Jump to content
The Official Site of the Vancouver Canucks
Canucks Community

Building Optimal 21-22 Roster through Analytics

Rate this topic


Recommended Posts

Triple P line! Cool analysis. Still, something doesn’t feel right about Pearson on the top line and Boeser on a third line. But the rationale is ok. 
Maybe TG will read this thread and give it a try.

Link to comment
Share on other sites

You know what, I have no idea what most of the means. But I really appreciate the effort you put into this. Personally the look of that ideal lineup just doesn't do anything for me.

 

I do have time for Horvat with Boeser though. Hog on with them, but I'd love to see Pearson with them at some point. Stick Garland with Miller and Petey, I could see that being a great way to spread out the scoring. Pod and Dickson on the 3rd line with either Pearson and Hoglander. 

 

  • Upvote 1
Link to comment
Share on other sites

40 minutes ago, nuckfanfromafrica said:

 

Hockey is a strong link game - having your best players throughout your lineup increases the chances of winning. The default tendency is to load up a top line. In most cases, absolutely entertaining to watch with short term results. Unfortunately detrimental to prolonged team success.

 

 Statistically, a team's success rate is increased exponentially by spreading out talent across the line up. Playing team's best players across the line up  elevates less-elite line mates.

 

The Key however is to compose lines where playing styles mesh. Playing style combination yeilding the highest xG% and the lowest drop off in chain link from forward line 1 to 4 (variance),

 

Example:

 

Team Blue features

  • line A xG% 55.4%
  • line B xG% 55.0% 
  • Line C xG% 52.7%
  • Line D xG% 48.9%

               

Team Yellow features

  • line A xG% 58.9%
  • line B xG% 51.5%
  • Line C xG% 48.9%
  • Line D xG% 48.9%

 

Even though Team Yellow's line A has the highest xG% on the top line, it has the lower probability of long term success, because of a bigger drop off (variance) from line to line.

The most successful teams feature a high xG% on the top line and lowest variance from top line to bottom.

 

  • 20-21 Edmonton Oilers (Team yellow): Connor McDavid - Leon Draisaitl
  • 20-21 Tampa (Team Blue): Point - Kucherov, Stamkos-Cirelli, etc.

 

The criteria used for this composition are as follows:

 

1. Playing Style (based on 5v5 play):

         

 Metric:

          a. Playmakers:  CF% => 54.1%    and      xGF60 => 2.64

          b. Shooter:     52.0% <= CF% < 54.1%     and      2.51 <= xGF60 < 2.64

          C. Balanced:   50.2% <= CF% < 52.0%    and     2.35  <=  xGF60  <  2.51

          d. Dependent: CF% <=  45.6%   and    xGF60 <  2.02  

 

Current Roster forward composition CF% and xGF60 (stat average over the last 3 seasons):

 

  • Tanner Pearson (48.11%, 2.62) - falls within two playing styles depending on his position on ice (primarily a dependent & secondary a shooter). He tends to be dependent when lined up on the Left wing. Playing off wing as a RW he utilizes his shoot more. Video eye test below:

 

 

  • Elias Pettersson (53.7%, 2.74) - fits as both a playmaker (when playing as a center) and a shooter playing the wing.

 

  • Nils Hoglander (50.71%, 2.50) - limited sample size. fits into the dimension of balanced

 

  • Bo Horvat (49.28%, 2.73) - fits primarily as a playmaker then secondary as a dependent.

 

  • Boeser (51.2%, 2.54) - fits primarily as a shooter then secondary as balanced

 

  • J.T. Miller (50.90%, 2.65) - fits primarily as a playmaker then secondarily as balanced

 

  • Jason Dickinson (51.59%, 2.14) - fits primarily as balanced then secondary as dependent

 

  • Conor Garland (59.06%, 2.55) - fits primarily as a shooter when playing RW and playmaker on the LW. Eye test video below:

 

 

  • Brandon Sutter (44.42%, 2.10) - fits as a dependent

 

  • Tyler Motte (42.51%, 2.14) - dependent

 

  • Highmore (40.32%, 2.12) - dependent

 

  • Zack Macewen (45.15%, 2.01%)

 

  • Jonah Gadjovich (limited sample) - projects as a dependent

 

  • Podkolzin (no sample) - projects primarily as a playmaker and secondarily as balanced.

 

 

2. Playing Style composition xG%:

 

Probability that a shot will result in a goal based on the characteristics 5v5 expected goal. The line composition below are not based on LW-C-RW composition but more chemistry fit for each playing style.

