Content uploaded by Richard Pollard
Author content
All content in this area was uploaded by Richard Pollard on Dec 11, 2015
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=rjsp20
Download by: [Cal Poly San Luis Obispo] Date: 11 December 2015, At: 10:41
Journal of Sports Sciences
ISSN: 0264-0414 (Print) 1466-447X (Online) Journal homepage: http://www.tandfonline.com/loi/rjsp20
Long-term trends in home advantage in
professional team sports in North America and
England (1876 – 2003)
R Pollard & G Pollard
To cite this article: R Pollard & G Pollard (2005) Long-term trends in home advantage in
professional team sports in North America and England (1876 – 2003), Journal of Sports
Sciences, 23:4, 337-350, DOI: 10.1080/02640410400021559
To link to this article: http://dx.doi.org/10.1080/02640410400021559
Published online: 18 Feb 2007.
Submit your article to this journal
Article views: 1077
View related articles
Citing articles: 46 View citing articles
Long-term trends in home advantage in professional team sports in
North America and England (1876 – 2003)
R. POLLARD
1
& G. POLLARD
2
1
Department of Statistics, California Polytechnic State University, San Luis Obispo, CA and
2
Galaxy Creations, Santa
Monica, CA, USA
Abstract
Home advantage is quantitatively defined and calculated for each season since the start of the main professional sports in
North America and England. Over 400,000 games are analysed. The leagues represented are the National League (1876 –
2002) and American League (1901 – 2002) for baseball, the National Hockey League (1917 – 2003) for ice hockey, the
National Football League (1933 – 2002) for American football, the National Basketball Association (1946 – 2003) for
basketball, and the four levels of professional football, formerly called the Football League, in England (1888 – 2003).
Problems caused by unbalanced playing schedules are considered. The results are presented graphically to show long-term
trends and sudden changes. The highest levels of home advantage for all sports were in their early years of existence. Home
advantage in ice hockey, basketball and football in England has declined over the last two decades. In baseball there has been
very little change over the last 100 years, with home advantage consistently lower than in other sports. There was a large drop
in home advantage in football in England following the 7-year suspension of the league during the Second World War. The
trends and changes provide some evidence that travel and familiarity contribute to home advantage, but little in support of
crowd effects.
Keywords: Baseball, basketball, football, hockey, home advantage, soccer
Introduction
The existence of home advantage is well established
and its causes have been the subject of much recent
research. Although the magnitude of the advantage
has been quantified in different sports at different
times, no systemic attempt has been made to
interpret home advantage from a historical perspec-
tive. This paper presents a year-by-year
measurement of home advantage for the four major
professional sports in North America (baseball,
American football, ice hockey and basketball) and
for football in England. The starting point for each
sport is taken to be the foundation of its main
professional league. The aim of the study is to
identify variations and trends over time and to
investigate them in the context of any insight they
might provide into the hypothesized causes of home
advantage.
The first detai led quantitative consideration of
home advantage was carried out by Schwartz and
Barsky (1977). Although most of their data related to
professional sport in North America in 1971, they
did report 5-year home advantage figures for Amer-
ican football dating back to 1945. Other analyses on
competitions before 1971 were based on play-off
games with relatively small sample sizes. Several
publications have quantified home advantage in
professional baseball from 1900 by decade (e.g.
Thorn & Palmer, 1985). Pollard (1986) carried out a
similar analysis for the First Division of the Football
League in England from 1888 to 1984. This was
brought up to date to include the Premiership in a
paper by Pollard and Pollard (in press), in which
home advantage was also computed for both the FA
Cup and the European Cup between 1960 and 2002.
Dowie (1982) analysed home advantage in the four
divisions of the Football League between 1946 and
1981. Clarke and Norman (1995) compared the
home advantage of each of the 94 professional
football teams in England for each season from
1981 to 1990. The only other studies on home
advantage making use of long sets of historical data
relate to the Winter Olympics from 1908 to 1998
(Balmer, Nevill, & Williams, 2001) and the Summer
Olympics from 1896 to 1996 (Balmer, Nevill, &
Williams, 2003). The many other studies on home
advantage that have been conducted since 1980 have
Correspondence: R. Pollard, 2401 Cloverfield Boulevard, Santa Monica, CA 90405, USA. E-mail: rpollard@calpoly.edu
Journal of Sports Sciences, April 2005; 23(4): 337 – 350
ISSN 0264-0414 print/ISSN 1466-447X online # 2005 Taylor & Francis Group Ltd
DOI: 10.1080/02640410400021559
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
focused on data for at most a few seasons or, if for
longer peri ods of time, involving only a small sample
of games.
There appear to be many causes for home
advantage, each contributing a small amount and
interacting with each other in ways that vary from
sport to sport. Nevill and Holder (1999) provide a
general review for all sports, while P ollard and
Pollard (in press) review the specific causes for
soccer and propose a model for their interacting
effect on home advantage.
This paper documents the results of a season-by-
season analysis of home advantage beginning in 1876
and based on the results of over 400,000 games in
five professional team sports in England and North
America.
Methods
Calculation of home advantage
For sports in which the outcome of a game is
either a win or a loss, home advantage can be
quantified as the number of games won by the
home team expressed as a percentage of all games
played. Baseball and basketball fall into this
category.
For sports in which ties are allowed, different
numbers of points are awarded to each team
depending on the result. In this case, home
advantage is quantified as the number of points
obtained by the home team expressed as a
percentage of all points obtained in all games
played. Ice hockey and football (soccer) fall into
this category, although several different systems of
awarding points have been employed. Ties occur
only occasionally in American football, which
produces standings based on win percentage into
which ties are incorporated.
