers to understand what the metric is exactly when I talk about matching lines. What I'm keeping track of is the average power play time of opponent forwards, which is easier to calculate than you might think. This is the Montreal game, it's home, so Housley gets to match lines.

 

POULIOT 10.74

O'REILLY 10.17

OKPOSO 9.97

BEAULIEU8.38

SCANDELLA7.96

TENNYSON7.81

RISTOLAINEN7.31

Lehner 7.12

NOLAN6.66

JOSEFSON6.65

LARSSON6.64

KANE6.14

EICHEL5.83

POMINVILLE5.71

ANTIPIN4.98

MCCABE4.97

REINHART4.80

GRIFFITH4.38

GIRGENSONS4.23

 

So my theory is, I've talked about this a couple times now, is that Housley's strategy is to get the O'Reilly line against the other team's top line, the 4th line against the 2nd line, Eichel against the 3rd line, and Reinhart against the 4th line. These numbers above bear (bare? beer?) this out.

Game 2 was against the Islanders. It's a road game, so it's the Islanders given the opportunity to match lines. Johnson must have a penalty, but there are still two Beaulieus, my guess is they changed goaltenders, so the middle Beaulieu is still Lehner.. So for game 2, Reinhart sees the tough competition, because that's more what the Islanders want.

 

GIRGENSONS9.61

O'REILLY9.28

RISTOLAINEN9.00

GRIFFITH8.79

BEAULIEU8.51

REINHART8.37

OKPOSO8.20

LARSSON8.11

MCCABE7.95

JOHNSON7.55

Lehner 7.54

KANE7.44

EICHEL7.31

POMINVILLE7.17

ANTIPIN6.76

SCANDELLA6.75

POULIOT6.04

TENNYSON6.01

JOSEFSON5.59

MOULSON5.52

Edited October 20, 2017 by rakish

[3putt]

Here is where I have an issue understanding your theory. If a player is on the pp like Ovechkin, but is a liability defensively, we say that is tougher minutes? I think your model ignores the defensive skill sets of some of the players. ROR gets pp time Moulson gets pp time. I think ROR historically has been the tougher player to play against.

[Eleven]

I cannot imagine the effort you are putting into this.  Thank you, and I will try to follow as best as I can.

[rakish]

Game 3 was the ugly home game against New Jersey. I didn't follow the lines at all, I just watched (rakii draft pick) Jesper Bratt skate through the defense. A lot of this makes no sense here. Reinhart really played against NJ's power play players? Somehow Eichel didn't see any of them? Not confident about the information being produced.

 

REINHART9.27

BEAULIEU9.15

GRIFFITH8.90

POULIOT8.57

RISTOLAINEN8.42

GIRGENSONS8.22

LARSSON7.86

OKPOSO7.35

JOHNSON7.32

O'REILLY7.29

Lehner 6.92

TENNYSON6.75

MCCABE6.63

ANTIPIN6.29

SCANDELLA6.19

KANE6.14

JOSEFSON6.13

MOULSON6.11

POMINVILLE5.95

EICHEL5.95

Here is where I have an issue understanding your theory. If a player is on the pp like Ovechkin, but is a liability defensively, we say that is tougher minutes? I think your model ignores the defensive skill sets of some of the players. ROR gets pp time Moulson gets pp time. I think ROR historically has been the tougher player to play against.

 

Definitely.

 

It's a metric. I'm trying to guess at your first line by PP minutes.  I could guess using time on ice instead, or numbers (everyone over 90 is on the first line). I don't think it's terrible, but it aint perfect.

I cannot imagine the effort you are putting into this.  Thank you, and I will try to follow as best as I can.

 

As I said upstream, it's easier than it looks. Most of my work lately was for a Fortune 500 company, I can't use their data in my portfolio, so I need to have an ongoing site that I can show work.

Edited October 20, 2017 by rakish

[rakish]

Game 4 was the Sharks in San Jose, so we're back to the other team dictating lines.

 

Reinhart plays with Larsson and Pouliot, and they get to play against the 4th line.