 

Metric

 

1. Playmaker - Playmaker - Playmaker = 58.9%

2. playmaker - Balance - Playmaker = 57.2%

3. playmaker - Playmaker - Shooter = 55.4%

4. playmaker - shooter - shooter = 55.1%

5. Balanced - playmaker - shooter = 55.0%

6. balanced - shooter - shooter = 54.7%

7. playmaker - dependent - playmaker = 54.5%

8. balanced - playmaker - balanced = 52.2%

9. dependent - shooter - shooter = 52.1%

10. balanced - balanced - balanced = 51.5%

11. balanced - balanced - shooter = 50.9%

12. dependent - playmaker - shooter = 50.3%

13. shooter - shooter - shooter = 50%

14. balanced - dependent - playmaker = 48.9%

15. balanced - dependent - shooter = 48.5%

16. Dependent - playmaker - dependent = 45.8%

17. balanced - balanced - dependent = 45.4%

18. depensent - shooter - dependent = 44.7%

19. balanced - dependent - dependent = 44.6%

20. dependent - dependent - dependent = 41.8%

 

 

3. Forward Roster composition based on the 2 above metrics:

 

Optimal option:

 

Line a:          Pearson (LW) - Pettersson -   Podkolzin (RW)

                 7. Dependent       -  playmaker   - playmaker = 54.5%

 

Line b:      Miller (LW)  -  Dickinson - Garland (RW)

                5. playmaker - Balanced - shooter = 55.0%

 

Line c:      Hoglander (LW) - Horvat - Boeser (RW)

                5. balanced  - playmaker - shooter  = 55.0%

 

Line d:    Gadjovich (LW) - Sutter - Motte (RW)

               20. dependent - dependent - dependent = 41.8%

 

 

Extra forward on line d:  Highmore & Macewen.

 

Analysis: not much variance from line a - c. The avergae score for the top 3 lines is 54.83%. Weak link is line d. composition given that it comprises all forwards that are dependent. This line construction also adds the advantage of having a defensive presence on each line along with some grit whether on the wing on in the middle.

 

Popular option:

 

Line a:       Miller (LW) - Pettersson - Boeser (RW)

                 3.  playmaker - playmaker - shooter = 55.4%

 

Line b:       Hoglander(LW) - Horvat - Garland(RW)

                 5.    balanced -  playmaker - shooter  = 55.0%

 

Line c:      Pearson (LW) -  Dickinson - Podkolzin (RW)

                 14.  dependent -  balanced - playmaker  = 48.9% 

 

Line d:    Gadjovich (LW) - Sutter - Motte (RW)

               20. dependent - dependent - dependent = 41.8%

 

Analysis: feature 1 line with the highest % but the drop off from Line b to line c is significant. Average of the top 3 lines is 53.1%.

 

 

 

 

 

I would switch Garland and Boeser around since a line of Hoglander and Garland playing together would be quite small.  Great analysis though!

Link to comment
Share on other sites

37 minutes ago, nuckfanfromafrica said:

 

 

Tanner Pearson (48.11%, 2.62) - falls within two playing styles depending on his position on ice (primarily a dependent & secondary a shooter). He tends to be dependent when lined up on the Left wing. Playing off wing as a RW he utilizes his shoot more. Video eye test below:

 

37 minutes ago, nuckfanfromafrica said:

Conor Garland (59.06%, 2.55) - fits primarily as a shooter when playing RW and playmaker on the LW. Eye test video below:

 

 

 

Sooooooo

 

Garland Bo Pearson = 55.4% ????

Link to comment
Share on other sites

3 hours ago, nuckfanfromafrica said:

 

Hockey is a strong link game - having your best players throughout your lineup increases the chances of winning. The default tendency is to load up a top line. In most cases, absolutely entertaining to watch with short term results. Unfortunately detrimental to prolonged team success.

 

 Statistically, a team's success rate is increased exponentially by spreading out talent across the line up. Playing team's best players across the line up  elevates less-elite line mates.

 

The Key however is to compose lines where playing styles mesh. Playing style combination yeilding the highest xG% and the lowest drop off in chain link from forward line 1 to 4 (variance),

 

Example:

 

Team Blue features

  • line A xG% 55.4%
  • line B xG% 55.0% 
  • Line C xG% 52.7%
  • Line D xG% 48.9%

               

Team Yellow features

  • line A xG% 58.9%
  • line B xG% 51.5%
  • Line C xG% 48.9%
  • Line D xG% 48.9%

 

Even though Team Yellow's line A has the highest xG% on the top line, it has the lower probability of long term success, because of a bigger drop off (variance) from line to line.

