The Brian Burke Theory

“The only way you cannot have salary cap problems is to have a crappy team.” – Brian Burke, Executive Vice President and General Manager of the Anaheim Ducks


1. Introduction

The monopsonist National Hockey League is the sole buyer of the elite hockey talent residing in North America and also draws the elite from the rest of the world.[1] The salary cap has been put in place for two main reasons: to maintain a constant distribution of revenue between owners and players and to restore some competitive balance in the league. [2] There can no longer be a New York Yankees of hockey franchise, like the 2002 Detroit Red Wings who purchased a Stanley Cup with estimated total player salaries of $70 million, because of the cap system.[3] In theory, given the salary cap in the NHL, teams with higher average playing ability will be near the top of the league in points and will have lower than average marginal factor costs. Because there is a large cap range (between $28 million and $44 million) there tend to be large differences in salaries.[4] As such, those teams that spend more will tend to have more success.

It is not an understatement to call the general manager of a National Hockey League franchise the most valuable person on the team. It is the job of the general manager to compile a team whose on-ice product attracts fans and in turn helps maintain financial stability for the franchise.[5] Demand for hockey is dependent on the on-ice product. Put together a winning team that is entertaining to watch and the revenue will flow (see Figure B1). Some teams like the Toronto Maple Leafs have a relatively inelastic demand curve for gate receipts, but the majority of teams rely on wins increase these and other revenues (Figure B2).[6]

The addition of the salary cap in the National Hockey League has made the job of the General Managers of the league much more complex. General Managers must contemplate a strategy that will attract the best players while at the same time maintaining their budget. Brian Burke, Executive Vice President and General Manager of the 2007 Stanley Cup Champion Anaheim Ducks is quoted as saying, “in a cap system it’s really hard to repeat because other teams get to reload quicker.”[7] He refers to the fact that at the end of a season, some players become restricted and unrestricted free agents, and due to the cap restraint other teams are able to outbid a successful team for their talent, and so it is difficult to repeat as Stanley Cup Champions. An example of this is the signing of restricted free agent Dustin Penner by Kevin Lowe of the Edmonton Oilers. The Oilers were forced to forgo their first, second and third round draft picks in 2008 because of this signing.[8] General Managers and Owners have argued that since the inception of free agency, players’ salaries are higher than their marginal physical products.[9]

The hypothesis that playing ability and marginal factor costs are positively and negatively related to team points earned respectively is examined using diverse measures of playing ability for forwards, defensemen and goaltenders. Many distinct statistics were compiled for this study but only a select few were used. Players are measured by their contribution to team wins. Both forwards and defensemen are rated based on even-strength, short-handed, power-play, game-winning and game-deciding goals, the latter referring to winning shootout goals and are independent of a players total goals statistic. Power-play, short-handed and even strength assists as well as plus/minus, penalty minutes, blocked shots and an assortment of team statistics were taken into account in these calculations (see Appendix A). Forwards are additionally measured on face-off percentage and percentage of team face-offs taken as this contributes to team possession. All statistics are measured on a per-game basis.

Statistics such as hits, giveaways and takeaways were considered, but were deemed to be poor measures of a player’s contribution to goals for or against his team. There was no insightful way to directly apply these statistics to goals for or against contribution.

Goaltenders are rated on goals against average, save percentage and points earned for their team while in goal. Goaltenders are compared against league-wide averages on a game by game basis.

Some players thrive in penalty killing and other defensive situations while others are more offensively minded. Varied measures have been taken into consideration to attempt to credit all types of players and not just point-getters. Some players are important to their teams because they have the ability to gain momentum through a fight, a big hit or drawing a penalty. These are very difficult things to measure and will be ignored in this study.

2. Playing Ability and Marginal Factor Cost

The intuition behind this measure of playing ability is as follows; a player is as good as the expected number of goals he will contribute to per game. What this means is that if a player contributes negatively to his team, this will show up in his playing ability. His playing ability is a function of his skill and the skill of his teammates.

