Original research

International Journal of Sports Science

& Coaching
The undervalued set piece: Analysis of 2021, Vol. 16(3) 830–839
! The Author(s) 2021
soccer throw-ins during the English
Article reuse guidelines:
Premier League 2018–2019 season sagepub.com/journals-permissions
DOI: 10.1177/1747954121991447
journals.sagepub.com/home/spo

Joseph Antony Stone1 , Adam Smith2 and Anthony Barry3

Abstract
Set pieces in soccer (i.e., free kicks and corners) have been examined in detail and are a common focus for coaches during
training and performance preparation. However, limited evidence is available on the impact of throw-ins on soccer
performance and if coaches should dedicate time in training towards this specific set piece. Therefore, this research
aimed to firstly examine if throw-in performance is linked with soccer performance, and secondly the effect throw-in
direction and length has on first contact success rate, possession retention, mean time in possession and shot creation.
16,154 throw-ins from 380 English Premier League matches during the 2018–2019 season were analysed. Higher final
league position was correlated to increased throw-in first contact success and possession retention. 83% of throw-in’s
resulted in a successful first contact, 54% resulted in possession being retained and 8.8% of throw-ins led to a shot at goal
from the possession achieved after a successful first contact. Throw-in’s which went backwards or laterally in direction
resulted in increased first contact success, retaining of possession, and shot creation. The least efficient throw-in was
forwards and long, which resulted in both reduced first contact success and possession retention. Findings highlight, that
throwing the ball laterally or backwards should be a focus for coaches and players during attacking training. In contrast, a
team’s defensive strategy should reduce the opportunities to throw backwards or laterally with a higher press and look to
force a long forward throw-in, therefore, increasing the likelihood of winning possession and counter attacking.

Keywords
Association football, performance analysis

Introduction
touched the ball when the whole of the ball passes
Performance analysis is now a central element of sport over the touchline, on the ground or in the air
science support for soccer coaches and therefore (Law 15).10 Recently, McKinley11 highlighted in the
research has undergone rapid expansion over recent Major League Soccer (MLS) between 2015 and 2019
years, with studies investigating performance indicators almost 64,000 throw-ins were taken. This results in an
related to possession, tactical behaviour, positional average of 44 throw-ins occurring each game,
demands and the match location.1–4 Furthermore,
with set pieces accounting for 30% to 40% of goals
scored in elite soccer5 recent research has focused on
Reviewers: Claudio Casal (Catholic University of Valencia, Spain)
set piece examination of corner kicks,6 free kicks,7,8
Tim Swartz (Simon Fraser University, Canada)
and penalty kicks.9 Findings highlight these set piece Hui Zhang (Zhejiang University, China)
game events as critical components of successful offen- 1
Sport and Human Performance Research Group, Sport and Physical
sive performance in soccer and hence form a key focus Activity Research Centre, Sheffield Hallam University, Sheffield, UK
area during professional soccer training and perfor- 2
Wigan Athletic Football Club, Wigan, UK
3
mance preparation. However, one set piece which has Chelsea Football Club, London, UK
had limited investigation is the throw-in and therefore
Corresponding author:
it is unknown if coaches should dedicate time in train- Joseph Antony Stone, Sport and Physical Activity Research Centre,
ing towards this specific set piece. A throw-in is Sheffield Hallam University, Sheffield, UK.
awarded to the opponents of the player who last Email: joseph.stone@shu.ac.uk
Stone et al. 831