The most successful teams feature a high xG% on the top line and lowest variance from top line to bottom.

 

  • 20-21 Edmonton Oilers (Team yellow): Connor McDavid - Leon Draisaitl
  • 20-21 Tampa (Team Blue): Point - Kucherov, Stamkos-Cirelli, etc.

 

The criteria used for this composition are as follows:

 

1. Playing Style (based on 5v5 play):

         

 Metric:

          a. Playmakers:  CF% => 54.1%    and      xGF60 => 2.64

          b. Shooter:     52.0% <= CF% < 54.1%     and      2.51 <= xGF60 < 2.64

          C. Balanced:   50.2% <= CF% < 52.0%    and     2.35  <=  xGF60  <  2.51

          d. Dependent: CF% <=  45.6%   and    xGF60 <  2.02  

 

Current Roster forward composition CF% and xGF60 (stat average over the last 3 seasons):

 

  • Tanner Pearson (48.11%, 2.62) - falls within two playing styles depending on his position on ice (primarily a dependent & secondary a shooter). He tends to be dependent when lined up on the Left wing. Playing off wing as a RW he utilizes his shoot more. Video eye test below:

 

 

  • Elias Pettersson (53.7%, 2.74) - fits as both a playmaker (when playing as a center) and a shooter playing the wing.

 

  • Nils Hoglander (50.71%, 2.50) - limited sample size. fits into the dimension of balanced

 

  • Bo Horvat (49.28%, 2.73) - fits primarily as a playmaker then secondary as a dependent.

 

  • Boeser (51.2%, 2.54) - fits primarily as a shooter then secondary as balanced

 

  • J.T. Miller (50.90%, 2.65) - fits primarily as a playmaker then secondarily as balanced

 

  • Jason Dickinson (51.59%, 2.14) - fits primarily as balanced then secondary as dependent

 

  • Conor Garland (59.06%, 2.55) - fits primarily as a shooter when playing RW and playmaker on the LW. Eye test video below:

 

 

  • Brandon Sutter (44.42%, 2.10) - fits as a dependent

 

  • Tyler Motte (42.51%, 2.14) - dependent

 

  • Highmore (40.32%, 2.12) - dependent

 

  • Zack Macewen (45.15%, 2.01%)

 

  • Jonah Gadjovich (limited sample) - projects as a dependent

 

  • Podkolzin (no sample) - projects primarily as a playmaker and secondarily as balanced.

 

 

2. Playing Style composition xG%:

 

Probability that a shot will result in a goal based on the characteristics 5v5 expected goal. The line composition below are not based on LW-C-RW composition but more chemistry fit for each playing style.

 

Metric

 

1. Playmaker - Playmaker - Playmaker = 58.9%

2. playmaker - Balance - Playmaker = 57.2%

3. playmaker - Playmaker - Shooter = 55.4%

4. playmaker - shooter - shooter = 55.1%

5. Balanced - playmaker - shooter = 55.0%

6. balanced - shooter - shooter = 54.7%

7. playmaker - dependent - playmaker = 54.5%

8. balanced - playmaker - balanced = 52.2%

9. dependent - shooter - shooter = 52.1%

10. balanced - balanced - balanced = 51.5%

11. balanced - balanced - shooter = 50.9%

12. dependent - playmaker - shooter = 50.3%

13. shooter - shooter - shooter = 50%

14. balanced - dependent - playmaker = 48.9%

15. balanced - dependent - shooter = 48.5%

16. Dependent - playmaker - dependent = 45.8%

17. balanced - balanced - dependent = 45.4%

18. depensent - shooter - dependent = 44.7%

19. balanced - dependent - dependent = 44.6%

20. dependent - dependent - dependent = 41.8%

 

 

3. Forward Roster composition based on the 2 above metrics:

 

Optimal option:

 

Line a:          Pearson (LW) - Pettersson -   Podkolzin (RW)

                 7. Dependent       -  playmaker   - playmaker = 54.5%

 

Line b:      Miller (LW)  -  Dickinson - Garland (RW)

                5. playmaker - Balanced - shooter = 55.0%

 

Line c:      Hoglander (LW) - Horvat - Boeser (RW)

                5. balanced  - playmaker - shooter  = 55.0%

 

Line d:    Gadjovich (LW) - Sutter - Motte (RW)

               20. dependent - dependent - dependent = 41.8%

 

 

Extra forward on line d:  Highmore & Macewen.