The formula takes the sum of a player’s goals (non power play or game winning), assists (non power play), and +/- per game into consideration. Next, it subtracts the product of average number of penalties by the player per game and one minus the net team penalty kill percentage for that season to calculate goals against caused by the player. The player’s power play contributions are calculated by taking the ratio of his power play goals and assists (per game) to the team’s average net power play goals per game and added to the equation. The next crucially important calculation measures the player’s clutch performances (not ability). In other words, the number of game winning goals or game deciding shootout goals the player has scored and averaging this on a per game basis. For example, if a player has played in 10 games and has 1 GWG and 2 GDGs he will have an additional 0.3 goals per game tallied for his team as a contribution to wins per game. Note that this is a somewhat questionable statistic as players may score game winning goals early in the game, and this unfortunately cannot be taken into consideration. Blocked shots are then added to this mix. The product of blocked shots per game and one minus the team save percentage is added to the total. The intuition behind this is that if a team stops 90% of their shots, then a player who blocks 25 shots in a season has saved approximately 2.5 goals against.

This total will demonstrate the player’s total positive or negative contribution to team goals per game. Because this number can be negative, the average goals needed for a .500 record, approximately 2.7, is added to this number such that a players’ marginal factor cost is and playing ability can’t be negative.[10] This final tally is multiplied by 82; the number of games per team in a season, and this is the player’s playing ability.

Additionally add the product of face-off win percentage minus 0.5, and percentage of team face-offs taken prior to multiplying by 82 for forwards. This is just a small measure of contribution to team possession of the puck. Face-offs are an important part of the game and can have a large impact on goals scored in a game.[11]

Goaltenders playing ability is a function of many variables. The goaltender’s goals against average is subtracted from the league goals against average. Instinctively, if a goaltender’s goals against average is less than the league average he is more likely to win. That number is then added to the goaltenders save percentage minus the league average save percentage, and again if the goaltender stops more than the average goaltender in the league he has a higher winning probability. Two times the goaltender’s total wins plus his overtime losses are divided by 82 games calculating the average points earned for his team per team game played. This number is added to the others and 2.7 is added to the total to keep consistent player measurements. This number is multiplied by 82 games and this represents the goaltenders playing ability.

The calculation of estimated average playing ability separated forwards, defensemen and goaltenders from each other and weighted them accordingly. A team is assumed to have twelve forwards, six defensemen and one goaltender play during a game. Average playing ability of the forwards was multiplied by twelve and the average playing ability of defensemen was multiplied by six. The goaltenders playing ability was calculated as a weighted average, taking into consideration games played. The summation of these weighted average estimated playing abilities were taken for each team representing estimated total playing ability. An example calculation is presented in Appendix C.

Marginal factor cost (or price per unit of playing ability) is calculated by taking the team salaries and dividing them by each team’s respective estimated total playing ability (See Appendix A for all equations).


3. Considerations


Some unobserved data not taken into account must be considered. Offsetting penalties, including fighting majors, where both teams get penalties and neither has a power play are not recorded. This could factor as a slight negative impact on a player’s estimated playing ability. There are players acquired mid-season in trades or waivers who didn’t play the full season for their team. Some players were injured for long periods of time during the season and teams were forced to dress players normally playing in the minors, and this could have significantly affected their team standings. If a player or goaltender played less than twenty games, there stats were disregarded as they were considered outliers. Players with missing data were also disregarded.
Those players who are the elite on the ice generally play against the elite of the other team. This will skew their statistics but unfortunately can’t be taken into account here. Ice time, a statistic that is readily available, will not be considered in this study since better players will generally play more. A measure called production, which measures the average amount of time on ice between points for a particular player has not been taken into account as it does not measure goals against while the player is on-ice. Opponents the team faces are not considered. Age is not taken into account but must be reflected on for salary reasons since young players generally make less.[12] The Pittsburgh Penguins had a much younger team than most and therefore lower salaries, but were still a very competitive team.[13]

Marginal factor costs of individual players were calculated and recorded but will not be taken into consideration as a weighted average estimate of overall playing ability and total player salaries was a more promising estimate due to differing sample sizes.