accounting for almost 5% of all passes. This means Method

throw-ins occur more frequently per match than
corner kicks (10),6,12 free kicks (25–35)7,8,13 and goal Sample
kicks (17).14 This highlights the importance throw-ins The 20 English Premier League teams were included in
may have on a team’s possession and the outcome of the sample. Raw coded data on throw-in phases of
matches. play was exported from each of the 380 games
Research on throw-ins has typically focused on a during the 2018/2019 English Premier League season
biomechanical analysis of throwing the ball as long as from the Statsbomb database (https://statsbomb.com).
possible15 with the notion of creating a similar goal Permission to use the data was granted by Statsbomb.
scoring opportunity to a corner kick (i.e. the ability This resulted in a sample of 16,380 phases of play start-
to deliver the ball into the 18-yard box with pre-plan ing from a throw-in. After excluding throws-ins from
routines).16,17 Yet, long throws into the 18-yard box are injury clearances (i.e. possession freely given back to
likely to be a small proportion of the total (approx. the opposition following the ball being kicked out of
44 per match) throw-ins taken per match.11,18 Rather, play due to an injury) a total of 16,154 throw-in’s were
throw-ins are more commonly used to restart a team’s included in the sample (see Table 1) and resulted on
possession.11 With the importance of ball possession average of 808 throw-in’s per team (range 716–912
and shot creation being demonstrated as two factors throw-ins). The Local University ethics committee
that can discriminate between winning, drawing, and granted approval for the study.
losing teams,2 throw-in strategy could therefore direct-
ly influence a team’s ability to retain possession and Measures and procedures
build goal scoring opportunities.
Raw data from each throw-ins phase of play was
The location of the throw-in has been showed to
exported from the Statsbomb database (www.stats
influence game tactics, with throw-ins in the defensive
bomb.com). The phase of play was defined from the
area of the pitch taking longer to take (i.e. increased
start of the throw-in action, to the point the team which
game interruption) than in other areas of the pitch.14
threw the ball lost possession of the ball. Raw data
Despite not examining the actions of the resulting included, the team, opposition team, throw in location
throw-in, Siegle and Lames14 suggested as the team (x, y), outcome of the throw, throw-in outcome loca-
not in ball possession frequently sees a throw-in in tion (x, y), angle of throw in, length of throw-in, time in
the defensive area as an opportunity to conquer the
ball, they create pressure that might lead to a longer Table 1. Total number of throw-ins and mean throw-ins per
throw-in duration. This fits with anecdotal evidence match during the 2018–2019 English Premier League Season.
from soccer coaches that suggests throw-ins in defen-
Throw-ins Mean
sive areas have traditionally been taught to “work the
meeting throw-in’s
line” and “play in their half”. The emphasis being to Team inclusion criteria per match
throw the ball as long as possible in the forward direc-
tion (“down the line”) away from the teams’ own goal. AFC Bournemouth 804 21
However, empirical evidence is required to support the Arsenal 804 21
Brighton & Hove Albion 831 22
effectiveness of this strategy and help inform coaches
Burnley 867 23
tactics. Furthermore, the potential importance of the Cardiff City 768 20
throw-in on soccer matches was recently highlighted in Chelsea 734 19
practice by professional soccer teams starting to hire Crystal Palace 800 21
coaches specialising in throw-in strategy.19 Yet with the Everton 902 24
very limited research to date (for an exception see Fulham 741 20
McKinley11 online article) empirical understanding on Huddersfield Town 912 24
how throw-ins could affect soccer performance is Leicester City 841 22
Liverpool 884 23
needed to aid with future coaching practice.
Manchester City 716 19
Therefore, this research aimed to firstly examine if Manchester United 825 22
throw-in performance is linked with soccer perfor- Newcastle United 805 21
mance, and secondly the effect throw-in direction Southampton 764 20
and length has on first contact success rate, Tottenham Hotspur 810 21
possession retention, mean time in possession, and Watford 737 19
shot creation during the English Premier League West Ham United 792 21
Wolverhampton Wanderers 817 22
2018–2019 season.
832 International Journal of Sports Science & Coaching 16(3)

