1. Shortening the NBA Game:
A Look at the Effects on Competitive Balance
Jordan Aronson
Tanya Downey
Caleb Engelbourg
Michael P. McGrath
David S.K. Schneider
University of Massachusetts Amherst,
Mark H. McCormack of Sport Management
Abstract:
Maintaining competitive balance is an issue that runs through all sport leagues, but in no league
is it more prevalent than the National Basketball Association (NBA). The NBA has looked at
shortening the length of a season and also explored the idea of shortening the length of each
game for a variety of reasons, including increasing competitive balance. A simulation provides
an effective approach to examine the impact that shortening the game length would have on the
NBA’s competitive balance. In simulating the NBA season in this study, Net Points Per
Possession (NPPP) was used as a variable to assess team strength. An NPPP value was
randomly assigned to each team in the league, and then simulated through an NBA season with
game lengths of 48-minutes, 44-minutes, and 40-minutes. The results showed the decrease in
game-length resulted in more league wide parity, with significant changes in win totals for the
best and worst teams in the league. Other implications of shortening the length of the game and
possibilities for further research are discussed.

2. 1. Introduction
“The special problem for sports leagues is the need to establish a degree of competitive balance
on the field that is acceptable to fans” (Fort & Quirk 1995). The special problem for the National
Basketball League (NBA) is that the relative standard deviation of basketball is persistently
higher than any of the other three big four sports in North America – Major League Baseball
(MLB), National Football League (NFL), and the National Hockey League (NHL) – (Rockerbie
2014). Theory and empirical research suggest that fans expect a degree of uncertainty in the
outcome of the game and fairness in the rules and conditions of the competition (Zimbalist
2003). Thus, competitive balance is important to any league’s success because of its effect on
league economics; more fans translates to higher revenues.
With the desire for more competitive balance from a league standpoint, there have been talks by
players, coaches, and owners about shortening the amount of games played in a season and
shortening the length of each game. During the 2011 lockout shortened season, the 66 game
schedule provided good empirical data for the NBA to analyze. Although shortening the season
is one option, “cutting the number of games wouldn’t inject as much unpredictability as a
minutes reduction” (“What’s the” 2014). A minutes reduction would allow for more randomness,
as the better team would have less time to allow their superior talent to win the game.
The NBA has already looked at shortening the length of games, experimenting with both 44-
minute games and 40-minute games. This year, the NBA had their first experiment with a 44-
minute preseason game between the Brooklyn Nets and Boston Celtics. Though the transition
to 44-minute games is not imminent, this experiment shows that the NBA is willing to collect
data to see the possible effects on the game. NBA President of Basketball Operations Rod
Thorn told USA Today that the conversation about the length of games was a talking point at
the most recent owner’s meeting, indicating that the coaches and owners would like to consider
the idea of reducing game length “as a means to reduce minutes for some players and maybe
improve the flow of the game” (Zillgitt 2014).
Another outlet where the NBA can observe the effects of a shortened game is in the Liga
Nacional de Basquete, the top Brazilian basketball league with which the NBA recently formed a
partnership. The partnership will send both money and employees from the NBA for at least two
years, giving the NBA another way to expand globally. This will allow the league to gauge
interest in unifying the rules of the International games and the NBA (“The Case” 2014). The
NBA has already considered adopting other International Basketball Federation (FIBA) rules
including the basket interference rule and the ban of live-ball timeouts late in the game. The 40-
minute game would align the length of the NBA game with that of FIBA games, allowing for
more continuity between International play and the NBA, which aligns with the NBA goal of
expanding the league brand globally.
There are other reasons besides competitive balance that lend way to the NBA exploring a
shorter game length. With a recent injury to Paul George during the summer and Kevin Durant
dropping himself from the USA summer roster due to energy concerns, there has been

