Tuesday, December 14, 2010

Statistical Analysis: Do the refs favor the Pittsburgh Penguins?

(Ed. Note: Angela Havely is a self-identified "stats junkie" who is a corporate attorney when she's not crunching hockey numbers. "Over the past year, in my spare time, I've used available statistics to investigate a number of claims found on the Puck Daddy blog (and its comment sections), but I've kept the results to myself. I've been content just knowing things others don't," she wrote. "Perhaps inspired by Yahoo's request for contributions, however, I wrote an easy-to-read but detailed summary of my latest findings. My research focused on claims I've seen repeated time and time again about the Pittsburgh Penguins and their allegedly cozy relationship with officials."

This lengthy, academic study makes for interesting Sunday coffee reading and is something we'd love to see more from the readers. If you're a reader with something to say, first stop should be the Yahoo! Contributor Network, where you'll get some dough for your work. For Puck Daddy contributions, please email to puckdaddyblog@yahoo.com. Thanks for reading, and here's Angela's take on the Pittsburgh Penguins and perceived ref bias.)

• • •

By Angela Havely

Contextual Summary

Much has been made in both the mainstream and sub-mainstream media this season of a perceived chasm between penalties called in the Pittsburgh Penguins' favor and penalties they've earned or deserved. Derision has come in the form of a Mike Richards diatribe, accusations that Sidney Crosby is a serial diver, and a subtle suggestion by Puck Daddy that Pittsburgh's accumulation of power-play minutes might be indicative of a team catching breaks.

It would be disingenuous, of course, to refer to the Pittsburgh Penguins as oft-maligned, given their overall popularity and the lopsided attention paid them throughout the league. However, the data through roughly one-third of the season suggests the tinfoil hats are, in this case, picking up junk signals.

Macro-Analysis

Only seven NHL teams average fifteen (15) penalty minutes or more per game. Surprisingly, four of those teams (Rangers / Flyers / Penguins / Islanders) play in the Eastern Conference's Atlantic Division. As seen in the NHL.com table below, the Pittsburgh Penguins average approximately 16.4 penalty minutes per game, while the Rangers, Flyers and Islanders average 15.6, 15.8 and 17.2 minutes per game, respectively. Only two teams, the New York Islanders and the St. Louis Blues, spend more time on the penalty kill per game than Pittsburgh, and no team has been assessed more major penalties.

Table 1:            Penalty Statistics (Available via League Website)

The Atlantic Division can lay claim to another dubious honor: one-third ownership of the league's most penalized players, including the two most distinguished in this respect, Zenon Konopka and Sean Avery (see Table 2):

(LARGE VIEW)

Table 2:            Individual Penalty Statistics (Available via League Website)


(LARGE VIEW)

Rounding out the Atlantic Division's contributions to the "dirty thirty" are Brandon Prust, Scott Hartnell, Jody Shelley, Deryk Engelland, David Clarkson, Trevor Gillies, Mike Rupp and Danny Briere. As you may have noticed, Philadelphia alone supplies 10 percent of the league's thirty most penalized players. The Penguins, Rangers and Islanders each throw in a pair, and, as usual, the Devils send Clarkson.

A simple glance reveals that A) the Penguins are one of the most penalized teams in the league and B) they play in what is easily the most penalized division.

Micro-Analysis

The team statistics above can be misleading. One or two high-penalty games can shade trends that otherwise support cyberspace's popular assertion against the Pens. Such is the danger of averages. The division sees so many penalties that hotly contested Atlantic Division contests may prop up penalty numbers while Pittsburgh gets away with proverbial murder against non-divisional opponents. Likewise, penalties called on Pittsburgh in road games may paper over unfair treatment of opponents at Consol Energy Center. One must look closely at Pittsburgh's 2010-2011 season, to date, if such possibilities are to be eliminated.

Table 3:             Regular Season Penalty Analysis of the Pittsburgh Penguins v. Opponents[1]

OPPONENT

VENUE

MINUTES (PENS)

MINUTES (OPP)

PPG (PENS)

PPG (OPP)