 

KANE7.39

RISTOLAINEN7.33

O'REILLY7.28

OKPOSO7.19

SCANDELLA7.09

NOLAN5.90

JOSEFSON5.78

POMINVILLE5.55

EICHEL5.55

Lehner 5.53

GRIFFITH5.30

GIRGENSONS4.80

GORGES4.73

MCCABE4.66

BEAULIEU4.33

TENNYSON4.08

POULIOT3.66

REINHART3.59

LARSSON3.53

[SwampD]

I'm having a hard time understanding what this has to do with line matching. We know that teams have PP and PK units and certain guys are on them and others aren't. How does power play time relate to which line Housley sends out there 5v5 when the opposing teams 3rd line is out there?

 

EDIT: Or is this strictly about PP/PK usage?

Edited October 20, 2017 by SwampD

[Taro T]

The prehistory of this thread comes from me and Taro talking about the line matching during the preseason. It's hard to pay attention watching online, so I had to look at how Housley is matching lines using the NHL's html reports. Normally I would sit on this for a while, because it's difficult to tell if the numbers are right, but this is interesting for me, so I'm going to go ahead and post, and if I decide the numbers are wrong, there's worse things in life than being wrong on the internet.

 

I will cut and paste the first couple games from the New Lineup thread, which was successful, I think, in getting readers to understand what the metric is exactly when I talk about matching lines. What I'm keeping track of is the average power play time of opponent forwards, which is easier to calculate than you might think. This is the Montreal game, it's home, so Housley gets to match lines. Beaulieu is there twice, because Lehner's name hasn't been use in the reports, so I would need to manually fix that, until Lehner gets an assist, or a penalty, or a shot. It's more likely that Lehner's number is more toward the middle (since he is on the ice for all Canadien players), and the real Beaulieu is near the top (which I don't understand either)

 

POULIOT 10.74

O'REILLY 10.17

OKPOSO 9.97

BEAULIEU8.38

SCANDELLA7.96

TENNYSON7.81

RISTOLAINEN7.31

BEAULIEU7.12

NOLAN6.66

JOSEFSON6.65

LARSSON6.64

KANE6.14

EICHEL5.83

POMINVILLE5.71

ANTIPIN4.98

MCCABE4.97

REINHART4.80

GRIFFITH4.38

GIRGENSONS4.23

 

So my theory is, I've talked about this a couple times now, is that Housley's strategy is to get the O'Reilly line against the other team's top line, the 4th line against the 2nd line, Eichel against the 3rd line, and Reinhart against the 4th line. These numbers above bear (bare? beer?) this out.

Game 2 was against the Islanders. It's a road game, so it's the Islanders given the opportunity to match lines. Johnson must have a penalty, but there are still two Beaulieus, my guess is they changed goaltenders, so the middle Beaulieu is still Lehner.. So for game 2, Reinhart sees the tough competition, because that's more what the Islanders want.

 

GIRGENSONS9.61

O'REILLY9.28

RISTOLAINEN9.00

GRIFFITH8.79

BEAULIEU8.51

REINHART8.37

OKPOSO8.20

LARSSON8.11

MCCABE7.95

JOHNSON7.55

BEAULIEU7.54

KANE7.44

EICHEL7.31

POMINVILLE7.17

ANTIPIN6.76

SCANDELLA6.75

POULIOT6.04

TENNYSON6.01

JOSEFSON5.59

MOULSON5.52

The data you had for game 1 definitely matched the eyeball test.

 

W/ Larsson centering the 4th line tomorrow, highly expect to see similar usage against the 'Nucks.

 

Not to create additional work for you, but would also be interested in seeing how matchups relative to the other team's PK played out. Particularly for road games would expect Eichel's #'s to face the top PKers, followed by O'Reilly's, Reinhart's, then Larsson's. But some opponents, like the Pens that haven't been faced yet, would probably put Crosby's line out against Eichel.

I'm having a hard time understanding what this has to do with line matching. We know that teams have PP and PK units and certain guys are on them and others aren't. How does power play time relate to which line Housley sends out there 5v5 when the opposing teams 3rd line is out there?

