Further to: The limits of Corsi

March 4, 2014 in Unfiltered by Corey Pronman

In 2012, Tom Awad wrote a very important piece he coined “The Limits of Corsi”. He discussed a concept I’m sure most would agree with. Not every team regresses 100% to a 1000 PDO. For some teams, some of the components are skill. Boston has great goaltending, New Jersey doesn’t have great forwards, and to some degree there will be some persistence in their percentage stats. Tom in that column discussed why Columbus’ poor performance wasn’t all due to bad luck. 

Using the same methods Tom used in his column, I wanted to take a revisited look at Toronto. A team that had great notoriety in the preseason and start of the season due to their possession metrics pointing towards a demise. Yet Toronto has fought on (mind you it’s a poor conference) and after a little over 100 games, seem to be defying their possession metrics. They’ve been 29th in Fenwick Close for the last two seasons. So what gives, are they for real?

Here is a table showing each team’s shot and goal numbers from the last two seasons, splitting the goal differential up into two components, the part due to their possession/shot metrics and the part due to PDO. To calculate the first one, you assume league average (and can do it by situation) and subtract the remainder to get the second component.

Shot Dif Goal Dif GD  (shots) GD (PDO)
TOR -815 5 -70 75
PIT 190 89 16 73
ANA 169 74 15 59
WSH -394 17 -34 51
CHI 599 100 52 48
STL 345 75 30 45
BOS 330 73 28 45
CBJ -270 11 -23 34
TBL -110 19 -9 28
MTL 38 29 3 26
COL -272 -12 -23 11
DAL -141 -10 -12 2
MIN 15 -2 1 -3
SJS 579 45 50 -5
EDM -616 -59 -53 -6
NYR 340 23 29 -6
DET 147 3 13 -10
LAK 502 33 43 -10
OTT -24 -13 -2 -11
PHI -28 -14 -2 -12
BUF -746 -77 -64 -13
PHX -1 -17 0 -17
VAN 94 -10 8 -18
WPG -18 -20 -2 -18
NJD 332 -22 29 -51
CGY -244 -76 -21 -55
NYI 158 -42 14 -56
NSH -60 -63 -5 -58
CAR 61 -54 5 -59
FLA -155 -105 -13 -92

Toronto leads the league in goals due to PDO. In Tom’s example using Columbus over 2.5 seasons, Columbus had a Goals (PDO) of -119. Toronto is roughly on pace for that over 2.5 seasons. 

The standard dev for Goals (PDO) is 47, and to estimate the amount of noise/luck on Toronto’s, we take their sqrt (GF+GA)  of 331+ 326 which equals 25.6. The standard deviation of skill for Toronto on Goals (PDO) would be (47^2 -25.6^2) or 40. The exact number coincidentally for Columbus in Tom’s example.

Toronto’s mark of 75 Goals (PDO) would be nearly 3 standard devs which happens 0.3% of the time or it would have a 3 in 1000 chance of happening by just luck.

———–

In English Please:

I don’t think it’s controversial to say that the Leafs shooting and save percentage are not 100% due to luck. They’ve shown some very quality goaltending from James Reimer and Jonathan Bernier over the last 100 games and their shooter have finished at a high rate.

What I am not saying, is that they have super special shooting/goalie skills that will continue to play at this very high level for years. You should still bet against PDO. 

However not all teams regress all the way to a 1000 PDO. Some do have particular skills to some degree in the components.

I thought Toronto would be terrible this season. I think my reasoning at the time was sound. Bernier was very unproven, we hadn’t seen a long stretch of dominance from Reimer, Toronto was way outchanced in the shortened season, and it’s always a good bet to say a team’s SH% will regress all the way to the league mean.

Time and time again, we’ve seen fans cry their team has unique SH% talents that will persist just to see it not happen. Think of Colorado and Minnesota from years’ past. Toronto doesn’t usually happen, but they have happened. Most of the time teams don’t maintain high SH% and SV% marks. What we can learn from the case of Toronto, and Columbus, Boston, Vancouver ( see Tom’s column) etc. is maybe we should be delving deeper than we’ve gone so far to look for indicators of when a team may actually have some skill in a particular area.

Corey Pronman