Philadelphia Flyers

Consol Energy Center

10

12

1

1

Montreal Canadiens

Consol Energy Center

15

17

0

0

New Jersey Devils

Prudential Center

4

12

1

0

Toronto Maple Leafs

Consol Energy Center

11

11

1

1

New York Islanders

Consol Energy Center

28

23

1

1

Philadelphia Flyers

Wells Fargo Center

19

25

2

1

Ottawa Senators

Consol Energy Center

22

34

2

1

Nashville Predators

Bridgestone Arena

15

17

0

1

St. Louis Blues

Scottrade Center

11

13

0

0

Tampa Bay Lightning

St. Pete Times Forum

26

22

0

1

Philadelphia Flyers

Consol Energy Center

20

24

0

1

Carolina Hurricanes

RBC Center

16

6

0

0

Dallas Stars

American Airlines Center

34

32

0

0

Anaheim Ducks

Honda Center

9

13

1

0

Phoenix Coyotes

Jobing.com Arena

13

15

1

0

Boston Bruins

Consol Energy Center

21

19

1

1

Tampa Bay Lightning

Consol Energy Center

16

12

0

1

Atlanta Thrashers

Philips Arena

17

17

1

0

New York Rangers

Consol Energy Center

11

23

0

0

Vancouver Canucks

Consol Energy Center

15

9

0

0

Carolina Hurricanes

Consol Energy Center

4

4

1

0

Florida Panthers

BankAtlantic Center

10

6

2

0

Buffalo Sabres

HSBC Arena

10

12

0

0

Ottawa Senators

Consol Energy Center

8

10

2

0

Calgary Flames

Consol Energy Center

15

13

1

0

New York Rangers

Madison Square Garden

18

14

0

0

Atlanta Thrashers

Consol Energy Center

9

9

0

1

Columbus Blue Jackets

Nationwide Arena

17

15

3

1

New Jersey Devils

Consol Energy Center

15

19

0

1

Toronto Maple Leafs

Consol Energy Center

48

32

0

1

Buffalo Sabres

HSBC Center

21

19

0

1

TOTALS

508

509

21

15

Penguins spent fewer minutes in the penalty box (508) than teams they competed against (509). That much is undeniable. However, a difference of one minute hardly indicates a conspiracy. (For perspective, revisit Table 1 and note the 219% swing in penalty minutes between the league's most penalized team, St. Louis, and its least, Florida.)

Through 31 games, 187 individual penalties have been assessed against Pittsburgh, while 189 such penalties have been called against opponents. The Penguins and their opposition have accumulated a nearly identical number of major penalties, 34 to 33, because fighting majors are often assigned simultaneously to one player from each team. Pittsburgh served 150 minor penalties as their rivals served 157, and they earned three game misconducts to two, respectively. Each side attempted two penalty shots, but only the Stars' Loui Eriksson was successful, in an emotional contest featuring a much-publicized fight between Sidney Crosby and Matt Niskanen.

Analysis of the Averages

The Penguins average 16.387 minutes per game in the penalty box while their opponents rack up 16.42, for a staggering 0.2 percent difference. The Pens average 1.1 major penalties per contest, as the clubs they've played tally 1.06. With respect to minor penalties, the Pens serve an average of 4.84 when they take the ice compared to 5.06 for opponents, a difference of approximately 4.35 percent. (Again, for perspective, there is a 166 percent spread between the NHL team with the fewest minor penalties [Florida] and the team with the greatest [Philadelphia].)

Pittsburgh does average less time in the box per game than its opposition...by two seconds.

As mentioned earlier, however, averages potentially camouflage all manner of evil. To more confidently study the meaning behind the numbers, further analysis is necessary:

Median Analysis

The median in a set of numbers is found by chronologically arranging the numbers and simultaneously eliminating the highest and lowest values in the sequence until the central number is uncovered. Median analysis eliminates oddball games that might profoundly affect the average. The median number of penalty minutes per game for the Pittsburgh Penguins was 15, below the average of 16.387. In the same vein, the median number of penalty minutes per game for opponents of the Pittsburgh Penguins was also 15, significantly lower than their average of 16.42. So, what's driving up the averages?

Median analysis suggests the penalty minutes per game of both Pittsburgh and Pittsburgh's opponents have been inflated by an anomalous high-penalty game or two (likely, in this case, 48 minutes against the Pens when they recently played Toronto and 34 minutes against the Ottawa Senators early in the season). When these two games are completely removed and the averages recalculated, the number of penalty minutes per game, for Pittsburgh, falls from 16.387 to 15.10. For opponents, the number is reduced from 16.42 to 15.275 over the remaining 29 games. The medians and averages are much more closely matched using the recalculation.

Pittsburgh and its competitors were penalized similarly over the first 31 contests, with each penalized a median of 15 minutes. The averages may not accurately reflect a typical Penguins game -- they are skewed upward by one or two high-penalty games.

Mode Analysis

The mode in a set of values is, quite simply, the value that appears most often. It is the number most likely to be pulled at random from the set. The mode number of penalty minutes per game for the Pittsburgh Penguins is again 15, identical to the median (15) and less than the average (16.387). In five of the games studied, the Pens were assessed 15 total penalty minutes. The mode number for competitors was only 12, closer to the median of 15 (but still lower) than the average (16.42).

Although Pittsburgh averages fewer penalty minutes overall than its competition, they were most often levied 15 minutes of penalties, while opposing teams were most often charged with only 12.

Value-by-Value Analysis

The relatively small sample size affords the luxury of a value-by-value analysis. For our purposes, the most meaningful information from each game's data is whether a) the Pens are the least penalized team in the contest and, when they are, b) is the game at home and/or against a divisional opponent? Here, things get more interesting... 

The Pittsburgh Penguins are the least penalized team in 45 percent of the games, or 14 of 31. The teams are penalized equally in four of those meetings, meaning Pittsburgh is the most penalized team in 42 percent of games, or 13 of them.