3. increased discourse on the subject of player exhaustion due to the year-long playing schedule
for elite players.
With ongoing talks about the option of shortening game length, a simulation is an applicable
measure to see the effects of a shorter game on competitive balance. The Bradley-Terry is a
common way to simulate paired comparisons such as sports games and is an effective way to
simulate many seasons (Albert & Marchi 2014). This paper utilizes a simulation of the NBA
season at three different game lengths (48, 44, and 40 minutes) to analyze the effects that a
shorter game would have on competitive balance.
2. Methods
This paper investigates the length of NBA games with respect to potential remedies for the
competitive balance issue. Currently, the NBA has an 82 game season. Each game is four
quarters (overtime is excluded for this study), with each quarter lasting 12 minutes. This paper
sets out to find the differing effects on competitive balance for 12-minute quarters (48-minute
game), 11-minute quarters (44), and 10-minute quarters (40); respectively 3936, 3608, and
3280 minutes per season.
2.1 Selecting the league and league schedule
The original framework of a Bradley-Terry model was modified to fit the needs of the simulation.
Instead of using the standard Bradley-Terry model variable, team talents, Net Points Per
Possession (NPPP) was replaced in the creation of the league distribution. NPPP allows for a
simulation based on possession instead of just team talents, enabling the modification of game
length for different seasons. NPPP is calculated by the formula:
Where offensive efficiency is equal to a team’s points per 100 possessions, and defensive
efficiency is equal to a team’s points against per 100 possessions.
Calculations to find NPPP were taken from the 2002-2003 season to the 2013-2014 season.
This worked well as a cutoff because the NBA changed the 10-second backcourt violation to
eight seconds in 2001. This rule change helped to speed up the NBA game and had an effect
on the pace factor for the league. NPPP has proven to be a good indicator of winning
percentage over the 12 seasons of NBA data collected. There is a 0.957 correlation between
NPPP and winning percentage, indicating that NPPP is a good measure of team strength. This
relationship is also shown in Graph 1.

4. Graph 1. NBA NPPP vs Winning Percentage (2002-2013)
When selecting the league, the averages of 1,000 random draws from the normal distribution
were taken.
Here the mean is 0 because the net score within the league is a zero sum game, and our
standard deviation is based on NPPP for team season long averages. Teams were randomly
assigned one of the thirty NPPP values from the draw and the league composition stays
constant for each season’s simulation at each game length. The results of the league are shown
in Table 1, below.

5. Table 1. League NPPP Distribution
Each teams schedule was modeled after the 2013-2014 NBA schedule, keeping the schedule
constant for each simulation at every game length in order to accurately compare the
competitive balance effects of playing the same games for each team. This schedule is based
on four games against the other four division opponents, four games against six out-of-division
conference opponents, three games against the remaining four conference teams, and two
games against teams in the opposing conference.
2.2 Simulating the Season
The simulation has three key parameters: Pace, mean difference in NPPP, and standard
deviation of NPPP. These three key parameters are used in a normal distribution to determine a
winner of each game. The pace factor determines the number of possessions per game, which
in this simulation is defined as a possession for both the home and away team. The difference
Team% League% NPPP%
Washington* East* ,0.112*
Golden*State* West* ,0.083*
New*Orleans* West* ,0.070*
Atlanta* East* ,0.060*
Minnesota* West* ,0.052*
L.A.*Lakers* West* ,0.046*
Cleveland* East* ,0.040*
Memphis* West* ,0.034*
Dallas* West* ,0.029*
Miami* East* ,0.024*
Toronto* East* ,0.020*
Oklahoma*City* West* ,0.015*
CharloLe* East* ,0.011*
Houston* West* ,0.006*
Philadelphia* East* ,0.002*
Sacramento* West* 0.002*
New*York* East* 0.006*
Portland* West* 0.011*
L.A.*Clippers* West* 0.015*
Milwaukee* East* 0.020*
Utah* West* 0.024*
Detroit* East* 0.029*
Chicago* East* 0.035*
Brooklyn* East* 0.040*
Indiana* East* 0.046*
San*Antonio* West* 0.053*
Boston* East* 0.061*
Orlando* East* 0.070*
Denver* West* 0.082*
Phoenix* West* 0.111*