This method of calculating home advantage has
drawbacks. When teams of greatl y differing abilities
play each other, the better team is likely to win
whether or not the game is played at home or away.
Difference in ability will outweigh the relatively small
advantage of playing at home when determining the
result of a match. This factor is of particular
importance if a comparison of home advantage
between teams is being made. Since this is not the
focus of this study and since there is, in any case,
unlikely to be large differences in ability between
teams competing in elite professional leagues, this
simple and unadjusted measure of home advantage is
appropriate. The issue of team quality relating to
home advantage has been explored by Clarke and
Norman (1995) for soccer and by Madrigal and
James (1999) and Harville and Smith (1994) for
basketball.
Types of playing schedule
Balanced schedules. When teams form themselves into
a league, there are various ways in which the playing
schedule during a season can be arranged. If each
team plays each other team the same number of
times at home and away, then the schedule is said to
be balanced. The Football League in England has
used this system every season since its inception in
1888, each team playing each other team twice
during a season, once at home and once away. Most
football leagues throughout the world follow this
model, although the United States, Scotland and
Ireland are among current exceptions. Balanced
schedules allow a straightforward and unbiased
calculation of home advantage.
Unbalanced schedules. There are several ways in which
the playing schedule can lose its balance:
1. Unequal number of games played by each team.
2. Unequal number of games played at home and
away.
3. Each team plays the same number of games at
home and away, but not against the same
opponents.
In the early days of the four North American sports,
especially nineteenth-cen tury baseball, the first two
types of imbalance were certainly present. Since the
1960s, as the number of professional teams has
increased, leagues have split into conferences and
divisions, so that the third type of imbalance has now
become a factor.
The potential problem that unbalanced schedules
might present for interpretation of home advantage is
likely to be minimal, since this study is concerned
with complete leagues rather than a comparison
between teams within a league. If an unbalanced
schedule increases the home advantage for some
teams, it is likely to have lowered it for others, so the
net effect on the overall figure for the league should
be very small.
Sources of data
The five sports to be considered have different
characteristics and the dat a for each cover different
time periods. The following sections consider each
sport separately with special attention given to the
way in which changes in the structure of each league
might have affected the comparability of home
advantage calculations over time.
Baseball. The oldest continuous professional sports
league in the world is the National League, which
came into being with eight teams in 1876 and has
338 R. Pollard & G. Pollard
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
operated every year since then. Several other profes-
sional leagues were formed in the nineteenth century
but none survived for more than a few years. In 1901,
the American League started play, operating in
parallel with the National League each year since
then and constituting what is now known as Major
League Baseball (MLB). This study uses the
National League (1876 – 2002) and the American
League (1901 – 2002) as its data sets for professional
baseball.
Although the playing schedules in the early years
of the National League showed some imbalance,
near perfect balance was achieved after a reorganiza-
tion of the league in 1900, a situation that continued
until 1969 when both the American and National
Leagues split into two regional divisions; there was a
further split into three divisions in 1994. In 1997, the
slight imbalance that this had created was increased
when teams from each league started playing against
each other for the first time. In 1994 , a player strike
brought the season to an abrupt end in August,
resulting in imbalance in the final standings of that
season for all three reasons given above.
Ice hockey. The National Hockey League (NHL)
began on 19 December 1917 with four teams. Two
weeks later this was reduced to three when the home
stadium of the Montreal Wanderers was burned to
the ground. Their initial four games were included in
the standings and their next two games were forfeited
to maintain a balanced schedule. The remainder of
the season continued as planned, leaving a total of 34
games actually played. The NHL continued with a
balanced schedule involving only a few teams until
1926 – 27 when the league split into two divisions.
This slight imbalance continued until the 1938 – 39
season, when the league reverted to a single division
with a perfectly balanced schedule. In 1967 – 68, the
NHL started a series of expansions, first into two
divisions, then in 1974 – 1975 into two conferences
each with two divisions, finally becoming three in
1998 – 99. Inevitably the schedule s are no longer
balanced. Typically a team will play a balanced
schedule within its division, with fewer games against
teams in other divisions and in the other conference.
In the NHL, points are awarded to teams
depending on the result of each match. From
1917 – 18 up to 1998 – 99, two points were awarded
for a win, one for a tie and none for a loss. From
1999 – 2000, a slight modification was made. For a
game tied at the end of regulation time, but then
decided in the 5-min overtime period, the winning
team still obtains two points, but the losing team is
now awarded one point. Only 10% of games have
been ending in this way, and those that do are won
by the home team in the same proportion as in other
games. Thus the small number of extra points gained
in this way will produce a negligible change in the
calculation of home advantage.
American football. The National Football League
(NFL) began in 1922, but with a schedule that was
so unbalanced and brief that an estimate of home
advantage is not worthwhile. In 1933, the NFL was
reorganized into a league consisting of 10 teams and
two divisions. Several important rule changes were
implemented at the sam e time, so that 1933 can be
used as a suitable starting point for the NFL in this
study. Although the playing schedule was still
unbalanced, it was much less so than in previous
seasons. In 1936, teams finally started to pla y an
equal number of games, but it was not until the 1948
season that the same number of games was played
home and away, although not against the same
teams. In 1960, the American Football League
(AFL) began, with two divisions operating in parallel
with the NFL. There was a further change in 1967
when the NFL split into two conferences each with
two divisions. Finally, in 1970 the NFL and AFL
merged into a single league with two conferences,
each with three divisions, a situation which prevailed
until 2002 when each conference expanded to four
divisions. In most seasons since 1948, each team has
played a balanced sc hedule only within its division,
the schedule for the remaining games being deter-
mined in a variety of ways. Seasons 1982 and 1987
were unbalanced for all three reasons given above
due to player stri kes. This study uses all games
played in the NFL and AFL as its data source for
American football.