the match, actions during the possession from the between overall team performance and throw-ins, sep-
throw-in, and the outcome of possession from the arate spearman correlation coefficients were performed
throw-in. Microsoft Office Excel (Version 14.7.1, between final league position (ranked 1–20) and first
Microsoft Cooperation, United States) was used to cal- contact success, possession retention success, mean
culate performance indicators from the raw files for time in possession, and throw-ins resulting in a shot
each of the 20 teams. Based on the performance indi- from the possession achieved after a successful first
cators, three independent variables were examined, contact. Second, to test the relationship between
length (short, medium, long), direction (backwards, league position and throw-in strategy, separate spear-
lateral, forwards) and pitch location (4 areas, see man correlation coefficients were performed between
Figure 1). The effect of these independent variables final league position, and percentage of throw-ins
was examined via four dependant variables, first con- (directions and lengths). Due to the low number of
tact success, possession retention success, mean time in throw-ins taken in the defensive 18-yard area, and the
possession, and shot creation (see Table 2 and Figure 1 expectation of throw-ins in the attacking 18-yard area
for categories and definitions). to have more of an emphasis on direct set pieces and
Statsbomb are one of the leading suppliers of statis- not possession retention, these two zones were excluded
tical data in professional football clubs, media outlets, from further analysis. Data was examined between the
and broadcasters. However, to ensure the reliability of two remaining locations (rest of the defensive half, and
the data set, three randomly selected matches were rest of the attacking half) to examine the influence of
independently coded by the lead author using a specific throw-in strategies on performance (examined
NacSport (NacSport Elite, Las Palmas de Gran via first contact success, possession retention and shot
Canaria, Spain) custom-notational analysis system creation). The majority of data was normally distribut-
examining throw-in location, length, direction and out- ed, examined via Shapiro-wilk tests (p > .05), therefore
come (i.e. first contact succuss and possession reten- parametric analysis was employed. Separate Three-way
tion). Cohen’s kappa coefficient was calculated, based (Location, Direction and Length) repeated measure
on analysis of 106 throw-ins, with a kappa value of Analysis of Variances (ANOVA) were used to exam-
k ¼ 0.97 representing excellent reliability.20 ined first contact success, possession retention, and
mean time in possession with pairwise post-hoc testing
Data analysis using a Bonferroni correction. Finally, due to the lower
Descriptive analyses was employed in Microsoft Office number of shots being created, the two locations were
Excel to calculate relative frequencies for each variable combined and a Two-way (Direction and Length)
and the calculation of performance mean success values ANOVA was employed for shot creation. If the
for each team, (based on each teams 38 games) for each assumption of sphericity was violated, a Greenhouse-
variable. The data was then transferred to SPSS Geisser correction was used. Partial Eta Squared (gp2)
(Version 24.00 SPSS Inc., USA) to perform statistical is presented for effect size estimations of main effects
analysis. First to establish if a relationship existed on ANOVAs.

Figure 1. Definitions of pitch location, direction of throw-in, and length of throw-in (adapted from Siegle and Lames14; McKinley11).
Stone et al. 833

Table 2. Operational definitions for throw-in lengths, directions and outcome variables (based on Statsbomb, 201921 and McKinley,
201811).

Category Operational definition

First contact Successful: A player from the same team which throws the ball into play makes first contact
with the ball post throw-in without an opposition player making contact.
Unsuccessful: A player from the opposition team which throws the ball into play makes first
contact with the ball post throw-in.
Success percentage: Calculated by dividing the number of successful first contacts in a cat-
egory (i.e. short) by the total number of actions (Successful þ Unsuccessful) performed in
that category and multiplying by 100
Time in possession The time (seconds) from the throw-in action to the end of possession. A possession was
defined as a passage of play during which one team is largely in control of the ball. This may
involve that team temporarily being dispossessed, but a new possession will only start if
the opposing team is then able to demonstrate that they are fully in control of the ball
(www.Stasbomb.co.uk).
Possession retention Successful: The ball is retained in possession (as defined above) for 7 seconds from the point
in which the ball is thrown.
Unsuccessful: The ball possession is lost (as defined above) with in 7 seconds from the point
in which the ball is thrown.
Success percentage: Calculated using only the throw-ins which achieved a successful first
contact (n ¼ 13,376). Calculated by dividing the number of successful possessions retained
in a category (i.e. short) by the total number of actions (excluding those this did not
achieve a successful first contact) performed in that category and multiplying by 100
Throw-in resulting in a shot Shot Creation: A shot was recorded when a player attempted a shot at goal which resulted
from the throw-in possession. Success percentage: Calculated based on all throw-ins taken
with throw-ins in each category resulting in a shot divided by total number of throws in
that category, multiplied by 100.
Throw in length Short: The ball was thrown a distance between 0–10 yards (0–9.1 m).
Medium: The ball was thrown a distance between 10–20 yards (9.1–18.2 m).
Long: The ball was thrown a distance of 20 yards or longer (18.2 m).
Throw in direction Forward: The ball is thrown between 0–60 in reference to the sideline towards the offensive
goal.
Lateral: The ball is thrown between 60–120 in reference to the sideline.
Backward: The ball is thrown between 120–180 in reference to the sideline towards the
defensive goal.