 

Analysis: not much variance from line a - c. The avergae score for the top 3 lines is 54.83%. Weak link is line d. composition given that it comprises all forwards that are dependent. This line construction also adds the advantage of having a defensive presence on each line along with some grit whether on the wing on in the middle.

 

Popular option:

 

Line a:       Miller (LW) - Pettersson - Boeser (RW)

                 3.  playmaker - playmaker - shooter = 55.4%

 

Line b:       Hoglander(LW) - Horvat - Garland(RW)

                 5.    balanced -  playmaker - shooter  = 55.0%

 

Line c:      Pearson (LW) -  Dickinson - Podkolzin (RW)

                 14.  dependent -  balanced - playmaker  = 48.9% 

 

Line d:    Gadjovich (LW) - Sutter - Motte (RW)

               20. dependent - dependent - dependent = 41.8%

 

Analysis: feature 1 line with the highest % but the drop off from Line b to line c is significant. Average of the top 3 lines is 53.1%.

 

 

 

 

 

Very good, it fits nicely with my idea that Petey shouldn’t play with Miller.

Green needs to spread the goodies so the other coach will scratch his head what to do.

  • Vintage 1
Link to comment
Share on other sites

13 hours ago, nuckfanfromafrica said:

 

Hockey is a strong link game - having your best players throughout your lineup increases the chances of winning. The default tendency is to load up a top line. In most cases, absolutely entertaining to watch with short term results. Unfortunately detrimental to prolonged team success.

 

 Statistically, a team's success rate is increased exponentially by spreading out talent across the line up. Playing team's best players across the line up  elevates less-elite line mates.

 

The Key however is to compose lines where playing styles mesh. Playing style combination yeilding the highest xG% and the lowest drop off in chain link from forward line 1 to 4 (variance),

 

Example:

 

Team Blue features

  • line A xG% 55.4%
  • line B xG% 55.0% 
  • Line C xG% 52.7%
  • Line D xG% 48.9%

               

Team Yellow features

  • line A xG% 58.9%
  • line B xG% 51.5%
  • Line C xG% 48.9%
  • Line D xG% 48.9%

 

Even though Team Yellow's line A has the highest xG% on the top line, it has the lower probability of long term success, because of a bigger drop off (variance) from line to line.

The most successful teams feature a high xG% on the top line and lowest variance from top line to bottom.

 

  • 20-21 Edmonton Oilers (Team yellow): Connor McDavid - Leon Draisaitl
  • 20-21 Tampa (Team Blue): Point - Kucherov, Stamkos-Cirelli, etc.

 

The criteria used for this composition are as follows:

 

1. Playing Style (based on 5v5 play):

         

 Metric:

          a. Playmakers:  CF% => 54.1%    and      xGF60 => 2.64

          b. Shooter:     52.0% <= CF% < 54.1%     and      2.51 <= xGF60 < 2.64

          C. Balanced:   50.2% <= CF% < 52.0%    and     2.35  <=  xGF60  <  2.51

          d. Dependent: CF% <=  45.6%   and    xGF60 <  2.02  

 

Current Roster forward composition CF% and xGF60 (stat average over the last 3 seasons):

 

  • Tanner Pearson (48.11%, 2.62) - falls within two playing styles depending on his position on ice (primarily a dependent & secondary a shooter). He tends to be dependent when lined up on the Left wing. Playing off wing as a RW he utilizes his shoot more. Video eye test below:

 

 

  • Elias Pettersson (53.7%, 2.74) - fits as both a playmaker (when playing as a center) and a shooter playing the wing.

 

  • Nils Hoglander (50.71%, 2.50) - limited sample size. fits into the dimension of balanced

 

  • Bo Horvat (49.28%, 2.73) - fits primarily as a playmaker then secondary as a dependent.

 

  • Boeser (51.2%, 2.54) - fits primarily as a shooter then secondary as balanced

 

  • J.T. Miller (50.90%, 2.65) - fits primarily as a playmaker then secondarily as balanced

 

  • Jason Dickinson (51.59%, 2.14) - fits primarily as balanced then secondary as dependent

 

  • Conor Garland (59.06%, 2.55) - fits primarily as a shooter when playing RW and playmaker on the LW. Eye test video below:

 

 

  • Brandon Sutter (44.42%, 2.10) - fits as a dependent

 

  • Tyler Motte (42.51%, 2.14) - dependent

 

  • Highmore (40.32%, 2.12) - dependent

 

  • Zack Macewen (45.15%, 2.01%)

 

  • Jonah Gadjovich (limited sample) - projects as a dependent

 

  • Podkolzin (no sample) - projects primarily as a playmaker and secondarily as balanced.