6. in the NPPP is the mean of the normal distribution and the standard deviation of NPPP on a
possession basis is the standard deviation.
The pace factor is found from the 12 seasons of data collected, from which an overall league
mean and standard deviation was found. From this, a random number of possessions for each
game was calculated for the simulation. The formula for game pace is below.
For the mean of the game result simulation, the NPPP of the away team is subtracted from the
NPPP of the home team. These values were taken from the simulated league draw from Table
1.
To calculate the standard deviation of NPPP on a possession basis, it was important to look at
play-by-play data. The play-by-play data for this study was taken from the 2009-2010 season,
and the standard deviation was calculated from this data. It is assumed that the standard
deviation of points scored on a possession does not fluctuate a significant amount season-to-
season.
These three parameters were used in a normal distribution to sum up n random points, where n
is the pace factor for this specific game. If the sum is greater than zero, the home team is
determined the winner, and if the sum is less than zero, the away team is determined the
winner. The equation for the simulation of one game is shown below.
This method is applied to each game of the simulation. The season was simulated 1,000 times
at each game length. In order to simulate different game lengths, the mean pace was scaled by
a factor of 11/12 and 10/12 for a 44-minute and 40-minute game, respectively.
3. Results
3.1 Win Total Differentials
The results of the simulation for average win totals by game length can be found in Table 2, and
the average difference for the different game lengths can be seen in Table 3.
Based on the simulation, there was more competitive balance in the shortened games. The 5
teams with the lowest win totals for the 48-minute game season saw their win totals increase by
an average of 0.68 wins per season over the 44-minute game simulation, while every team with

7. a negative NPPP saw an average increase in wins of 0.35 in the 44-minute game simulation.
The top five teams in terms of wins during the 48-minute simulation saw their average win total
decrease by 0.62 wins per season in the 44-minute game simulation, and every team with a
positive NPPP had an average decrease in wins of 0.35 per season.
The 5 teams with the lowest win totals for the 48-minute game season saw their win totals
increase by an average of 1.12 wins per season over the 40-minute game simulation, while
every team with a negative NPPP saw an average increase in wins of 0.58 in the 44-minute
game simulation. The top five teams in terms of wins during the 48-minute simulation saw their
average win total decrease by 1.04 wins per season in the 40-minute game simulation, and
every team with a positive NPPP had an average decrease in wins of 0.58 per season.
Table 2. Average Win Totals by Game Length
Team% NPPP%
48%min%Wins%
Avg%
44%min%Wins%
Avg%
40%min%Wins%
Avg%
Washington* +0.112* 20.04* 20.98* 21.65*
Golden*State* +0.083* 25.03* 25.58* 25.92*
New*Orleans* +0.070* 27.18* 27.69* 28.13*
Atlanta* +0.060* 29.01* 29.59* 30.01*
Minnesota* +0.052* 30.67* 31.53* 31.83*
L.A.*Lakers* +0.046* 32.11* 32.58* 32.84*
Cleveland* +0.040* 32.98* 33.30* 33.65*
Memphis* +0.034* 34.49* 34.70* 35.07*
Dallas* +0.029* 35.71* 35.76* 35.84*
Miami* +0.024* 35.92* 36.06* 36.30*
Toronto* +0.020* 36.90* 37.26* 37.09*
Oklahoma*City* +0.015* 37.99* 38.23* 38.14*
CharloKe* +0.011* 38.96* 39.02* 39.04*
Houston* +0.006* 39.89* 40.12* 40.06*
Philadelphia* +0.002* 40.41* 40.19* 40.40*
Sacramento* 0.002* 41.94* 42.08* 41.89*
New*York* 0.006* 42.46* 42.02* 42.02*
Portland* 0.011* 42.99* 43.09* 43.16*
L.A.*Clippers* 0.015* 44.46* 44.27* 44.08*
Milwaukee* 0.020* 44.55* 44.34* 44.36*
Utah* 0.024* 46.06* 45.68* 45.68*
Detroit* 0.029* 46.69* 46.54* 46.33*
Chicago* 0.035* 47.94* 47.51* 47.44*
Brooklyn* 0.040* 48.41* 48.19* 47.95*
Indiana* 0.046* 49.73* 49.34* 48.88*
San*Antonio* 0.053* 51.66* 51.18* 50.96*
Boston* 0.061* 52.87* 51.97* 51.64*
Orlando* 0.070* 54.38* 54.03* 53.59*
Denver* 0.082* 56.97* 56.31* 55.96*
Phoenix* 0.111* 61.60* 60.88* 60.12*

