In part 1, I divided all forwards into four groups, ranked by even-strength ice time, and observed that players who received the most ice time (Good Players, aka GPs) tended to outplay their opponents in shot volume, shot quality and shooting percentage. In part 2, I will analyze whether these players play more in offensive or defensive situations, the caliber of their opponents and teammates and the amount of special teams ice time they receive.
3. Situation, Opponents and Teammates
So far, I have looked at the on-ice results for the four tiers of players assuming that all minutes of ice time were played under the same circumstances. We know that this is not the case. Players are given tasks that match up well with their skills. Better players will be given the most difficult matchups, defensively minded players will start in the defensive zone, and so on. Can we quantify this?
We can with Delta. Delta is designed to jump from Delta to DeltaS, DeltaSO and DeltaSOT, each time adjusting for S (Situation), O (Opponents) and T (Teammates). Each of these adjustments is measured in goals.
2009-10 Delta results
Group DeltaDS DeltaDO DeltaDT Total
1st tier -24 103 -44 35
2nd tier -9 55 -38 9
3rd tier 20 -7 7 20
4th tier 13 -146 83 -50
Sadly for my pretensions to teaching you something new, pretty much everything you see in this table is what you would expect. I will explain in detail: DeltaDS is “Difficulty of Situation”, referring to Zone Starts. A positive value means that you started more often in the defensive zone (more difficult), while a negative value means that you started more often in the offensive zone (easier). We see that GPs had a slight bias towards starting in the offensive zone.
As an aside, there is a meme going around on the internet that players who start uniquely in the offensive zone are worse or sheltered, while players who start exclusively in the defensive zone are excellent. This is incorrect: players start in the offensive zone because their offensive skill is greater than their defensive skill. This could be because their defensive skill is lacking (Marc-Andre Bergeron), it could be because they are really good offensively (Marian Hossa), or it could be a bit of both (Patrick Kane). The proof is that Good Players, to whom we give the most ice time, do not have a leaning for defensive zone starts. You do, however, have to take Zone Starts into account when looking at a player’s results. Players who start exclusively in the defensive zone will tend to have poor Corsi and Delta ratings and, inevitably, poor +/- as well. This is not because they suck; it’s because their job is difficult.
Looking at difficulty of Opponents (DeltaDO), we see that power vs. power is indeed the preferred matchup method in the NHL: GPs played against the strongest opposition of the four groups, facing adversaries who had +103 Delta overall, while 4th tier players were trusted with the weaklings. By the same token, GPs had the best teammates (the number is negative because it is measured as a difficulty: good teammates make your job easier). We can now explain one of the mysteries from a previous table: why the on-ice save percentage for Good Players was lower than for other players. If you look at the table of shooting percentages and PDO above, you can see that the opponents’ shooting percentage was highest against 1st tier players and slowly crept down by group. This is not because GPs allow more dangerous shots than their teammates: it’s because they play against better players! Naturally, whoever is tasked with shutting down first-line players will see the opponents finish their chances at first-line rates.
While the exact numbers differ, we can see that the overall situation in 2008-09 was similar. Good Players got favorable Starts, better Opponents and better Teammates, while weaker players got the reverse:
2008-09 Delta results
Group DeltaDS DeltaDO DeltaDT Total
1st tier -29 92 -92 -29
2nd tier 8 41 -27 22
3rd tier 14 4 30 48
4th tier 6 -103 73 -24
4. Special Teams
So far, we have filtered GPs by even-strength ice time. But we know that one of the major ways in which good players distinguish themselves from their peers is by their special teams play. Penalty-killing ice time is a sign of defensive awareness and your coach’s confidence in you; power-play ice time shows offensive skill. We would probably also expect GPs’ results to be a bit better than their peers on special teams, much as we saw at even strength. (A note on my methodology: if I had wanted to find the truly best players, I should have ranked them by total ice time, not just even-strength. But I wanted to see how their special teams ice time ranked by tier, so I couldn’t include that in the initial sorting function. In practice, guys with high overall ice time, like Jeff Carter, Kyle Okposo or Joe Pavelski, are better than guys with just high even-strength ice time like Milan Lucic or Wayne Simmonds).