Results time in possession after the throw-in (rs (20) ¼ .738,

p < 0.001) and throw-ins resulting in a shot from the
Descriptive analysis is presented in Table 3. A total of
possession achieved after a successful first contact (rs
16,154 throw-in’s were taken during the 2018–2019
(20) ¼ .710, p < 0.05) (see Figure 2). The higher
season (excluding injury clearances), in which 83%
ranked teams had greater success rates in all four
(13,376 throws) resulted in a successful first contact,
54% (8847 throws) resulted in possession being variables.
Final league position was correlated with percentage
retained for 7 seconds or longer, with 8.8% (1422) of
throw-ins resulting in a shot from the possession of throw-ins performed backwards (rs (20) ¼ .662,
achieved after a successful first contact. The most p ¼ 0.001), forwards (rs (20) ¼ .767, p < 0.001) and lat-
common direction of throw was forwards (41.3%) eral (rs (20) ¼ .474, p ¼ 0.035) (Figure 3). Higher
with 78.5% of throw-ins taken in the rest of the attack- ranked teams performed more backwards throw-ins,
ing and defensive areas. whereas lower ranked teams favoured a forward direc-
tion. No relationship was shown between league posi-
Relationship between throw-ins and final tion and lengths of throw-in (p > 0.05).
league position
There was a relationship between league position and First contact success
first contact success (rs (20) ¼ .868, p < 0.001), posses- The three-way repeated measure ANOVA showed an
sion retention success (rs (20) ¼ .768, p < 0.001), mean interaction for direction * length * location for first
834 International Journal of Sports Science & Coaching 16(3)

Table 3. Descriptive analysis of throw-in strategy (n ¼ 16,154), first contact success (n ¼ 13,376) and possession retained (n ¼ 8847)
in relation to throw-in length, direction, and pitch location.

Throw-in strategy First contact success Possession retained

Number of Number of Number of

Percentage throws Mean percentage throws Mean percentage throws

Throw in length
Short 19.4% 3134 97.3% 3050 62.8% 1920
Medium 41.7% 6736 89.3% 6020 64.0% 3859
Long 38.9% 6284 69.5% 4306 70.6% 3068
Direction
Backwards 29.7% 4805 99.5% 4781 83.5% 4044
Lateral 29.0% 4677 89.2% 4165 64.0% 2687
Forwards 41.3% 6672 67.6% 4430 48.3% 2116
Pitch location
Attacking 18 Yard 15.0% 2419 86.0% 2077 62.8% 1311
Rest of attacking half 42.1% 6793 87.7% 5942 69.6% 4140
Rest of defensive half 36.4% 5873 78.3% 4612 65.0% 3034
Defensive 18 Yard 6.6% 1069 69.0% 745 48.5% 362

Figure 2. Correlations between final league position and first contact success (a), possession retention for 7 s (b), mean time in
possession from the throw-in (c), and (d) throw-ins resulting in a shot from the possession achieved after a successful first contact.

contact success F(2.623, 49.830) ¼ 20.773, p < 0.001, There was a main effects for direction F(1.6,
gp2 ¼ .522. There was also a two-way interaction for 20.397) ¼ 537.408, p < 0.001, gp2 ¼ .966 and length
direction * length F(4, 76) ¼ 125.534, p < 0.001, gp2 ¼ F(1.599, 30.384) ¼ 218.496, p > 0.001, gp2 ¼ .920, but
.869 and location * direction F(1.708, 32.452) ¼ no main effect for location F(1, 19) ¼ 2.562, p < 0.05,
38.617, p < 0.001, gp2 ¼ .670 for first contact success. gp2 ¼ .119. Post-hoc tests showed differences between
But location * length was not significant F(1.5, all lengths (all p < 0.05) of the throw-in, with first con-
28.504) ¼ 1.964, p > 0.05, gp2 ¼ .094. tact success rate decreasing as throw-in length
Stone et al. 835

Figure 3. Correlations between final league position and percentage of throw-ins performed in the backwards (a), forwards (b) and
lateral (c) direction.