Not to speak for Rakil, but he seems to be using opponent's PP TOI as a proxy for the offensive quality of those players. Wherein the expectation that the guys getting PP time are the opponents best players & those that get the most time are their top 5v5 guys, the next most are 2nd liners, next the 3rd line, & the weakest guys get the least PP time.

[rakish]

My strategy, Swamp, is this. I'm defining the opponents offensive capability by measuring how much PP time they get. It will get some anomalies, Moulson gets more PP time than his 'line he plays on' suggests, but I think overall, it captures what I trying to show.

 

What I'm trying to show is my theory that what is considered the 4th line (Josephson and Nolan) are playing better offensive players than Reinhart's line. I have found in the past some coaches try to do this to give players (Reinhart here, Lupal in Toronto when I looked at that a few years back, O'Reilly in Colorado a few years back). I think it's a strategy that can work if Reinhart's line is able to put the puck in the net.

Game 5 was the loss to LA.

 

KANE9.28

MOULSON8.45

EICHEL8.33

POMINVILLE8.18

O'REILLY7.81

GIRGENSONS7.61

SCANDELLA7.56

GORGES6.97

RISTOLAINEN6.86

Lehner 6.60

TENNYSON6.17

MCCABE6.01

BEAULIEU5.46

REINHART5.40

LARSSON4.29

GRIFFITH3.89

NOLAN2.45

POULIOT2.33

JOSEFSON1.96

[dudacek]

My strategy, Swamp, is this. I'm defining the opponents offensive capability by measuring how much PP time they get. It will get some anomalies, Moulson gets more PP time than his 'line he plays on' suggests, but I think overall, it captures what I trying to show.

 

What I'm trying to show is my theory that what is considered the 4th line (Josephson and Nolan) are playing better offensive players than Reinhart's line. I have found in the past some coaches try to do this to give players (Reinhart here, Lupal in Toronto when I looked at that a few years back, O'Reilly in Colorado a few years back). I think it's a strategy that can work if Reinhart's line is able to put the puck in the net.

This would make a heckuvalot of sense if Reinhart was given a winger (#9) who can score.

It's why I've been advocating three balanced lines. If ROR is neutralizing the top guys, then either Eichel or Reinhart should be getting a mismatch.

 

Right now, Sam is stuck with two guys who can't finish, and the redundancies of Kane/Eichel have been thoroughly discussed.

[Randall Flagg]

I'm following what Swamp is saying a bit - Power plays often regularly have guys from 3 different lines on them as well as all D-pairings, probably that one more often than forwards. So I think it would be tough to draw which lines Phil uses guys against and which lines get sent out against our guys on the road from that.

[rakish]

It's not only about the 3 balanced lines Dudacek, it's about having a 4th line that can play offensive players. Josephson is fast enough to play their second line, which give Reinhart the opportunity to play against their 4th line.

 

Game 6 against Anaheim

 

EICHEL8.13

SCANDELLA8.06

POMINVILLE7.90

KANE7.84

BEAULIEU7.82

RISTOLAINEN7.78

REINHART7.76

O'REILLY7.68

GIRGENSONS7.13

TENNYSON6.90

JOHNSON6.73

BAILEY5.60

POULIOT5.14

LARSSON5.01

JOSEFSON4.78

NOLAN4.63

GORGES4.35

MCCABE4.34

MOULSON4.23

I didn't upload Vegas yet, and I'm almost done for the night, but I think you guys know where I'm going with this. I wanted to get his out before the next home game. I recognize the flaws in the thought process, not sure it's a solvable issue.

 

I removed Taro's opposite list, I can see how it would be confusing, if you really care, I think Taro copied it in his content below

Edited October 20, 2017 by rakish

[LGR4GM]

Hmm interesting.....

[Taro T]

It's not only about the 3 balanced lines Dudacek, it's about having a 4th line that can play offensive players. Josephson is fast enough to play their second line, which give Reinhart the opportunity to play against their 4th line.