8. Table 3. Average Win Differential
3.2 Statistical Significance
Using a paired t-test, each of the bottom and top eight teams in the league had a statistically
significant difference in average win totals at both the 44 and 40-minute game lengths. This
analysis was completed with an alpha level of 0.01.
4. Discussion
Based on the results, shortening the NBA game would have a significant effect on competitive
balance in the league. For the shortened games, there was compression for the top and bottom
five teams in the league of an average of .65 and 1.08 wins on average for a 44 and 40-minute
game, respectively. While this number may seem relatively small compared to an 82-game
schedule, any increase in parity for the NBA could help ticket sales, TV ratings, and the
common sentiment that the NBA lacks competitive balance.
Team% NPPP%
48%vs%44%min%
Win%Difference%
48%vs%40%min%Win%
Difference%
Washington* +0.112* 0.94* 1.61*
Golden*State* +0.083* 0.55* 0.89*
New*Orleans* +0.070* 0.51* 0.95*
Atlanta* +0.060* 0.57* 1.00*
Minnesota* +0.052* 0.86* 1.16*
L.A.*Lakers* +0.046* 0.47* 0.73*
Cleveland* +0.040* 0.32* 0.67*
Memphis* +0.034* 0.22* 0.59*
Dallas* +0.029* 0.05* 0.12*
Miami* +0.024* 0.14* 0.38*
Toronto* +0.020* 0.36* 0.19*
Oklahoma*City* +0.015* 0.24* 0.15*
CharloKe* +0.011* 0.06* 0.08*
Houston* +0.006* 0.23* 0.17*
Philadelphia* +0.002* +0.22* +0.01*
Sacramento* 0.002* 0.14* +0.05*
New*York* 0.006* +0.44* +0.45*
Portland* 0.011* 0.10* 0.17*
L.A.*Clippers* 0.015* +0.19* +0.38*
Milwaukee* 0.020* +0.21* +0.19*
Utah* 0.024* +0.38* +0.38*
Detroit* 0.029* +0.14* +0.35*
Chicago* 0.035* +0.43* +0.50*
Brooklyn* 0.040* +0.22* +0.46*
Indiana* 0.046* +0.39* +0.86*
San*Antonio* 0.053* +0.48* +0.70*
Boston* 0.061* +0.89* +1.23*
Orlando* 0.070* +0.36* +0.80*
Denver* 0.082* +0.66* +1.01*
Phoenix* 0.111* +0.72* +1.48*

9. The simulation shows that the smaller the sample size, in this case minutes per game, the more
unpredictability in the game outcome. Even though our results may not have shown a large
difference in win totals, the amount that does differ could have a major impact on which teams
qualify for playoff contention, the seeding of playoff teams, and draft position. The loss of one
win for top teams could impact playoff seeding, as there were three pairs of playoff teams in the
2013-2014 season that ended with the same record. This included the 4 and 5 seed of the
Western conference, where a one-win difference would have meant a switch in home court
advantage. While shortening the game will not completely solve the competitive balance issue
in the NBA, it can help to flatten the standings, allowing more randomness on a game-to-game
basis.
The development of more league parity could also have a significant economic impact. In his
article Diamond Dollars: The Economics of Winning in Baseball, Vince Gennaro discusses how
winning baseball games in the regular season can yield larger revenue streams. He writes, “At
various levels of competitiveness—conveniently measured by annual win totals—fans will
allocate varying amounts of time and money to support their favorite ball club. The more
successful the team, the greater the fan and sponsor support, and hence the higher the revenue
totals” (Gennaro 2007). As the league playoff picture remains clouded for longer due to the
increased uncertainty of games, the bottom tier teams should see a positive economic impact
due to their lengthened period of playoff contention, while the top tier teams see a relatively
smaller negative impact, since after reaching the win total needed to qualify for the playoffs,
they have already accomplished their regular season goal.
The NBA must consider other factors that may be impacted by the implementation of a shorter
game length. The NBPA has raised concerns over the current length of the NBA season, citing
player health issues. Shortening game length would also accomplish a reduction in playing time,
with a 44-minute game constituting a 6.83 game decrease in season length, and a 40-minute
game would be a 13.67 game reduction. The belief is that shorter games would limit the wear
and tear on players, although the question to consider is whether shorter games will actually
limit the number of minutes played per game for the premier players in the league.
An additional appeal of the shorter game is fitting an NBA game into the coveted two-hour
television time window. The experimental Celtics-Nets game lasted only 1 hour and 58 minutes
(Mazzeo 2014), which is 17 minutes less than the current average NBA game length (Keh
2014). As a result of the shorter quarters, there were two less mandatory television timeouts,
one each in the second and fourth quarters. In a USA Today article, Jeff Zillgitt wrote that he
believes fitting an NBA game into a 2-hour timeslot would result in a better fan experience at the
arena and at home (Zillgitt 2014). Enhanced fan experience may lead to an increase in both
attendance and television viewership. Increasing television viewership is important because the
NBA recently signed a new 9-year, $2.66 billion per year television rights deal with ESPN and
TNT. As noted earlier in Zach Lowe’s article, a reduction in game minutes should help drive ad
revenue due to the scarcity of advertisement space (“The Case” 2014).