Some notable assumptions must also be considered. All statistics are from the 2006-07 NHL season. Games played by a goaltender are calculated as the summation of wins, losses and overtime losses and will not include those games in which a goaltender was not credited with a win, a loss or an overtime loss. Those are games where either the goaltender is pulled and a lead change occurs or the goaltender replaces another goaltender and no lead change occurs.

Another assumption to consider is the calculation of goals per game to end with 50% of the points (a .500 record). The difference between .500 and a team’s actual point percentage was calculated and multiplied by their recorded average goals per game to find their goals per game above .500. Their average goals per game for .500 are then calculated by subtracting this value from their actual average goals per game. The league average of these numbers was calculated and estimated at approximately 2.7. Something to consider is that since three-point games are available given single points for overtime/shootout losses, the summation of point percentage above .500 is greater than zero.

The salary cap midpoint was then taken into account as a separate measure. The range of values that could be spent on players in the 2006-07 NHL season was between $28,000,000 and $44,000,000, and the midpoint was $36,000,000.[14] New marginal factor costs were calculated taking the midpoint as team salary. This had very obviously the same results as the average playing ability. Graphs showing the results can be found in Appendix D.

4. Results

The results of the study were somewhat as predicted. Those teams with higher average playing abilities tended to be near the top of the league in points (see Figure D1). An R2 coefficient of 0.673 was calculated using Microsoft Excel. Marginal factor costs showed a slight negative correlation with team points (R2 = 0.078) as hypothesized (Figure D2). This result was slightly skewed since the Washington Capitals spent the least amount of any team in the league by a margin of $5,624,894.[15] Removing them as an outlier found more promising results with respect to marginal factor cost as can be seen in Figure D5 (R2 = 0.235).

The teams that spent more did tend to have more points, but this was not an overly significant finding (Figure D3, R2 = 0.109). Setting the salaries of each team equal to the midpoint of the salary cap brought on expected results. There was a tendency for teams with lower marginal factor costs to have more points (Figure D4, R2 = 0.684). This is virtually the same as comparing average playing abilities to points.

5. Conclusions and Future Considerations

This study has shown that given the salary cap, marginal factor costs will have a slight negative correlation with points earned. Also, average playing ability is a significant estimate of points for a team in the National Hockey League. This can be explained intuitively; since playing abilities have been calculated ex-post to the season, teams with higher average playing abilities will have done better than other teams or their numbers would have been lower with respect to points earned. Since goals for and against are accounted for, playing ability should be a good indicator of points.

Future studies should be done using samples from other seasons. All post-lockout era data should be compared with pre-lockout data to see if the salary cap has had an impact on competition as was intended. It would be shrewd to assume that playing ability measures would be positively correlated to points, however due to the large differences in salaries in the pre-lockout era; it is more than likely marginal factor costs were be positively correlated with points given unlimited player salaries and bidding wars due to free agency.[16]

