A week or so ago I was reading an article on another blog that displayed the Penguins’ possession stats at 5v5. They were impressive across the board. Yet there we were, sitting 5-5-1 at the time with one regulation win.
Scanning team Corsi rankings this morning, I noticed the Flyers…seventh overall in points percentage…are dead last (44.55) and the eleventh-place Capitals aren’t much better (25th at 47.51). The Islanders, also ahead of us in the standings, are only marginally better than the Caps with a Corsi of 48.72.
Meanwhile, the Pens with their glistening 53.64 Corsi sit a distant 20th in points percentage.
It got me to thinking. How important is puck possession with regard to winning the Stanley Cup?
I decided to do a little digging.
Before I review the results, a little background. Corsi is an advanced statistic used in the National Hockey League to measure shot attempt differential while at even strength play. This includes shots on goal, missed shots on goal, and blocked shot attempts towards the opposition’s net minus the same shot attempts directed at your own team’s net.
The first table displays the Corsi and corresponding Corsi ranks for Stanley Cup champions, President’s Trophy winners and our Penguins dating back to 2007-08, when the statistic was first kept.
Not surprisingly, there is a strong correlation between Corsi and capturing the Cup. From ’07-08 through ’15-16, Cup winners ranked a shade above 5th in team Corsi, and that includes our ’08-09 Pens…an anomaly at 19th. Three Cup winners led the league in Corsi, three others…including the ’15-16 Pens…finished 2nd.
However, over the past few seasons there’s been a pronounced shift. The average Corsi rank of the past four Cup winners is 13.75, and that includes last season’s Lightning squad that ranked 5th.
There’s a similar skew among Presidents’ Trophy winners. Going back to the ’14-15 Rangers (20th in Corsi), top regular-season point getters have ranked an average of 12th in team Corsi. The present league-leader in points, Toronto, ranks 16th.
New Pens president of hockey operations Brian Burke took some heat recently from metrics-oriented blogs with the following observation:
“Pittsburgh’s the last team to win with a small, fast team. The last three Cup winners have been big, they play long pants hockey, they play ugly hockey. I think we need to look at that and see, have we kept up with the Joneses here?”
You know what? The numbers appear to support Burke’s claim.
|
|
Stanley Cup Champion |
Presidents’ Trophy Winner |
Penguins |
|||||
|
Season |
Team |
Corsi |
Rank |
Team |
Corsi |
Rank |
Corsi |
Rank |
|
20-21 |
|
|
|
TOR |
49.75 |
16 |
53.64 |
5 |
|
19-20 |
TBL |
52.25 |
5 |
BOS |
50.85 |
12 |
50.86 |
11 |
|
18-19 |
STL |
51.50 |
10 |
TBL |
51.59 |
9 |
49.73 |
15 |
|
17-18 |
WSH |
47.96 |
24 |
NSH |
51.48 |
10 |
52.28 |
5 |
|
16-17 |
PGH |
50.14 |
16 |
WSH |
51.81 |
4 |
50.14 |
16 |
|
15-16 |
PGH |
52.72 |
2 |
WSH |
51.04 |
14 |
52.72 |
2 |
|
14-15 |
CHI |
53.60 |
2 |
NYR |
49.48 |
20 |
52.82 |
5 |
|
13-14 |
LAK |
56.82 |
1 |
BOS |
53.85 |
4 |
48.64 |
22 |
|
12-13 |
CHI |
54.14 |
4 |
CHI |
54.14 |
4 |
48.99 |
18 |
|
11-12 |
LAK |
54.74 |
2 |
VAN |
52.75 |
7 |
54.23 |
3 |
|
10-11 |
BOS |
50.71 |
14 |
VAN |
52.23 |
6 |
52.73 |
4 |
|
09-10 |
CHI |
56.54 |
1 |
WSH |
52.88 |
3 |
52.06 |
6 |
|
08-09 |
PGH |
48.07 |
19 |
SJS |
54.22 |
5 |
48.07 |
19 |
|
07-08 |
DET |
58.77 |
1 |
DET |
58.77 |
1 |
45.48 |
29 |
|
Average |
|
|
7.8 |
|
|
8.2 |
|
11.4 |
I also looked at the overall league standing (based on points percentage) for teams that led the league in Corsi and for those that had the worst Corsi.
