The Impact of Power Plays in
NHL Hockey, or: No Dogs Play
Hockey
Jordan Pedersen, Tom Geiger,
and Waleed Khoury
Some Background



A team is said to be on a
power play when at least
one opposing player is
serving a penalty AND
The team has a numerical
advantage on the ice
Whenever both teams
have the same number of
penalties being served,
there is no power play
Methodology



Using League Data for
this season and the
preceding four seasons
Focusing on Rankings
(based on points) and
power play conversion
percentages
Regular Season
Refresher on 2-Sample T Test


Used to analyze whether the difference in means
between two independent groups of data is statistically
significant
1-sample t-tests are for measuring the difference in
mean for a single group versus a hypothesized mean
Use of the Test



Test will be used to see if there is a
statistically significant difference between
power play conversion percentages for top
15 teams vs. bottom 15 teams (based on
their Point total)
Therefore, our question is…
Does the ability to convert power plays
really affect a team’s ability to win games?
Current Season

Open up PP%Top and Bottom.mtw
We’ll be doing 2008-2009

A surpriiiiithe?

Mandatory Class Exercise

¼
¼
¼
¼

1: Ties and dogs were banned after the 2004-2005 lockout



of the class does 2007-2008
does 2006-2007
does 2005-2006
does 2003-2004
1
Relationship between Points (Not
Goals) and PP%


Linear Regression Model
Remember, a team is awarded 2 points if
they win, 1 point if they lose in
OT/shootout, and 0 points if they lose in
regulation
2008-2009
Fitted Line Plot
P = 44.03 + 1.975 PP%
110
S
R-Sq
R-Sq(adj)
100
P
90
80
70
60
50
12
14
16
18
20
PP%
22
24
26
10.9875
25.6%
22.9%
2007-2008
Fitted Line Plot
P = 57.51 + 1.896 PP%
120
S
R-Sq
R-Sq(adj)
110
P
100
90
80
70
15.0
17.5
20.0
PP%
22.5
25.0
9.68195
16.8%
13.9%
2006-2007
Fitted Line Plot
P = 32.40 + 3.356 PP%
120
S
R-Sq
R-Sq(adj)
110
100
P
90
80
70
60
50
12
14
16
18
PP%
20
22
24
13.6885
30.6%
28.1%
2005-2006
Fitted Line Plot
P = 6.66 + 4.802 PP%
130
S
R-Sq
R-Sq(adj)
120
110
P
100
90
80
70
60
50
12
14
16
18
PP%
20
22
12.4745
44.9%
43.0%
2003-2004
Fitted Line Plot
P = 50.53 + 2.204 PP%
110
S
R-Sq
R-Sq(adj)
100
P
90
80
70
60
10
12
14
16
PP%
18
20
22
14.9092
12.5%
9.4%
A Note on Our Failure


We tried to compare shooting percentages
on and off a power play, but the NHL is
skimpy (unhelpful) when it comes to data
Shots on goal during power plays were
unavailable
Conclusions




In general, better teams
tend to be better at
converting power plays
But a direct correlation
was not to be found
More sophisticated data is
needed from the league
We attribute this lack of
data to lack of dogs in
the NHL
Citations


http://en.wikipedia.org/wiki/Power_play_(
sport)
Tommy, AJ, and Catherine’s Presentation
(for definition of 2-sample t tests)
Food For Thought…


Why is this dog smoking and playing hockey?
What statistical model would most accurately
predict this behavior?