[1] National Hockey League, http://en.wikipedia.org/wiki/National_Hockey_League
[2] Salary Cap, http://en.wikipedia.org/wiki/Salary_cap#Salary_cap_in_the_NHL
[3] Detroit Red Wings, http://sportsillustrated.cnn.com/hockey/news/2002/09/11/2002_redwingspreview/
[4] NHL Salary Cap Hits $44 Million, http://www.cbc.ca/sports/story/2006/06/27/nhl-salarycap.html
[5] Brian Burke, http://www.speakers.ca/burke_brian.aspx
[6] In class-discussion, February 28, 2008, Economics of Sports, Professor John Palmer
[7] The Element: Brian Burke Interview, http://ducks.nhl.tv/team/console?type=fvod&id=5299
[8] Burke Still Smarting From Penner Deal, http://www.tsn.ca/nhl/news_story/?ID=229580&hubname=
[9] Leeds & von Allmen The Economics of Sports, 3rd Edition, page 265
[10] We ignore this step when calculating marginal revenue product as a player can contribute negatively to team wins, it is important in calculating a player’s price per unit of playing ability as that number can’t be negative.
[11] Face-Offs, http://www.behindthenet.ca/faceoff.html
[12] NHL, NHLPA agree to tentative deal, http://www.cbc.ca/sports/story/2005/07/13/nhl050713.html
[13] 2006-07 Pittsburgh Penguins season, http://en.wikipedia.org/wiki/2006-07_Pittsburgh_Penguins_season
[14] NHL salary cap hits $44 million, http://www.cbc.ca/sports/story/2006/06/27/nhl-salarycap.html
[15] 2006-07 (Unofficial) Final Cap Numbers, http://www.nhlscap.com/2006-07_cap.htm
[16] Leeds & von Allmen The Economics of Sports, 3rd Edition, page 265

References

National Hockey League. (2008). STATS. Retrieved March 14 – 20, 2008, from World Wide Web: http://www.nhl.com/nhlstats/app

Sports Matters. (Tuesday, January 17, 2006). NHL Team Salaries. Retrieved March 18, 2008, from World Wide Web: http://sportsmatter.blogspot.com/2006/01/nhl-team-salaries.html

TSN. (2008). Payroll Commitments. Retrieved March 18, 2008, from World Wide Web: http://www.tsn.ca/

Anaheim Ducks. (2008). State of the Franchise – Brian Burke. Retrieved April 1, 2008, from World Wide Web: http://ducks.nhl.tv/team/console

Wikipedia. (2008). National Hockey League. Retrieved April 1, 2008, from World Wide Web: http://en.wikipedia.org/wiki/National_Hockey_League

Wikipedia. (2008). Salary Cap. Retrieved April 1, 2008, from World Wide Web: http://en.wikipedia.org/wiki/Salary_cap#Salary_cap_in_the_NHL

Sports Illustrated. (2002). Detroit Red Wings. Retrieved April 1, 2008, from World Wide Web: http://sportsillustrated.cnn.com/hockey/news/2002/09/11/2002_redwingspreview/

CBC. (June 27, 2006). NHL Salary Cap Hits $44 Million. Retrieved April 1, 2008, from World Wide Web: http://www.cbc.ca/sports/story/2006/06/27/nhl-salarycap.html

Speakers’ Spotlight. (2008). Brian Burke, Anaheim Mighty Ducks Executive Vice President & GM. Retrieved April 1, 2008, from World Wide Web: http://www.speakers.ca/burke_brian.aspx

Anaheim Ducks. (2008). The Element: Brian Burke Interview. Retrieved April 1, 2008, from World Wide Web: http://ducks.nhl.tv/team/console?type=fvod&id=5299

TSN. (February 13, 2008). Burke Still Smarting From Penner Deal. Retrieved April 1, 2008, from World Wide Web: http://www.tsn.ca/nhl/news_story/?ID=229580&hubname=

Leeds and von Allmen. (2008). The Economics of Sports, 3rd. Edition. Pearson Education, Inc. (pg. 265).

CBC. (July 14, 2005). NHL salary cap hits $44 million. Retrieved April 1, 2008, from World Wide Web: http://www.cbc.ca/sports/story/2006/06/27/nhl-salarycap.html

NHLSCAP.com. (2008). 2006-07 (Unofficial) Final Cap Numbers. Retrieved April 1, 2008, from World Wide Web: http://www.nhlscap.com/2006-07_cap.htm

Behind the Net.ca. (2005). Face-Offs. Retrieved April 1, 2008, from World Wide Web: http://www.behindthenet.ca/faceoff.html