The results were somewhat more predictable. Teams that paced the league in Corsi generally (but not always) finished higher in the standings, and teams that had the worst Corsi were usually among the worst teams.
In conclusion, is puck possession important? Yes. But it’s not the be-all-end-all factor that some might lead you to believe.
|
|
Best Corsi |
Worst Corsi |
||||
|
Season |
Team |
Corsi |
Standing |
Team |
Corsi |
Standing |
|
20-21 |
COL |
56.63 |
8 |
PHI |
44.55 |
7 |
|
19-20 |
VEG |
54.76 |
8 |
DET |
45.99 |
31 |
|
18-19 |
SJS |
54.87 |
6 |
OTT |
44.90 |
31 |
|
17-18 |
CAR |
54.48 |
21 |
NYR |
45.92 |
24 |
|
16-17 |
LAK |
54.99 |
22 |
ARI |
45.04 |
28 |
|
15-16 |
LAK |
56.37 |
8 |
COL |
44.20 |
21 |
|
14-15 |
LAK |
55.34 |
18 |
BUF |
37.53 |
30 |
|
13-14 |
LAK |
56.82 |
9 |
TOR |
42.82 |
22 |
|
12-13 |
LAK |
56.30 |
8 |
CGY |
47.36 |
26 |
|
11-12 |
DET |
54.90 |
9 |
MIN |
44.26 |
24 |
|
10-11 |
SJS |
53.56 |
5 |
ANA |
44.34 |
9 |
|
09-10 |
CHI |
56.54 |
3 |
FLA |
44.57 |
28 |
|
08-09 |
DET |
57.25 |
3 |
PHX |
45.40 |
26 |
|
07-08 |
DET |
58.77 |
1 |
ATL |
42.89 |
28 |
|
Average |
|
|
9.2 |
|
|
23.9 |
Hey Rick,
Thanks for the digging, Rick. You know I am a stat geek.
However, I have never been all that impressed with CORSI. Stats don’t lie, they are what they are, empirical data. The problem with stats in general and CORSI in particular is in how they applied and CORSI is way over generalized. In effect it often is used like some people, rather than using the proper tool, use a screw driver to hammer a nail or worse (maybe a saw to hammer the nail – very dangerous).
Whether you are talking team v team or player v player, there are far too many confounding variables for blanket application of CORSI. CORSI does not account for shot quality. The Pens last game against the Isles, our black and gold held a 52.7% edge in CORSI but the Isle held the edge in HD shots 12 – 7, Sully was lucky to sneak that game to a SO and a win.
CORSi also doesn’t account for game situations such as who is winning. In the Pens last game, they never trailed. It was the Caps that had to keep fighting back. Our home town heroes lost the CORSI battle, only 49.41 and the HD battle 6 – 11 but managed their 2nd regulation win.
More importantly, Hockey games are won and lost based on goal differential, not shot attempt differential.
When looking at players, not only do you need to look at the above variables, but now you have to look at the with or without you differences in their line mates and their opponents. You saw some of the stats with which I was playing while I was still fearful of a Simon return; how his CORSi numbers were bouyed up in blowout games (wins or losses by more than 1 goal) but when the game was on the line during 1 goal difference games his CORSI was actually avg to below avg. How Simon’s CORSi numbers were actually below the numbers one would expect based on the avg CORSI of his line mates and below the avg numbers his opponents gave up. Simon’s numbers were also highly inflated by the disproportional number of offensive zone starts he had. To make matters worse, the most important stat, goal differential, for Simon was inversely proportional to his CORSI. If there ever was a Perimeter player, it was Simon.
CORSI can be interesting when looked at properly, when game situations are controlled for. No sense looking at 5-on-5 numbers if you are taking lots of stupid penalties or drawing tons of penalties. CORSI is also kind of useless in stat padding games.
Rick – Very good assessment.
Not trying to beat a dead horse but I’ve said many times that
the “numbers lie”!! I personally lived through it.
Go Pens