In divisional games, however, the Pens snag far fewer unfavorable calls and are the lesser penalized team in six of eight such meetings (75 percent). Only one team in the division averages more penalty minutes per game than the Penguins, regardless of whom they're playing (see Table 1), so this is something of a surprise. When Pittsburgh is the most penalized divisional team, the game is always against a New York rival (once against the Islanders at home, once against the Rangers at Madison Square Garden).

Finally, Puck Daddy once ran a piece highlighting an added benefit for NHL teams playing at home: a certain degree of amnesty from the officials. The general rule doesn't apply to the Penguins. Pittsburgh is either less likely to behave at home or is less coddled than standard home teams. In the sample studied, 16 games are played in the Steel City and in nine the Pens serve more penalty minutes or the same penalty minutes as visitors. They have no detectible home-ice advantage, at least with respect to the officials.

The Penguins are the most penalized team in 42 percent of games, and the least in 45 percent of games. They get no special treatment at home, but, when head to head, do consistently better with the refs than division rivals.

Graphic Analysis of the Data Set

Chart 1:             Range of Penalty Minutes (x-axis) / No. of Games in that Penalty Range to 12/12/10 (y-axis)


Graphic Analysis of the League

The average number of penalty minutes per game, league-wide, is 13.17, lower than Pittsburgh's average by more than three minutes. The median number of penalty minutes is 13.1, nearly identical to the average. The graphic distribution in the following chart plots a bell-like curve peaking at the group average, something akin to a statistically "normal" distribution (even given the small amount of data). The standard deviation, given the average of all 30 teams, is 2.3. It is a relatively low standard, meaning the teams very much cluster around the 13.17 average. For 68 percent of the teams, in a normal distribution, average penalty minutes per game should fall between 10.87 and 15.47, a standard deviation of one or less. For 95 percent of teams, the standard deviation should be two or less, within the range of 8.57 to 17.77 minutes. The Pittsburgh Penguins and their opponents both fall within two deviations. (If we use the recalculated averages from above, thereby eliminating the two anomalous high-penalty games, both fall within the first deviation.) In the NHL, only one team falls more than two standard deviations from average: the Florida Panthers, with its eight penalty minutes per contest.

Chart 2.             Range of Average Penalty Minutes (x-axis) / No. of NHL Teams in that Range (y-axis)

Significant Observations

Buried in the data, there is an interesting facet worth mentioning: a significant shift that occurs between the first 15 games of the season and the most recent 16. Between Game 1 of the regular season and Game 15, the Pens were the least penalized team on 10 occasions (nearly 67 percent of the first 15 games) but they struggled, finishing an unfavorable 7-7-1. They simply couldn't translate their luck with the refs into wins on the schedule. Games 16-31 saw a reversal of those numbers. The Pens were the least penalized team in only four of those contests, or approximately 25 percent of the time. The team went on a 14-1-1 tear over the latter sequence, in spite of the refs.

The breakdown of penalties in divisional games is also significant. As you may recall, the Pens were the lesser penalized team in six of eight divisional games. When one excludes the aberrational 48-minute game against Toronto, however, and instead uses the recalculated average of 15.10, suddenly three of the division's five teams accumulate more average penalty minutes than the Penguins. Perhaps it's not so surprising, then, that Pittsburgh fares better against most of the teams in its division. But it's an open question.

Tin Foil Hat Factor

Two-thirds of the season looms ahead and the sample from which to draw conclusions is small. The current data set lacks a "smoking gun" that would definitively support claims of referee favoritism, especially in the latter half of the games studied, wherein the Penguins were penalized equally to or more than the competition for three-quarters of the stretch.

If any conclusion negative to the officiating could be drawn, it is this: The Penguins were given the benefit of the doubt to open the season, but that advantage (and a largely unfruitful advantage at that) eroded as the season developed. However, even that conclusion is a difficult one to credibly draw, as several other explanations may be possible (i.e. the Penguins played a less physical game to open the season, so were less penalized but ultimately less successful / the Penguins started the year with more seasoned players but began inserting rookies and otherwise inexperienced players, due to injury, as the season progressed, etc.). I'm curious to find out what the numbers say to you, the hockey fans, and what other arguments you'd make for either side.

In short, there are a few interesting or notable statistics, but nothing outlandish or patently suspicious. The divisional results may raise a few eyebrows, and prevent a complete foreclosure on the paranoia, but a statistical argument for referee favoritism could more easily be made about the Panthers than the Penguins. After a review of the data, any claim of officiating that is ‘in the bag' for Pittsburgh deserves four out of five tinfoil hats:

[Special thanks to Ryan Havely for his contributions of both time and talent.]

[1] Italics are used to denote a game in which Pittsburgh was the more penalized team. Bold indicates a game in which Pittsburgh was the less penalized team. Division games are underlined.

Luca Sbisa Marco Scandella Nick Schaus Luke Schenn

No comments:

Post a Comment