Standings in the NFL are expressed as a winning
percentage. Ties occur infrequently. From 1933 to
1971, tied games were omitted completely from the
calculation of winning percentage, as if the games
had not taken place. Since 1972, they have been
included as a ‘‘half-win’’ and ‘‘half-loss’’.
The NFL has consistently played the least balanced
schedule, as well as far fewer games each season than
any of the other sports. This means that season by
season home advantage calculations fluctuate more
than in other sports and should be considered the
least reliable of the five sports under study.
Basketball. The inaugural season of the National
Basketball Association (NBA) was 1946 – 47, when
10 teams competed in two divisions, each team
playing 60 games with a playing schedule that was
balanced, a situation that the NBA has never been
able to repeat. From 1948 – 49 until 1973 – 74, many
NBA games took place on neutral territory. The
number varied from season to season (a high of 31%
of games in 1954 – 55) but averaged 13% over the
period. All these games are excluded from the
analysis and represent an increase in imbalance.
Long-term trends in home advantage 339
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
The only major change in league format occurred in
1970 – 71 when the NBA split into two conferences
each with two divisions. An important rule change,
the 3-point shot, wa s introduced in 1979 – 80 but is
unlikely to have had an effect on home advantage.
Football. Three years after the legalization of profes-
sionalism, the Football League in England was
created in 1888 – 89 with a series of games on 8
September 1888. The league consisted of a total of
12 teams from Lancashire and the Midlands, each
team playing each other once at home and once
away. This perfectly balanced playing schedule was
the creation of William Macgregor, the founder of
the league, and has remained unchanged to the
present day. In 1892 – 93, a second division was
formed, with promot ion and relegation operating
between Division 1 and Division 2. The following
year Arsenal became the first team from the south of
England to be admitted to the Football League;
expansion continued and by 1914 – 15 each division
consisted of 20 teams. The next four seasons were
cancelled due to the First World War, but by 1921 –
22 the league had expanded to four divisions with the
additions of Division 3 South and Division 3 North,
promotion and relegation taking place between these
divisions and Division 2. Seven seasons (1939 – 40 to
1945 – 46) were lost as a result of the Second World
War. In 1958 – 59, a further change occurred when
the two regional third divisions were re-arranged into
Division 3 and Division 4. The next reorganization
was in 1992 – 93, with the creation of the Premier
League (replacing the old Football League Division
1), and the reduction of the Football League to three
divisions (renamed Divisions 1, 2 and 3). With
promotion and relegation between the four newly
named divisions still intact, the net effect on the
structure was minimal.
The four divisions desc ribed above form the basic
data set for football, but are analysed separate ly due
to differences in level of play. To avoid confusion
due to the renaming of the various divisions, the
four dat a sets will be referred to by their levels as
follows:
Level 1: Division 1 (1888 – 1992); Premier League
(1992 to present).
Level 2: Division 2 (1892 – 1992); Division 1 (1992
to present).
Level 3: Division 3 (1920 – 1921); Division 3 (North
and South, 1921 – 1958); Division 3 (1958 – 1992);
Division 2 (1992 to present).
Level 4: Division 4 (1958 – 1992); Division 3 (1992
to present).
In football, points are awarded according to the
result of each game. The original system of two
points for a win, one for a draw (tie) and none for a
loss was adopted only after the start of the inaugural
1888 – 89 season, and against the wishes of several
teams that would have preferred no points for either
team in the event of a draw. This system lasted until
1981 – 82, when it was decided to award three points
for a win instead of two. This has two implications
for home advantage. First, awarding three points
instead of two for a win is certain to increase the
calculated value of home advantage bei ng used in
this study, provided more games are won at home
than away. Typically, the increase will be small,
about one percentage point. Secondly, one of the
reasons for the change was to discourage defens ive
play, especially by the away team. Since special
tactics by the away team has bee n advanced as a
possible contributing factor to home advantage, it is
possible the new points system might have contrib-
uted to a change in home advantage.
As with the other sports, minor rule changes have
occurred over the years. The most significant change
in football was the modification of the offside rule in
1925 – 26, which resulted in a 40% increase in the
number of goals scored, with possible consequences
for home advantage calculations which are based on
the result of a match and not the margin of victory
Location of data. To perform the analysis, it was
necessary to obtain home and away standings for each
year in which the leagues of each sport have been in
operation. The baseball standings were extracted
from Total Baseball, a regularly updated publication,
which also provides a wealth of other data comparing
various performance measures at home and on the
road. The basketball standings were obt ained from
the annual Sporting News Official NBA Guide. The
data for ice hockey (from 1926 – 27) and American
football were taken from www.shrpsports.com, an
excellent web site giving not only the annual
standings classified by home and away, but also the
results for all games for all seasons. This web site in
addition pr ovides the same information for baseball
(from 1901) and for basketball. The standings for ice
hockey before the 1926 – 27 season were extracted
from the web site www.geocities.com/Colosseum/
Count/9902/History/Schedules. The standings and
complete results for football in England are available
from Laschke (1980) up to the 1978 – 79 season and
from the annual publication Rothmans Football Year-
book thereafter.
Sample size and standard error
The annual number of games played in the leagues
under consideration range from under 50 in the early
days of ice hockey and American football to around
1200 in several leagues today. Since the standard
340 R. Pollard & G. Pollard
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
error of a sample home advantage percentage
depends both on the number of games (n) and the
actual home advantage percentage (p), it will vary
from season to season and from sport to sport. The
binomial formula for the standard error of a
proportion is the square root of (p(1 - p)/n) and
requires that p is constant. Applied to home
advantage, p represents the overall proportion of
games won by the home team in a league. This will
clearly vary from game to game depending on the
relative strengths of the teams involved. It can be
shown that the above formula for the standard error
is the upper bound of the true standard error in such
a situation. Table I gives the standard error for a
selection of values of n and p. The table is based on
the binomial formula and will therefore slightly
overestimate the true standard errors. Since overall
home advantage will seldom be outside the range of
50% to 70%, the number of games is seen to be the
main influence on the standard error.
