|
We thought we'd start off the new year with a new statistic, one that we hope can start to answer questions including which players are the league's best playmakers, and which players pass when they should be shooting (and vice versa). In so doing, we discovered Brian Campbell's unnaturally strong season in Florida last year, and an underrated playmaking star in Colorado.
Since shots can be an effective way to track a player's goal-generating performance in a way that's independent of his shooting luck, let's define a sister statistic called "passes", which is the number of passes that directly resulted in shots (you can call them "shot assists" if you want). Unlike shots, this statistic isn't tracked by the NHL, but we can create an estimate by dividing a player's assists by the average number of assists per goal (1.7) and then dividing that by the average on-ice shooting percentage of his linemates. This approach, which could admittedly be further refined by separating out power play time, should provide a decent estimate of how many of the player's passes resulted in shots on goal.
Passing leaders, 2011-12
Player Team Passes
Brian Campbell Florida 476
Joe Thornton San Jose 416
Henrik Sedin Vancouver 408
Claude Giroux Philadelphia 402
Ryan Getzlaf Anaheim 360
Erik Karlsson Ottawa 358
Anze Kopitar Los Angeles 349
Mark Streit NY Islanders 341
Alex Pietrangelo St. Louis 330
Ryan O'Reilly Colorado 329
This new statistic really highlights Brian Campbell's resurgence in Florida, resulting in 476 passes—the highest total in the past four seasons. While Campbell saw a steady decrease in Chicago from 321 in 2008-09 to 232 in 2009-10, and finally 140 in 2010-11, his heavy-minutes, first-line usage in Florida gave him the opportunity for maximum offensive contribution.
The remainder of the top 10 offers up few surprises except young Ryan O'Reilly, who played on Colorado's tough minutes line along with Calder Trophy winner Gabriel Landeskog. O'Reilly, who managed just 136 and 114 passes in the previous two seasons, shot up to 329, ahead of established playmakers like John Tavares (324), Eric Staal (320), and Evgeni Malkin (319).
Pass-to-Shot Ratio
The interesting thing about top playmakers like Joe Thornton and Henrik Sedin is that their own shooting percentage is consistently higher than the average of their linemates—so perhaps they pass too often?
Pass-to-shot ratio is simply a player's passes divided by his shots, and in order to maximize offense you would hope that it would match up with the relationship between the shooting percentage of a player's linemates and his own. Unfortunately for forwards, as you can see, it often doesn't!
Pass-to-shot ratio leaders, 2011-12
Forward Team PSR SH% On-ice SH%
Henrik Sedin Vancouver 3.6 12.4% 9.5%
Joe Thornton San Jose 2.7 11.5% 8.2%
David Desharnais Montreal 2.5 16.3% 10.4%
Alex Tanguay Calgary 2.3 15.4% 10.5%
Kyle Wellwood Winnipeg 2.3 19.4% 7.8%
Nicklas Backstrom Washington 2.1 14.7% 8.6%
Marcus Johansson Washington 2.1 15.6% 10.0%
Jussi Jokinen Carolina 2.0 10.2% 8.5%
Ryan Getzlaf Anaheim 1.9 5.9% 7.4%
Mike Ribeiro Dallas 1.9 12.7% 9.5%
Minimum 100 Passes
In fairness, forwards tend to have higher shooting percentages (average 10.2%) than on-ice shooting percentages (average 8.0%) for two reasons. The former is boosted by the power play while the latter is not (once again: this new statistic can be refined), and also because defensemen take a lot of low probability shots.
Even once that's taken into account, of the 10 forwards with the highest pass-to-shot ratio, only Ryan Getzlaf and possibly Jussi Jokinen appear to clearly be making passing decisions wisely. It's puzzling why certain players pass twice as often as they shoot when they're clearly capable scorer. Arguably the Vancouver Canucks could have scored four more goals if Henrik Sedin were closer to pass/shot parity—which equates to more than a point in the standings and a million dollars in value.
On the flip side there are obviously some players with very low pass-to-shot ratios—but with a few exceptions they aren't the players who would have guessed.