 

Game 6 against Anaheim

 

EICHEL8.13

SCANDELLA8.06

POMINVILLE7.90

KANE7.84

BEAULIEU7.82

RISTOLAINEN7.78

REINHART7.76

O'REILLY7.68

GIRGENSONS7.13

TENNYSON6.90

JOHNSON6.73

BAILEY5.60

POULIOT5.14

LARSSON5.01

JOSEFSON4.78

NOLAN4.63

GORGES4.35

MCCABE4.34

MOULSON4.23

I didn't upload Vegas yet, and I'm almost done for the night, but I think you guys know where I'm going with this. I wanted to get his out before the next home game. I recognize the flaws in the thought process, not sure it's a solvable issue.

 

Taro wanted to see the opposite, which is seeing how teams played their defense against Buffalo. So this is how much time their defensemen get on the PK, which of course, will have the same flaws as the reverse.

 

This is the kings game:

 

KANE7.20

POMINVILLE6.68

EICHEL6.68

SCANDELLA6.57

MOULSON6.38

RISTOLAINEN6.36

O'REILLY6.28

TENNYSON6.25

Lehner 6.23

BEAULIEU6.09

GIRGENSONS6.04

GORGES6.00

REINHART5.97

NOLAN5.82

MCCABE5.76

LARSSON5.53

POULIOT5.40

JOSEFSON5.37

GRIFFITH5.37

Danke. :beer:

[rakish]

Vegas game, which was on the road.

 

POULIOT9.92

BAILEY9.80

LARSSON8.51

EICHEL8.46

KANE8.41

MCCABE8.40

GORGES8.36

REINHART8.31

POMINVILLE8.30

RISTOLAINEN8.25

O'REILLY8.20

SCANDELLA8.12

BEAULIEU8.11

JOHNSON7.94

OKPOSO7.77

TENNYSON7.57

MOULSON5.87

NOLAN5.58

GIRGENSONS0.89

[SwampD]

My strategy, Swamp, is this. I'm defining the opponents offensive capability by measuring how much PP time they get. It will get some anomalies, Moulson gets more PP time than his 'line he plays on' suggests, but I think overall, it captures what I trying to show.

 

What I'm trying to show is my theory that what is considered the 4th line (Josephson and Nolan) are playing better offensive players than Reinhart's line. I have found in the past some coaches try to do this to give players (Reinhart here, Lupal in Toronto when I looked at that a few years back, O'Reilly in Colorado a few years back). I think it's a strategy that can work if Reinhart's line is able to put the puck in the net.

Got it. I'm just having a hard time drawing any conclusions about it as of yet, because it could mean so many things.

 

It could show that Housley thinks the 4th line is better defensively and he has that shutdown "just don't let them score" line.

It could also show that (and I think this is the case because most of our games have been away and had the last change) the other team got the matchups they wanted and put out a better defensive line against Sam and and a better offensive line against our 4th.

 

It is interesting, though.

[woods-racer]

I'm so frickin lost.

 

In the simplest of terms.... what your showing is a Sabre player with a higher number played against the opposing teams better players, right?

[rakish]

right, so pay attention to which team is home, because when home, Housley gets to match lines, on the road, the opponent.

[woods-racer]

It also shows that the number *flips* when on the road.  Showing who the opposing team believes our weakest players are, is how I'm reading it.

 

It's a really cool way of matching up to the eyeball test.

 

Thanks Rakish!

Edited October 20, 2017 by Woods-Racer

[LGR4GM]

Rakish what if you added PP time together with total 5v5 ice time? Obviously defenders will get lifted up but that isn't necessarily a bad thing. 

[Samson's Flow]

I'm so frickin lost.

 

In the simplest of terms.... what your showing is a Sabre player with a higher number played against the opposing teams better players, right?

What the numbers show is a proxy for quality of competition. The higher numbers (Vegas ex. Pouliot/Larsson/Bailey) mean those players played against the opponent's best players (the ones that get the most PP TOI) while the lower numbers mean those Sabres players (Vegas ex. Okposo/Moulson) played against weaker opposition.

 

The thought is this should be able to identify what players Housley trusts to "shut down" the opponents top lines, and who he would like to use to take advantage and score against lesser opponents.