Home advantage percentage in baseball, basketball
and American football is based on wins and losses
and calculated from a simple proportion from which
the values in Table I are derived. Because ties and
draws in ice hockey and soccer lead to home
advantage percentages based on points, standard
error calculatio ns should be derived from the multi-
nomial distribution (Pollard, 1986). This would
require a separate calculation for each season
depending on the points system in use and on the
proportion of games won, tied and lost. It would also
be subject to the same problem of non-constant
probabilities. The number of games played will again
be the main influence on the standard errors. It
seems reasonable that the values in Table I can at
least be used as a guide to the magnitude of the true
standard errors for these sports.
Analysis
The procedure of analysis for each sport was the
same. Home advantage was calculated for each
season and the results presented graphically to
provide an initial picture of trends or sudden changes.
The graphs were constructed on the same vertical and
horizontal scales to aid comparison between sports.
Results
Table II (North American sports) and Table III
(football in England) provide, for each season and for
each sport, the number of games played and the home
advantage. Theannualhomeadvantages are illustrated
graphically inFigures1 and 2.There are several general
trends in home advantage common to more than one
sport. These will be presented first, followed by a more
detailed look at each individual sport.
General trends
In each sport, home advantage was at or near its
highest in the early years of each league. Every league
had its season of highest home advantage during the
first 10 years of its existence. The only exception is
Level 4 of the Football League, but this was created
by a reorganization of Level 3 and not as a new
league. Several leagues have experienced a drop in
home advantage over the last two decades. This is
especially evident in ice hockey and basketball as well
as in football in England. However, this trend is not
evident in baseball or American football
Baseball
Home advantage was at its highest in the early years
of the National League during the nineteenth
century, averaging around 60%. The first three
seasons of the American League had a similar figure,
but since 1904 home advantage has been amazingly
constant at an average of about 54% for both leagues,
with annual fluctuations being mostly within two
standard errors of this value.
Ice hockey
Home advantage was very high in the first seven
seasons of the NHL, reaching 75.0% in 1922 – 23,
the highest figure recorded for any year in any sport
in this study. As the league began to expand, home
advantage declined, but by the 1930s had settled at
an average of around 60%, a figure that continued
until the mid-1970s. Since then, there has been a
steady decline that has levelled off at around 55%
since the mid-1990s.
American football
Because of the smaller number of games played
per season, home advantage in the NFL shows
much more annual variability than in the other
sports. There are no obvious trends and certa inly
Table I. Approximate standard errors of home advantage
percentage
Actual home advantage (p)
Games played (n)50 60 70
50 7.1 6.9 6.5
100 5.0 4.9 4.6
200 3.5 3.5 3.2
500 2.2 2.2 2.0
1000 1.6 1.5 1.4
1200 1.4 1.4 1.3
Long-term trends in home advantage 341
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
Table II. Annual home advantage and games played in North American professional sports leagues
Home advantage (%) Games played
Year* NL AL NHL NFL NBA NL AL NHL NFL NBA
1876 52.5 257
1877 65.5 177
1878 55.0 180
1879 55.1 316
1880 56.3 332
1881 55.4 334
1882 59.0 334
1883 61.0 390
1884 56.8 447
1885 54.1 442
1886 60.2 480
1887 57.1 492
1888 55.3 532
1889 58.7 518
1890 62.9 531
1891 56.3 545
1892 58.3 903
1893 62.2 773
1894 61.2 783
1895 62.2 783
1896 60.2 778
1897 62.2 788
1898 59.4 897
1899 59.0 903
1900 58.1 554
1901 54.4 57.9 553 539
1902 52.9 62.9 548 542
1903 52.8 59.9 551 548
1904 54.7 53.5 611 608
1905 54.8 55.8 611 606
1906 52.1 56.3 606 600
1907 55.1 53.8 603 599
1908 51.0 56.9 616 612
1909 51.1 56.3 610 607
1910 55.0 57.0 614 609
1911 51.3 54.1 608 610
1912 52.5 51.9 608 611
1913 51.2 50.9 603 609
1914 56.6 53.8 615 612
1915 57.1 53.7 611 611
1916 54.0 57.6 611 615
1917 49.7 51.5 70.6 612 612 34