10. 5. Conclusion
Through our study, we were able to determine that the NBA can reasonably expect to see
increased competitive balance throughout the league should they choose to implement shorter
game lengths than their current 48-minute structure. As previously mentioned, increased parity
throughout the league could serve as an economic boost both to individual franchises, and the
league as a whole.
However, the implementation of a shorter game length would not be simple. There are a
number of other factors to be considered when examining the NBA game length, some of which
were discussed in this paper. Other implications include the impact on NBA statistics and
historical comparisons, the impact on the NBA Draft lottery, and playing time adjustments for
both premier and bench players (as well as the resulting salary ramifications of adjusted playing
time and subsequent production).
Further research on this topic can be done in analyzing the impact of shorter games on overtime
games and on playoff series outcomes. There is a possibility of more overtime games with
fewer possessions in the game, something not considered in this study. Additionally, playoff
series may become more competitive, increasing the probability of upsets. More work in
examining these issues would be useful to understanding the full impact of a shortened game.

11. Works Cited
Basketball Reference. N.p., n.d. Web. 11 Dec. 2014. <http://www.basketball-reference.com>.
ESPN. N.p., n.d. Web. 11 Dec. 2014. <http://espn.go.com>.
Fort, Rodney, and James Quirk. "Cross-subsidization, Incentives, and Outcomes in Professional
Team Sports Leagues." Journal of Economic Literature XXXIII (September 1995): 1265-
99. JSTOR. Web. 10 Dec. 2014.
Gennaro, VInce. "Diamond Dollars: The Economics of Winning in Baseball (Part 1)." The
Hardball Times. N.p., 22 Mar. 2007. Web. 11 Dec. 2014.
Keh, Andrew. "N.B.A. to Experiment With Shorter Game in Nets Exhibition." New York Times
[New York] 14 Oct. 2014: B13. New York Times. Web. 11 Dec. 2014.
Lowe, Zach. "The Case Made For The 40-Minute Game." Grantland. N.p., n.d. Web. 10 Dec.
2014.
Lowe, Zach "What the NBA’s Partnership With Brazil Means for the Future of the 44-Minute
Game." Grantland. N.p., 27 Oct. 2014. Web. 10 Dec. 2014. <http://grantland.com/the-
triangle/44-minute-game-brazil/>.
Marchi, Max, and Jim Albert. Analyzing Baseball Data With R. N.p.: n.p., 2014. Print.
Mazzeo, Mike. "Johnson: Game 'pretty much same.'" ESPN New York. ESPN, 20 Oct. 2014.
Web. 11 Dec. 2014.
"NBA’s 44-minute game clocks in at under 2 hours, players barely notice." ProBasketball Talk.
NBC Sports, 19 Oct. 2014. Web. 11 Dec. 2014.
Rockerbie, Duane W. "Exploring Interleague Parity in North America: The NBA Anomaly."
Journal of Sports Economics (2014): 1-16. SAGE journals. Web. 10 Dec. 2014.
Zillgitt, Jeff. "NBA to experiment with 44-minute preseason game." USA Today. N.p., 14 Oct.
2014. Web. 10 Dec. 2014. <http://www.usatoday.com/story/sports/nba/2014/10/14/nba-
44-minute-game-experiment-nets-celtics/17246813/>.
Zimbalist, Andrew. "Competitive Balance Conundrums: Response to Fort and Maxcy's
Comment." Journal of Sports Economics 4.2 (May 2003): 161-63. SAGE journals. Web.
8 Dec. 2014.