When Sinatra sings against Nelson Riddle’s strings

Georgexs:
LT: (from Friday)

“Is there a long list of successful offensive players who are poor five-on-five performers?”

A long list? I’d guess no. Over their careers, successful offensive players have to be good 5v5 performers. 5v5 is too big a part of the game, too big a portion of the minutes that a forward plays. If he can’t consistently generate offense at 5v5, then he’s going to have a hard time accumulating enough total offense to rank among the “successful” performers. At least in the long term.

In the short term? The 75th percentile for 5v5 P60 for forwards in the past 12 seasons has been in the 1.7 to 1.9 range. Roughly 140 forwards score above that rate each season. Do successful offensive players occasionally fall in that range, i.e., produce 5v5 offense that lags behind over 100 other forwards in a given season? Yes. Hall, Crosby, Ovechkin, Bergeron, Stamkos, Marchand, Malkin, Kessel, Backstrom, Seguin, Toews, Pavelski, Getzlaf… they all had blips, landed around the 75th percentile mark in a season and bounced back. (The list is longer than where I stopped, btw.)

This touches on the noisy nature of 5v5 P60. It’s the weakest scoring metric to use when evaluating forwards, as the correlation numbers I’ll share in a bit will show.

Ryan said on Friday that producing at 5v5 is the hardest aspect of the game. If that’s the case, what can we say about forwards who score in the top ranks at 5v5 in a given season, i.e., forwards who’ve managed to excel at the “hardest aspect of the game”? 5v5 P60 should identify the best, most successful offensive forwards, right? Here, then, are the top 30 forwards at 5v5 P60 from last season, among forwards who played at least 10 games. (If I included all forwards, regardless of GP, the top 5 on 5v5 P60 would be Ryan Poehling, Brendan Gaunce, Zac Dalpe, Dominic Toninato, and Adam Johnson.)

Player, 5v5 P60

Nikita Kucherov, 3.37
Max Comtois, 3.33
Sidney Crosby, 2.92
Andrew Shaw, 2.9
Mitchell Marner, 2.87
John Tavares, 2.87
Johnny Gaudreau, 2.84
Connor McDavid, 2.82
Patrick Kane, 2.8
Max Domi, 2.8
Artemi Panarin, 2.8
Auston Matthews, 2.76
Brayden Point, 2.75
Patrice Bergeron, 2.69
Steven Stamkos, 2.67
Jake Guentzel, 2.67
Tomas Tatar, 2.66
Brett Connolly, 2.66
Brad Marchand, 2.66
Tomas Hertl, 2.61
Timo Meier, 2.58
Sean Monahan, 2.58
Leon Draisaitl, 2.57
Taylor Hall, 2.54
Viktor Arvidsson, 2.52
Paul Stastny, 2.52
Andreas Johnsson, 2.49
Alex Tuch, 2.48
Matt Duchene, 2.47
David Krejci, 2.47

The list above is supposed to represent successful offensive performers, as measured by 5v5 P60. Let’s compare it to the list of the top 30 forwards as measured by Pts/GP, again for forwards who played at least 10 games last season:

Player, Pts/GP

Nikita Kucherov, 1.56
Connor McDavid, 1.49
Patrick Kane, 1.36
Leon Draisaitl, 1.28
Sidney Crosby, 1.27
Brad Marchand, 1.27
David Pastrnak, 1.23
Patrice Bergeron, 1.22
Nathan MacKinnon, 1.21
Johnny Gaudreau, 1.21
Steven Stamkos, 1.20
Mikko Rantanen, 1.18
Aleksander Barkov, 1.17
Brayden Point, 1.16
Mitchell Marner, 1.15
Jonathan Huberdeau, 1.12
Taylor Hall, 1.12
Blake Wheeler, 1.11
Artemi Panarin, 1.10
Alex Ovechkin, 1.10
Auston Matthews, 1.07
John Tavares, 1.07
Jack Eichel, 1.06
Evgeni Malkin, 1.06
Sean Monahan, 1.05
Claude Giroux, 1.04
Alexander Radulov, 1.03
Gabriel Landeskog, 1.03
Mark Scheifele, 1.02
Sebastian Aho, 1.01

Which list would you say is more exclusive, tougher to break in to? Which list is better represented by “successful offensive performers”? Which metric better aligns with your idea of highly skilled, offensive forwards?

There are 16 names common to both lists, leaving 14 distinct names that are unique to each list, Which unique list of 14 would you bet on for future offensive performance?

Let’s go back a season to see what would have happened if we bet on the players on the 5v5 P60 list who didn’t make the Pts/GP list. In 2017-18, the following forwards ranked in the top 30 on 5v5 P60 but didn’t rank in the top 30 on Pts/GP:

Player, 17-18 5v5 P60, 18-19 5v5 P60

Valentin Zykov, 3.34, 0.76
Anthony Cirelli, 2.88, 1.69
Evgenii Dadonov, 2.74, 2.31
Yanni Gourde, 2.65, 2.08
Matthew Peca, 2.53, 1.35
Filip Forsberg, 2.53, 2.19
Ryan Spooner, 2.5, 0.92
William Karlsson, 2.49, 1.93
Ty Rattie, 2.49, 1.16
Jordan Eberle, 2.49, 1.41
Eric Staal, 2.48, 1.68
Thomas Vanek, 2.47, 1.87
Jaden Schwartz, 2.46, 1.77
Reilly Smith, 2.45, 1.91

As you can see, none of these names ranked in the top 30 on 5v5 P60 in 18-19. The mean 5v5 P60 for this group in 18-19 was 1.65, compared to 2.61 the season before, a pretty strong regression to the mean. This gives you a sense of how problematic 5v5 P60 is for forecasting future performance. Players yo-yo in a way that makes you question whether the metric is doing a good job of capturing underlying skill: skill, where it exists, should show through in repeated measurements.

The case against 5v5 P60 is much stronger than the examples I’ve given here. I’m hoping the examples get you thinking: a good, accurate metric shouldn’t run roughshod over your intuition and it shouldn’t produce such random results in consecutive measurements.

The issue I see with your argument isn’t a 5v5 vs All situations issue, it’s a per game vs per minute issue as all minutes aren’t created equally.

Try that list with 5v5 pts/gp and you’ll likely have a better picture of the best players than all situations pts/gp.

If you want to go to 5v5 pts/60 you need to separate the players who play against top competition from those who don’t via their 5v5 TOI. If you draw a line in the sand for forwards at approximately 12.5 minutes for top 6 forwards, or 13.72 for top line forwards. Anyone who plays less 5v5 minutes than that should not be included in your list of elite players. That will give you a list similar to the 5v5 pts/GP list.

Your argument changed two factors and argued that one was at fault (5v5 vs all situations) when I believe it was the other at fault ( per game vs per minute).