1918 56.9 56.0 74.1 504 502 27
1919 54.1 56.0 66.7 556 557 48
1920 53.3 53.3 62.5 614 614 48
1921 53.9 54.7 64.6 612 614 48
1922 54.6 55.7 75.0 615 616 48
1923 49.9 51.9 70.8 615 611 48
1924 53.7 53.7 57.2 613 613 90
1925 56.0 56.0 59.9 612 612 126
1926 58.1 55.0 53.2 611 611 220
1927 56.5 55.7 52.0 614 614 220
1928 53.6 50.6 56.1 612 615 220
1929 54.3 54.6 58.9 611 610 220
1930 57.1 57.1 58.9 616 616 220
1931 58.3 58.1 64.8 614 613 192
1932 55.7 55.0 68.3 616 613 216
1933 57.4 54.3 61.1 61.5 612 604 216 52
1934 55.7 53.9 54.4 51.7 605 610 216 60
1935 56.1 53.3 59.9 49.0 613 606 192 49
1936 54.4 56.0 54.2 57.7 616 612 192 52
(continued)
342 R. Pollard & G. Pollard
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
Table II. (continued)
Home advantage (%) Games played
Year* NL AL NHL NFL NBA NL AL NHL NFL NBA
1937 52.8 55.9 58.9 48.1 612 614 192 52
1938 51.9 55.7 58.0 51.9 603 605 168 52
1939 56.2 51.1 62.8 63.5 610 612 168 52
1940 50.5 55.4 61.0 66.7 612 617 168 51
1941 53.0 54.9 63.7 56.6 615 616 168 53
1942 55.4 53.9 67.3 50.0 606 609 150 54
1943 55.4 55.0 63.3 62.2 614 613 150 37
1944 53.0 56.8 59.0 57.4 615 616 150 47
1945 53.9 59.6 61.7 61.2 614 604 150 49
1946 55.9 54.5 58.3 53.9 61.0 617 616 180 52 331
1947 55.4 52.6 55.6 63.8 55.7 616 616 180 58 192
1948 50.2 51.7 61.1 59.3 61.6 614 615 180 59 360
1949 53.4 58.4 57.9 50.9 67.8 616 616 210 57 561
1950 55.4 54.1 57.9 57.7 74.9 614 616 210 78 354
1951 52.0 53.2 58.1 47.8 72.7 619 616 210 69 330
1952 52.7 57.3 58.3 48.6 70.1 615 616 210 72 351
1953 56.2 49.1 65.0 55.1 64.3 616 613 210 69 324
1954 52.8 53.4 57.9 61.4 70.2 615 616 210 70 288
1955 58.2 54.2 61.4 65.2 63.8 615 616 210 69 288
1956 55.8 51.3 59.5 60.0 69.4 616 616 210 70 288
1957 51.5 53.6 53.3 56.3 63.8 616 614 210 71 288
1958 53.4 56.4 55.5 60.9 64.9 616 615 210 69 288
1959 55.7 52.4 60.5 53.5 63.9 618 616 210 71 300
1960 55.5 53.7 60.7 54.7 64.7 616 616 210 128 316
1961 53.6 56.3 62.4 55.3 60.8 616 807 210 150 360
1962 55.3 51.8 54.5 51.0 62.8 812 809 210 149 360
1963 55.7 54.6 65.5 58.2 59.5 811 808 210 146 360
1964 52.7 52.5 55.5 54.5 58.7 810 810 210 145 360
1965 53.8 53.5 60.5 51.7 68.9 809 810 210 147 360
1966 53.8 52.9 61.2 57.2 60.1 808 805 210 159 405
1967 57.4 55.2 60.6 54.3 57.0 809 808 444 164 492
1968 50.6 51.7 60.9 46.9 60.5 810 809 456 177 574
1969 54.1 55.6 59.1 60.9 60.2 972 971 456 174 574
1970 52.9 55.1 64.1 58.4 60.5 971 972 546 173 697
1971 52.4 51.6 60.8 57.5 58.2 971 966 546 174 697
1972 49.4 56.4 63.5 50.8 59.5 929 929 624 182 697
1973 52.4 53.7 59.8 60.7 61.0 970 972 624 182 697
1974 53.5 53.3 62.2 54.7 64.0 971 969 720 182 738
1975 55.9 52.1 59.4 55.5 65.6 970 963 720 182 738
1976 52.6 52.2 59.1 57.4 68.5 972 967 720 196 902
1977 56.3 52.7 58.8 57.7 67.6 972 1131 720 196 902
1978 57.3 57.4 58.3 58.3 66.5 970 1131 680 224 902
1979 53.6 54.3 59.3 58.9 65.2 969 1128 840 224 902
1980 57.2 51.6 59.3 54.7 62.1 972 1129 840 224 943
1981 54.1 50.5 60.2 62.3 59.9 640 749 840 224 943
1982 51.3 55.9 60.5 54.4 62.0 972 1134 840 126 943
1983 54.6 53.9 57.9 53.3 67.9 972 1134 840 224 943
1984 52.0 53.7 57.5 57.8 63.7 971 1133 840 224 943
1985 53.5 56.3 58.3 64.3 65.4 970 1131 840 224 943
1986 54.3 55.1 60.1 53.1 66.5 969 1133 840 224 943
1987 54.6 55.0 57.3 54.5 67.9 971 1134 840 210 943
1988 53.2 54.3 58.3 58.7 67.7 967 1131 840 224 1025
1989 54.3 55.6 58.1 57.4 64.4 970 1133 840 224 1107
1990 54.2 53.3 58.9 58.5 65.9 972 1133 840 224 1107
1991 54.9 52.7 60.5 58.9 63.1 970 1134 880 224 1107
1992 55.7 54.8 55.7 60.7 61.1 972 1134 1008 224 1107
1993 53.0 54.6 54.0 54.9 61.2 1133 1134 1092 224 1107
1994 50.7 52.6 58.3 57.1 59.7 802 794 624 224 1107
1995 52.7 53.6 56.1 60.0 60.4 1007 1009 1066 240 1189
1996 56.3 51.8 55.0 62.1 57.5 1134 1132 1066 240 1189
1997 55.4 51.7 53.6 60.8 59.5 1134 1132 1066 240 1189
(continued)
Long-term trends in home advantage 343
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
Table II. (continued)
Home advantage (%) Games played
Year* NL AL NHL NFL NBA NL AL NHL NFL NBA
1998 54.9 52.6 54.4 62.5 62.3 1297 1133 1107 240 725
1999 52.7 51.3 54.6 59.7 61.1 1295 1132 1148 248 1189
2000 54.3 53.8 54.6 55.6 59.8 1296 1132 1230 248 1189
2001 52.1 52.8 55.0 54.4 59.1 1296 1132 1230 248 1189
2002 54.5 53.7 54.0 58.0 62.8 1293 1132 1230 256 1189
* For seasons that cover two years (NHL and NBA), the year given represents the first year.