2009-10 per-60 min results at 5-on-4
TOI(s) GF GA EGF EGA Corsi +/- Delta Exp +/- (Corsi)
1st tier 179 7.06 0.80 6.83 0.80 46 6.26 5.89 5.61
2nd tier 138 6.35 0.67 6.38 0.77 43 5.69 5.49 5.22
3rd tier 84 6.07 0.72 6.19 0.82 41 5.35 5.25 4.95
4th tier 26 4.87 0.80 5.67 0.81 37 4.06 4.73 4.47
2009-10 per-60 min results at 4-on-5
TOI(s) GF GA EGF EGA Corsi +/- Delta Exp +/- (Corsi)
1st tier 57 1.08 6.05 1.02 6.08 -39 -4.97 -5.06 -4.75
2nd tier 66 0.67 5.94 0.84 6.20 -43 -5.27 -5.36 -5.20
3rd tier 63 0.69 6.50 0.73 6.44 -44 -5.81 -5.71 -5.37
4th tier 47 0.51 6.59 0.58 6.27 -44 -6.08 -5.69 -5.40
Power-play and penalty-killing use could not be more different. We see a clear correlation between even-strength ice time and power-play ice time: tier 1 players played 179 seconds, almost 3 minutes, per game on the power-play and the 2nd tier players got a respectable 2 min 18 s, vs. 1 min 24 s and a meager 26 seconds for 3rd and 4th tier players respectively. Of our 83 1st tier players, the only ones who got less than 90 seconds of power-play ice time per game were Miroslav Satan (69 s), Radek Dvorak (55 s), David Jones (55 s), Milan Lucic (39 s) and Wayne Simmonds (9 s!), and of these only Dvorak and Simmonds played the entire season. It seems that, for forwards, the skills useful at even-strength are roughly the same as those on the power play. We see this in the results as well: the 1st tier players had better Corsi, better shot quality and better shooting percentage than their brethren once again, mirroring the results at 5-on-5.
What about penalty-killing? Here the story is a little murkier. There was, in effect, no correlation between penalty-killing ice time and even-strength ice time, although this does mean that the lower on the totem pole you are, the bigger the fraction of your ice time spent with a man down. Despite, this, 1st tier players still obtained better results on the PK, in terms of reduced shots and goals against. A small fraction of this could be explained by more favorable zone starts, but not much. Teams killing penalties get very few offensive zone faceoffs.
What explains this discrepancy? First, distribution of skills. Many offense-first players do fit the stereotype of the scoring forward who can’t or won’t backcheck, while others simply lack the required defensive skill. Having uncanny passing ability and a rocket of a shot is enough to get you on the 1st line, but killing penalties will be Kryptonite. On the other hand, those players good enough to justify both a 1st line role and a regular shift on the penalty kill will tend to be exceptional defensive players, and it will show in their results.
Secondly, opportunity cost. 1st tier players are so valuable, and their skills so much better employed with offensive opportunities, that is makes sense to reserve them for even-strength and power-play situations and let others handle the load of penalty killing, unless they are truly exceptional. Sidney Crosby, Jonathan Toews and Pavel Datsyuk are all good at killing penalties, but is that really what you want them doing with their time and energy? There were only 8 first-tier players who saw more than 2 minutes of penalty-killing ice time per game: Ryan Kesler, Antoine Vermette, Tomas Plekanec, Patrick Marleau, Radek Dvorak, Rick Nash, Brandon Dubinsky and Mikko Koivu. All but Dvorak also saw significant power-play time as well. These 7, plus Mike Richards who just missed the cut, are the most versatile forwards you can have on your roster. If anybody was wondering why Plekanec, Koivu and Marleau all got huge multi-year contracts this summer, you have part of your answer.
A simpler way of putting it is: almost every 1st line forward has offensive skill, but not every 1st line forward has defensive skill.
We’re almost done! Next time I will look at Goals, Assists and Shots, the traditional stats, and conclude my study.
Tom Awad is an author of Hockey Prospectus.
You can contact Tom by clicking here or click here to see Tom's other articles.