 

Again Home games should be more representative since we get last change in those games, Away games are more of a mixed bag/who the opponent wants to play their top guys against.

[rakish]

Rakish what if you added PP time together with total 5v5 ice time? Obviously defenders will get lifted up but that isn't necessarily a bad thing. 

 

 

Could certainly, my first thought is that it wouldn't be better. This one only uses forwards, so my example below relates to the reverse chart that Taro wanted to see of defensemen.

 

There are some defensemen who play a lot (Shattenkirk is an example) but at the same time they try to keep him from the tough competition. A lot of people consider Shattenkirk a top line defenseman because of minutes played, but I think him as a second or third pair defenseman because he mostly plays against lower lines. I'm sure there are forwards who play a lot, but only because their defense is good. I don't know off hand.

 

So in separating forwards, it might help to add 5-5 minutes, but it might make them more homogeneous. It's hard now deciding if the numbers are right (

The NJ game looks all wrong,

the Girgensons Vegas game he gets almost 0, how'd that happen?

), adding even a little more complexity would make debugging even more difficult.

[Samson's Flow]

Could certainly, my first thought is that it wouldn't be better. This one only uses forwards, so my example below relates to the reverse chart that Taro wanted to see of defensemen.

 

There are some defensemen who play a lot (Shattenkirk is an example) but at the same time they try to keep him from the tough competition. A lot of people consider Shattenkirk a top line defenseman because of minutes played, but I think him as a second or third pair defenseman because he mostly plays against lower lines. I'm sure there are forwards who play a lot, but only because their defense is good. I don't know off hand.

 

So in separating forwards, it might help to add 5-5 minutes, but it might make them more homogeneous. It's hard now deciding if the numbers are right (

The NJ game looks all wrong,

the Girgensons Vegas game he gets almost 0, how'd that happen?

), adding even a little more complexity would make debugging even more difficult.

I attributed the Girgensons result in Vegas as a result of the injury. There weren't enough meaningful minutes played to get a result.

[LGR4GM]

Could certainly, my first thought is that it wouldn't be better. This one only uses forwards, so my example below relates to the reverse chart that Taro wanted to see of defensemen.

 

There are some defensemen who play a lot (Shattenkirk is an example) but at the same time they try to keep him from the tough competition. A lot of people consider Shattenkirk a top line defenseman because of minutes played, but I think him as a second or third pair defenseman because he mostly plays against lower lines. I'm sure there are forwards who play a lot, but only because their defense is good. I don't know off hand.

 

So in separating forwards, it might help to add 5-5 minutes, but it might make them more homogeneous. It's hard now deciding if the numbers are right (

The NJ game looks all wrong,

the Girgensons Vegas game he gets almost 0, how'd that happen?

), adding even a little more complexity would make debugging even more difficult.

hmmm, I'll think on it. 

[Randall Flagg]

I think it would be worthwhile for someone to write down Vancouver's line #s and keep track of who goes on the ice following a Vancouver change, and compare it with what rakish gets.

 

It's not gonna be me.

[woods-racer]

What the numbers show is a proxy for quality of competition. The higher numbers (Vegas ex. Pouliot/Larsson/Bailey) mean those players played against the opponent's best players (the ones that get the most PP TOI) while the lower numbers mean those Sabres players (Vegas ex. Okposo/Moulson) played against weaker opposition.

 

The thought is this should be able to identify what players Housley trusts to "shut down" the opponents top lines, and who he would like to use to take advantage and score against lesser opponents.

 

Again Home games should be more representative since we get last change in those games, Away games are more of a mixed bag/who the opponent wants to play their top guys against.

 

In a fleeting glimpse snap shot kind of way I get...

 

Vegas looked to put their best players against Pouliot, Larsson, Bailey as they found that to be to their advantage.

 

They found no threat in Okposo/Moulson and iced players accordingly.

 

I'm not adding Housley to my summation because there is very little he to do about it being it was an away game. I just find it fascination in an away game we can get a glimpse of how opposing coaches view our roster by the players they ice. To me that is more critical than how Housley will ice his team tonight.