For example, 1980 represents season 1980 – 1981.
Table III. Annual home advantage and games played in the Football League in England
Home advantage (%) Games played
Year Level 1 Level 2 Level 3 Level 4 Level 1 Level 2 Level 3 Level 4
1888 – 89 67.4 132
1889 – 90 64.8 132
1890 – 91 68.2 132
1891 – 92 69.0 182
1892 – 93 67.5 70.8 240 132
1893 – 94 69.6 66.7 240 210
1894 – 95 68.3 72.5 240 240
1895 – 96 72.7 70.4 240 240
1896 – 97 60.4 72.3 240 240
1897 – 98 69.4 69.2 240 240
1898 – 99 68.6 69.0 306 306
1899 – 90 68.0 69.9 306 306
1900 – 01 67.8 70.9 306 306
1901 – 02 72.5 70.6 306 306
1902 – 03 68.3 68.0 306 306
1903 – 04 65.8 70.4 306 306
1904 – 05 65.7 65.4 306 306
1905 – 06 68.8 64.7 380 380
1906 – 07 68.2 71.2 380 380
1907 – 08 64.9 67.4 380 380
1908 – 09 61.2 68.0 380 380
1909 – 10 65.0 67.5 380 380
1910 – 11 64.2 69.3 380 380
1911 – 12 65.8 66.1 380 380
1912 – 13 61.6 69.3 380 380
1913 – 14 64.2 67.8 380 380
1914 – 15 64.7 68.3 380 380
1919 – 20 64.7 65.0 462 462
1920 – 21 62.6 65.0 67.2 462 462 462
1921 – 22 64.9 64.8 67.9 462 462 924
1922 – 23 69.4 64.1 68.4 462 462 924
1923 – 24 64.5 67.6 70.3 462 462 924
1924 – 25 64.4 64.3 69.9 462 462 924
1925 – 26 67.3 67.1 69.9 462 462 924
1926 – 27 68.6 66.9 68.9 462 462 924
1927 – 28 66.2 66.5 68.0 462 462 924
1928 – 29 67.1 68.7 69.2 462 462 924
1929 – 30 65.9 68.6 68.3 462 462 924
1930 – 31 66.1 69.8 70.0 462 462 924
1931 – 32 68.4 64.9 69.7 462 462 924
1932 – 33 67.2 66.8 71.2 462 462 924
1933 – 34 68.1 68.6 68.2 462 462 924
1934 – 35 66.9 68.7 70.4 462 462 924
1935 – 36 68.1 68.1 69.2 462 462 924
(continued)
344 R. Pollard & G. Pollard
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
Table III. (continued)
Home advantage (%) Games played
Year Level 1 Level 2 Level 3 Level 4 Level 1 Level 2 Level 3 Level 4
1936 – 37 69.6 65.2 69.2 462 462 924
1937 – 38 64.8 68.9 68.5 462 462 924
1938 – 39 66.5 65.2 68.3 462 462 924
1946 – 47 60.0 65.4 62.5 462 462 924
1947 – 48 62.3 62.2 60.9 462 462 924
1948 – 49 64.6 66.6 64.9 462 462 924
1949 – 50 61.7 62.9 65.0 462 462 924
1950 – 51 59.5 64.7 66.5 462 462 924
1951 – 52 61.4 66.0 67.5 462 462 924
1952 – 53 64.5 63.5 68.2 462 462 924
1953 – 54 65.7 65.2 67.4 462 462 924
1954 – 55 60.3 65.7 64.5 462 462 924
1955 – 56 64.0 65.4 66.5 462 462 924
1956 – 57 65.0 62.7 65.9 462 462 924
1957 – 58 63.4 63.4 65.3 462 462 924
1958 – 59 61.6 66.8 67.3 63.0 462 462 552 552
1959 – 60 60.9 63.4 67.0 66.5 462 462 552 552
1960 – 61 63.5 64.9 68.2 65.6 462 462 552 552
1961 – 62 64.7 64.7 62.1 68.0 462 462 552 506
1962 – 63 61.3 66.6 64.5 65.3 472 462 552 552
1963 – 64 61.0 63.5 64.7 65.3 462 462 552 552
1964 – 65 65.8 65.8 64.2 66.8 462 462 552 552
1965 – 66 64.2 66.0 64.9 66.2 462 462 552 552
1966 – 67 62.1 66.8 66.8 64.8 462 462 552 552
1967 – 68 64.8 61.3 67.5 63.2 462 462 552 552
1968 – 69 65.6 63.7 66.8 62.9 462 462 552 552
1969 – 70 59.8 66.5 63.3 66.7 462 462 552 552
1970 – 71 61.1 64.0 61.4 65.2 462 462 552 552
1971 – 72 63.1 68.1 63.2 66.7 462 462 552 552
1972 – 73 65.2 64.6 67.4 66.2 462 462 552 552
1973 – 74 63.3 61.4 63.2 67.3 462 462 552 551
1974 – 75 64.3 65.3 70.0 65.4 462 462 552 552
1975 – 76 63.3 63.7 64.7 64.4 462 462 552 552
1976 – 77 66.8 63.4 64.9 68.1 462 462 552 552
1977 – 78 62.6 66.1 66.7 66.4 462 462 552 552
1978 – 79 60.8 62.6 63.8 63.7 462 462 552 552
1979 – 80 63.9 63.1 65.0 63.1 462 462 552 552
1980 – 81 65.4 62.6 62.7 63.8 462 462 552 552
1981 – 82 60.3 64.9 61.9 60.7 462 462 552 552
1982 – 83 68.7 64.7 67.7 65.5 462 462 552 552
1983 – 84 62.8 64.5 66.1 66.7 462 462 552 552
1984 – 85 64.0 61.7 63.0 64.6 462 462 552 552
1985 – 86 62.5 65.3 62.8 63.6 462 462 552 552
1986 – 87 66.3 64.7 62.4 62.6 462 462 552 552
1987 – 88 60.1 62.2 61.1 59.5 420 506 552 552
1988 – 89 56.7 63.4 62.9 64.3 380 552 552 552
1989 – 90 62.2 61.8 61.8 60.4 380 552 552 552
1990 – 91 64.6 61.8 61.0 66.6 380 552 552 552
1991 – 92 60.7 61.5 63.7 62.2 462 552 552 462
1992 – 93 61.5 62.0 61.1 57.9 462 552 552 462
1993 – 94 57.7 64.3 61.9 55.8 462 552 552 462
1994 – 95 59.8 65.1 60.1 59.1 462 552 552 462
1995 – 96 63.0 59.2 61.9 60.2 380 552 552 552
1996 – 97 59.3 62.2 63.1 60.7 380 552 552 552
1997 – 98 61.9 61.6 63.5 64.7 380 552 552 552
1998 – 99 60.7 61.2 57.9 60.0 380 552 552 552
1999 – 00 62.3 62.7 57.1 58.5 380 552 552 552
2000 – 01 62.8 59.0 58.3 65.1 380 552 552 552
2001 – 02 57.4 60.5 62.2 62.5 380 552 552 552
2002 – 03 62.0 58.4 57.0 57.8 380 552 552 552
Long-term trends in home advantage 345
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
no evidence of the recent decline seen in ice
hockey and basketball. In fact, the reverse appears
to be the case, with home advantage appearing to
increase slightly over the last 15 years.
Basketball
Home advantage in the NBA has shown more
changes over the years than other sports. After
346 R. Pollard & G. Pollard