 

 

2. Playing Style composition xG%:

 

Probability that a shot will result in a goal based on the characteristics 5v5 expected goal. The line composition below are not based on LW-C-RW composition but more chemistry fit for each playing style.

 

Metric

 

1. Playmaker - Playmaker - Playmaker = 58.9%

2. playmaker - Balance - Playmaker = 57.2%

3. playmaker - Playmaker - Shooter = 55.4%

4. playmaker - shooter - shooter = 55.1%

5. Balanced - playmaker - shooter = 55.0%

6. balanced - shooter - shooter = 54.7%

7. playmaker - dependent - playmaker = 54.5%

8. balanced - playmaker - balanced = 52.2%

9. dependent - shooter - shooter = 52.1%

10. balanced - balanced - balanced = 51.5%

11. balanced - balanced - shooter = 50.9%

12. dependent - playmaker - shooter = 50.3%

13. shooter - shooter - shooter = 50%

14. balanced - dependent - playmaker = 48.9%

15. balanced - dependent - shooter = 48.5%

16. Dependent - playmaker - dependent = 45.8%

17. balanced - balanced - dependent = 45.4%

18. depensent - shooter - dependent = 44.7%

19. balanced - dependent - dependent = 44.6%

20. dependent - dependent - dependent = 41.8%

 

 

3. Forward Roster composition based on the 2 above metrics:

 

Optimal option:

 

Line a:          Pearson (LW) - Pettersson -   Podkolzin (RW)

                 7. Dependent       -  playmaker   - playmaker = 54.5%

 

Line b:      Miller (LW)  -  Dickinson - Garland (RW)

                5. playmaker - Balanced - shooter = 55.0%

 

Line c:      Hoglander (LW) - Horvat - Boeser (RW)

                5. balanced  - playmaker - shooter  = 55.0%

 

Line d:    Gadjovich (LW) - Sutter - Motte (RW)

               20. dependent - dependent - dependent = 41.8%

 

 

Extra forward on line d:  Highmore & Macewen.

 

Analysis: not much variance from line a - c. The avergae score for the top 3 lines is 54.83%. Weak link is line d. composition given that it comprises all forwards that are dependent. This line construction also adds the advantage of having a defensive presence on each line along with some grit whether on the wing on in the middle.

 

Popular option:

 

Line a:       Miller (LW) - Pettersson - Boeser (RW)

                 3.  playmaker - playmaker - shooter = 55.4%

 

Line b:       Hoglander(LW) - Horvat - Garland(RW)

                 5.    balanced -  playmaker - shooter  = 55.0%

 

Line c:      Pearson (LW) -  Dickinson - Podkolzin (RW)

                 14.  dependent -  balanced - playmaker  = 48.9% 

 

Line d:    Gadjovich (LW) - Sutter - Motte (RW)

               20. dependent - dependent - dependent = 41.8%

 

Analysis: feature 1 line with the highest % but the drop off from Line b to line c is significant. Average of the top 3 lines is 53.1%.

 

 

 

 

 

Nicely done!

Link to comment
Share on other sites

Hey @nuckfanfromafrica

 

Where does this come from:

 

     a. Playmakers:  CF% => 54.1%    and      xGF60 => 2.64

          b. Shooter:     52.0% <= CF% < 54.1%     and      2.51 <= xGF60 < 2.64

          C. Balanced:   50.2% <= CF% < 52.0%    and     2.35  <=  xGF60  <  2.51

          d. Dependent: CF% <=  45.6%   and    xGF60 <  2.02  

 

 

Link to comment
Share on other sites

1 hour ago, grumpyone said:

whatever you do DON'T let the oilers or leafs see this....they might use it to figure out why they suck.

their problem is algebra, not stats. They thought McDavid/Matthews = Cup and forgot all the other variables. 

  • Hydration 1
  • Vintage 1
Link to comment
Share on other sites

Does this take into account how heavily players rely on their linemates to generate these stats? 

Guys like Pettersson is able to drive his own line more often than somebody like a Boeser for example. 

Link to comment
Share on other sites

Just watched Moneyball again last night.  I like the statistical breakdown, but it also illustrated just how dependent it was on playing the exact players in the exact situation, and even then, you couldn't take forgranted numerous other circumstances.  I like where you're going with this, I love the work, and love the explanation.  And if presented with that, would be curious to see what that lineup would do over a 20 game sample size.  It also helps to identify how weak the 4th line is compared to the top 9, and illustrates what you'd need to do to fix it, and what type of players you're looking at.  

  • Hydration 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
 Share

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...