September 30, 2010
Numbers On Ice
What Makes Good Players Good, Part Three

by Tom Awad

In part 1, we 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, we saw that GPs tend to play against stronger opponents and with stronger teammates, and to receive vastly more power-play ice time than their brethren. Here we will look at their traditional stats to see if they give us any insight on why GPs excel, and then we'll summarize our conclusions.

5. Goals, Assists and Points

Despite being the standard way that we measure offense, I have left goals and assists for the end of my analysis. The reason for this is that goals and assists are the end result, rather than the cause, of performance. It’s more interesting to know how and why a team scored than just the fact that they scored.

However, there is still a lot to be learned from goals and assists, and if nothing else it will help us put into context the numbers which we most commonly see attached to our favorite players. All numbers below are at even-strength:

2009-10 aggregate results

	
	 GP    TOI    GF     G    A   EG  Shots  SH%  ESH% Pts/60 Sh/60	Pts/G
1st tier 5998  85400  4073 1187 1813 1049 11493 10.3% 9.1% 2.11   8.1	73.7%
2nd tier 6545  84507  3549 1109 1383 1040 11292  9.8% 9.2% 1.77   8.0	70.2%
3rd tier 7187  84750  3138  890 1227  927 10087  8.8% 9.2% 1.50   7.1	67.5%
4th tier 9414  84622  2546  685  922  832  9392  7.3% 8.9% 1.14   6.7	63.1%

2008-09 aggregate results

	
	 GP   TOI   GF	   G	A   EG	Shots  SH%  ESH% Pts/60	Sh/60 Pts/G
1st tier 5984 80036 3668 1150 1580 1033	11014 10.4% 9.4% 2.05	8.3   74.4%
2nd tier 5899 72847 3085  943 1269  896	 9626  9.8% 9.3% 1.82	7.9   71.7%
3rd tier 6027 69688 2687  768 1081  769	 8211  9.4% 9.4% 1.59	7.1   68.8%
4th tier 8176 76836 2389  654  924  786	 8748  7.5% 9.0% 1.23	6.8   66.1%

(A reminder: when we speak of tiers in 2008-09, we are talking about the exact same players in each group as in 2009-10. This allows us to see if certain characteristics are sustainable over seasons.)

Obviously, GPs accumulate more goals and assists than other players. My colleague Rob Vollman has argued convincingly that the distinguishing feature of top-6 forwards is that they score at least 1.7 points per 60 minutes of even-strength time. We see exactly that from the table above. 2nd tier players averaged 1.8 Pts/60 both last season and the season before, while 1st tier players averaged 2.1. These are also the players, as we saw, who received 70% of the power-play ice time.

We also see that there is no aggregate difference between the quality of shots that 1st and 3rd tier players take, and only a small difference with 4th tier players (this is noted in the ESH%, which indicates the expected shooting percentage of these players given where their shots were taken from), but there is a heck of a difference between their shooting abilities: GPs sank 10.3% of their shots, far more than expected, while 4th tier players sank only 7.3% of theirs. Even taking into account the small difference in shot quality, this means a top line player is 40% more likely to score than a grinder from the exact same spot.

Points per goal is an interesting statistic analyzed by Tyler Dellow which asks what fraction of the offense is a player involved in. Tyler’s conclusion, which I agree with, is that you can judge a player better by the on-ice shooting percentages he drives than by the fraction of the points he is involved with. Our numbers here seem to agree with this: the spread in finishing ability, analyzed in section 2, is larger than the spread in Pts/G that we observe here. In other words, Joe Thornton is likely to have something to do with that goal the Sharks just scored even if he wasn’t credited with an assist. Maybe his presence kept the defense honest just because he’s Joe Thornton. Maybe he created a distraction. The point is good players contribute in all kinds of ways. (Don’t believe me? Watch Mario’s non-assist on Paul Kariya’s goal in the 2002 Olympics).

At the extreme, the best way for a player to have a high Pts/G is to be a generational talent playing with mediocre linemates. Sidney Crosby, come claim your title.

6. Conclusions

To summarize the conclusions of this study:

1. Ice time obviously correlates positively with player skill.
2. Good players have above average skill in puck possession, shot quality and finishing ability. The shot quality skill is minor, while the shot volume and finishing skills are major. Ergo, good players on average will have positive Corsi, Delta and +/- and above-average on-ice shooting percentages and PDO.
3. Good players show no ability to reduce the shooting percentage of their opponents.
4. Good players start slightly more often in the offensive zone than the defensive zone, play against above-average opponents and with above-average teammates.
5. Good players play significantly more on the power-play, and achieve better results than their peers. They do not play significantly more on the penalty-kill, but do achieve slightly better results when they do.
6. Good players will average 2.1 Pts / 60 min of Even-Strength ice time, and will score points on a slightly above-average fraction of their team’s goals while they are on the ice.

While I feel that I’ve spent four thousand words to confirm things we already knew, I hope I’ve at least helped to quantify how, and how much, top players distinguish themselves from their peers. Nearly everything we can think of as a potential skill actually is one, and has measurable differences among the NHL population; the difference is the magnitude of those skills. Knowing which is which helps set realistic expectations for the future.

Next time we'll take a look at Defensemen.

Tom Awad is an author of Hockey Prospectus. You can contact Tom by clicking here or click here to see Tom's other articles.