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
quickly climbing to a very high 74.9% in 1950, it
gradually decreased to around 60% by the late
1960s. Season 1965 – 66 stands out with a value of
68.9%, around 10% higher (representing about 3
standard errors) than the seasons immediately before
and after. After this there wa s a series of climbs and
falls until home advantage had increased to around
65% by the late 1980s. This was followed by another
decline that has currently levelled off at about 60%.
Football
The early years of the Football League in England
saw home advantage averaging close to 70%, with a
peak of 72.7% in 1895 – 96. For Level 1 there was
then a decline to values below 65%, but by the
1930s home advantage was averaging around 67%.
However, the 7-year suspension of the league
during the Second World War was followed by an
immediate drop in home advantage to its lowest
ever value of 60.0%. A slight increase followed, but
since the late 1980s annual values below 60% are
not uncommon.
The graph for home advantage in Level 2 is the
smoothest of the four levels and suggests a steady
decline fr om values around 70% during the 1890s to
around 60% at present. The drop after the Second
World War was less dramatic than in Level 1;
nevertheless, immediate post-war values were con-
sistently below the average in the 1930s.
At Level 3, home advantage fluctuated little during
the 1920s and 1930s, averaging just below 70%,
slightly higher than at Levels 1 and 2. As with Level 1
there was a big drop immediately after the Second
World War followed by an increase, but never
reaching pre-war levels. There has been a steady
decline since about 1960, with values currently
around 60%, similar to Levels 1 and 2. Level 4
started in 1958 and has followed a very similar path
to Level 3
Figure 1. Home advantage in professional team sports in North America. (a) Major League Baseball (National League); (b) Major League
Baseball (American League); (c) National Hockey League; (d) National Football League; (e) National Basketball Association.
Long-term trends in home advantage 347
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
Figure 2. Home advantage in professional football in England. (a) Football League Level 1; (b) Football League Level 2; (c) Football League
Level 3; (d) Football League Level 4.
348 R. Pollard & G. Pollard
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
Discussion
The three main causes of home advantage are
believed to be crowd effects, travel and familiarity.
Other causes such as referee bias, territoriality,
special tactics, rules and psychological factors have
been also been advanced. It is likely that they interact
with each other to pr oduce what we know as home
advantage. However, it has proved very difficult to
disentangle the individual contributions of these
effects. Against this background, the trends shown
in the graphs can be examined.
With the exception of American football , home
advantage was at or near its highest in the early years
of each league. In the case of the National League of
baseball and Levels 1 and 2 of the Football League,
this early period was in the nineteenth century before
the days of motorized transport. Travel must hav e
been difficult and tiring and a case can be made for
this being a factor in the high home advantage values
seen. Previous studies on the effects of travel have
focused on distance travelled and time zones crossed,
but mostly for games in the latter part of the
twentieth century. These have failed to show any
consistent effect and are summarized by Nevill and
Holder (1999). An analysis of home advantage in the
Winter Olympics from 1908 to 1998 showed no
change in the travel effect over time (Balmer et al.,
2001). Data on crowd size and density in the
nineteenth century are hard to obtain, but it is
unlikely crowds were larger than subsequ ently. The
effects of a home crowd were probably contributing
to home advantage, but it is difficult to see how these
could be the reason for the elevated early values in
any of the sports. Similarly, an initial lack of
familiarity with a home stadium would have been
expected to lower rather than heighte n home
advantage in the first few years of a league.
The second trend evident in more than one sport
is the decline in home advanta ge that has occurred in
the last 20 years or so. Only American football and
baseball fail to show this trend. During this period,
professional sport has developed into a business with
enormous financial rewards. No stone is left un-
turned to prepare players for games and it is likely
that they are better able to cope with the perceived
disadvantage of playing away from home. Surpris-
ingly, the decline has been greatest in ice hockey, a
sport that typically has a noisy and dense crowd, very
close to the players and favouring the home team.