Lowest pass-to-shot ratios, 2011-12
Forward Team PSR SH% On-Ice SH%
Mike Santorelli Florida 0.2 7.7% 4.6%
Rene Bourque CGY/MTL 0.3 11.4% 7.6%
Sean Bergenheim Florida 0.3 9.2% 6.5%
Eric Nystrom Dallas 0.4 15.7% 6.8%
Jeff Carter CBJ/LA 0.5 11.4% 8.4%
Jim Slater Winnipeg 0.5 11.0% 8.1%
Jason Blake Anaheim 0.5 6.4% 5.4%
Patric Hornqvist Nashville 0.5 10.9% 8.5%
David Clarkson New Jersey 0.5 13.2% 7.8%
David Dorsett Columbus 0.5 8.8% 6.5%
Minimum 100 Shots
With only a couple of exceptions at best, not many of them have the sniping skills to warrant their tendencies to take two or more shots for every pass. In other news, Jeff Carter was probably not the best linemate for sniper Rick Nash!
Wrapping Up
This new statistic is meant to relate to assists in the same way that shots relate to goals, except that we must estimate these totals until they're counted manually. Much like shots, the advantage of looking at passes is the removal of the influence of shooting percentages so you can truly see who is most adept at creating shots, and who is passing when they should be shooting.
That's it for our preliminary look at this new statistic. Stay tuned for further refinements, and check out the raw data for yourself over at Hockey Abstract.
Robert Vollman is an author of Hockey Prospectus.
You can contact Robert by clicking here or click here to see Robert's other articles.
| 
1 comment has been left for this article. 
NHL Preview: Impact Of The 48-Game Schedule
This is a nice idea, though as you note it needs a little refining.
It looks like you've used a player's overall assist total and his 5v5 on-ice shooting percentage. This has a couple of flaws...
For one, there's the obvious mismatch between overall and 5v5. This creates a bias towards players who spend a larger fraction of their time on the PP, since PP time gives them extra assists without impacting their 5v5 on-ice shooting percentage.
For another, the player's own shooting factors into his on-ice shooting percentage, which means that when his shots go in the net, it reduces your estimate of how many passes he made. (There's also the trickier issue that he doesn't get the chance for a rebound leading to a teammate's goal, which would count as an assist and some ~10-12 passes in this stat even though he didn't actually pass.)
For another, I think you should decide whether you want to count only the passes that lead directly to a shot (in which case you'd use primary assists only) or count all passes that could potentially lead to an assist (in which case there's no reason to divide by 1.7, since there can be multiple passes per shot).
Finally, I always prefer rate stats to counting stats, for the usual reasons.
This then gives the following calculation, using 5v5 stats that are all available on behindthenet.ca:
A1/60 [optionally + A2/60] / [(GFON/60 - G/60) / (GFON/60 + SvFON/60 - G/60 - Sv/60)]
Another systemic note: the estimation of passes removes some of the luck from a player's assist total, but certainly not all of it. It's entirely possible that Player X could find that his teammates fail to score on the shots that come from his passes but often score on his linemate's passes (either because he's not as good of a playmaker as his linemate or often because of luck), which would give him a misleadingly high on-ice sh% and therefore a misleadingly low pass total.
Finally, I'm a little suspicious of the interpretation of pass-to-shot ratio and individual shooting percentages as a read on who's shooting more than they should. There's a lot of interconnectedness in there (even after you fix the bugs in the data) -- a rebound on a player's saved shot can produce an assist (and end up counting as 12 passes in your analysis, whereas if his shot went in he'd get none), for example. Moreover, we know shooting percentages fluctuate from year to year -- this analysis seems to imply that after shooting 16.2% last year, when he found his shots not going in this year Getzlaf would have been wise to focus on passing over shooting this year when his shooting percentage dipped, whereas I would be inclined not to have a player change his strategy based on such fluctuations.
Overall, I like where this is headed -- I often do something similar implicitly when I look at on-ice sh% in projecting future assists, and I really like the idea of doing it explicitly (and trackably) as you have here. So please read all of this as helpful suggestion rather than complaint.