Home advantage in baseball has always been the
lowest of the major sports with little potential to show
a decline.
Turning to trends specific to individual sports, in
the Football League in England, of particular interest
is the big drop in home advantage that occurred after
the 7-year suspension of the league between 1939
and 1946. It could be argued that the benefits of
familiarity with playing conditions at a home stadium
would have largely disappeared in this time. At the
same time, many new players would have been
recruited with little familiarity with their home
stadium. Hence a drop in home advantage due to
loss of familiarity would be expected. All this is
consistent with the recent finding that when a team
moves to a new stadium, home advantage declines,
with a loss of familiarity the likely explanation
(Pollard, 2002). Crowds immediately after the war
were much larger than subsequently, so that crowd
effects, if they existed, would have suggested, if
anything, an increase rather a fall in home advantage.
Further lack of evidence for crowd effects as a caus e
of home advantage is the fact that home advantage
has always been very similar in the four levels of play
despite large differences in the magnitude and
density of crowd support. The conflicting evidence
of a crowd effect on home advantage is reviewed by
Nevill and Holder (1999). In general, it has been
surprisingly difficult to demonstrate its existence or
quantify its magnitude. Nevill, Newell and Gale
(1996) claimed that an analysis of the 1992 – 93
football season in England and Scotland demon-
strated an incre ase in home advantage with crowd
size. Howeve r, this was mainly due to the inclusion
of both the GM Vauxhall Conference and the
Scottish Second Division in the analysis. Furt her-
more, had they used figures from the previous season
for the Football League (1991 – 92 in Table III), the
opposite conclusion might have been reached. The
decline in home advantage had already started when
three points for a win wa s introduced in 1981 – 82, so
that its effect on home advantage is difficult to assess
(Figure 2). However, it is clear that the theoretical
increase due to the new method of calculating the
advantage failed to materialize. The change in the
offside rule that took place in the 1925 – 26 season,
and which was accompanied by a sharp rise in
goalscoring, did herald the start of a sustained period
of slightly increased home advantage for Levels 1 and
2 (Figure 2), while home advantage at Level 3
remained even higher, around 70%. Since this was
also a period of rapidly evolving tactics, there are no
obvious conclusions to be made.
For basketball, there is no clear reason for the
changing pattern of home advantage apparent in the
NBA between 1950 and 1990, although a steady rise
in the early 1970s did coincide with the split into
regional divisions (Figure 1). However, this is in
contrast to the NHL, in which the series of splits into
conferences and divisions has been mirrored by a
steady decline in home advantage since 1970 (Figure
1)
The NFL has had five seasons with a home
advantage figure bel ow 50%, two of whi ch were
Long-term trends in home advantage 349
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
consecutive (1951 and 1952). The most likely
explanation for this and other fluctuations shown in
Figure 1 is the very small number of games played
each season.
Conclusion
Only a general description of the main historical
trends in home advantage has been made. Tables I
and II are abbreviated versions of the full data sets for
each sport, which include the annual number of
games won, drawn (tied) and lost at home, as well as
the number of teams in each league and other
relevant information. These data sets lend them-
selves well for more specific investigations and are
available from the author on request.
Acknowledgement
I am grateful to Randy Rokosz, creator of the web
site www.shrpsports.com, for help in obtaining and
checking home and away records for the early years
of the NHL, NFL and NBA.
References
Balmer, N. J., Nevill, A. M., & Williams, A. M. (2001). Home
advantage in the Winter Olympics (1908 – 1998). Journal of
Sports Sciences, 19, 129 – 139.
Balmer, N. J., Nevill, A. M., & Williams, A. M. (2003). Modelling
home advantage in the Summer Olympic Games. Journal of
Sports Sciences, 21, 469 – 478.
Clarke, S. R., & Norman, J. M. (1995). Home ground advantage
of individual clubs in English soccer. The Statistician, 44, 509 –
521.
Dowie, J. (1982). Why Spain should win the World Cup. New
Scientist, 94, 693 – 695.
Harville, D. A., & Smith, M. H. (1994). The home-court
advantage: How large is it, and does it vary from team to team?
The American Statistician, 48, 22 – 28.
Laschke, I. (1980). Rothmans book of Football League records:
1888 – 89 to 1978 – 79. London: Macdonald and Jane’s.
Madrigal, R., & James, J. (1999). Team quality and the home
advantage. Journal of Sport Behavior, 22, 381 – 398.
Nevill, A. M., & Holder, R. L. (1999). Home advantage in sport:
An overview of studies on the advantage of playing at home.
Sports Medicine, 28, 221 – 236.
Nevill, A. M., Newell, S. M., & Gale, S. (1996). Factors
associated with home advantage in English and Scottish soccer
matches. Journal of Sports Sciences, 14, 181 – 186.
Pollard, R. (1986). Home advantage in soccer: A retrospective
analysis. Journal of Sports Sciences, 4, 237 – 248.
Pollard, R. (2002). Evidence of a reduced home advantage when a
team moves to a new stadium. Journal of Sports Sciences, 20,
969 – 973.
Pollard, R., & Pollard, G. (in press). Home advantage in soccer: A
review of its existence and causes. Revista Internacional de Futbol
y Ciencia.
Schwartz, B., & Barsky, S. F. (1977). The home advantage. Social
Forces, 55, 641 – 661.
Thorn, J., & Palmer, P. (1985). The hidden game of baseball. New
York: Doubleday.
350 R. Pollard & G. Pollard
Downloaded by [Cal Poly San Luis Obispo] at 10:41 11 December 2015