Figure 4. First contact success rate (percentage and absolute values) based on pitch location, throw-in direction and throw-in length.

increased. Post-hoc tests showed a significant differ- direction * length F(2.428, 46.130) ¼ 21.365, p < 0.001,
ence between the three direction (all p < 0.05) with gp2 ¼ .529 and location * direction F(2, 38) ¼ 4.221,
throwing the ball backwards (99.5%) resulting in the p < 0.05, gp2 ¼ .182. However, location * length was
highest first contact success rate with a 24.9% increase not significant F(2, 38) ¼ 2.069, p > 0.05, gp2 ¼ .098.
compared to throwing the ball forwards (74.6%). There was a main effect for direction F(2, 38) ¼
Hence, as Figure 4 demonstrates, when throwing back- 309.484, p < 0.001, gp2 ¼ .942. Post-hoc tests showed
wards, length of throw-in did not affect success rates a difference between the three direction (all p < 0.05)
regardless of location. However, when throwing for- with throwing backwards (83.0%) having higher suc-
wards, as the length increased, there was a reduction cess rates than lateral (67.7%) and forwards (50.3%).
in success, with the lowest success rate being forwards There was also a main effect for length F(1.418,
and long in the rest of the defensive half. 26.934) ¼ 9.90, p > 0.05, gp2 ¼ .343). Post-hoc tests
showed a difference between Short (70.5%) compared
to medium (66.2%) and long (64.3%) (p < 0.05). The
Retaining possession from a throw-in
main effect for location was not significant F(1, 19) ¼
The three-way repeated measure ANOVA showed an 0.406, p > 0.05, gp2 ¼ .021.
interaction for direction * length * location for posses- The three-way repeated measure ANOVA interac-
sion retention success F(2.647, 50.292) ¼ 4.02, tion for direction * length * location was non-
p < 0.05, gp2 ¼ .175 (see Figure 5). There was also a significant for mean time in possession F(2.428,
two-way interactions for possession retention for 46.139) ¼ 2.72, p ¼ 0.066, gp2 ¼ .125. However, there
836 International Journal of Sports Science & Coaching 16(3)

Figure 5. Possession retained success based on pitch location, throw-in direction and throw-in length. Percentage success, absolute
values and mean time in possession.

was an interaction for mean time in possession for Discussion

direction * length F(2.013, 38.25) ¼ 12.886, p < 0.001,
This research firstly examined if throw-in performance
gp2 ¼ .404, location * length F(2, 38) ¼ 5.154, p < 0.05,
was correlated to final league position, and then how
gp2 ¼ .213, and location * direction F(2, 38) ¼ 3.687,
throw-in direction and length affected first contact suc-
p < 0.05, gp2 ¼ .163.
cess rate, possession retention, and shot creation during
There was a main effect for direction F(2, 38) ¼
the English Premier league 2018-2019 season. On aver-
257.798, p < 0.001, gp2 ¼ .931. Post hoc analysis
age 43 throw-ins were taken per match, meaning
showed throwing backwards (24 s) had a longer mean
throw-ins occur more frequently than corner kicks,6
possession than forwards (13 s p < 0.001) or laterally
free kicks7,8 and goal kicks14 highlighting the influence
(19 s p < 0.001). Laterally also had a longer mean
throw-ins could have on professional soccer and a need
time than forwards (p < 0.001). The main effect for
for coaches to focus on this set piece. The importance
length was significant F(2, 38) ¼ 8.381, p > 0.05,
throw-ins could have on performance was indicated via
gp2 ¼ .306. Post hoc analysis showed longer time for
significant correlations with teams ranked highest in
short (20.0 secs) compared to long (17.8 secs)
the final league position having increased first contact
(p < 0.005). There was also a main effect for location
success, possession retention and shot creation. These
F(1, 19) ¼ 6.861, p > 0.05, gp2 ¼ .265 with longer time in
correlations suggest either first contact success, posses-
possession in the defensive half (19.4 secs) compared to
sion retention and shot creation results in teams win-
the attacking half (18.4 sec).
ning or drawing more matches (i.e. gaining more points
to be ranked higher in the league), or higher ranked
Throw-in resulting in a shot teams use more effective throw-in strategy resulting in
1053 throw-ins resulted in a shot originating from the greater success. To explore this further, initial data,
rest of the attacking and defensive area of the pitch suggested the differences in success rate were due to
after a first contact was won. Based on all throw-ins changes in throw-in strategy, with higher ranked
taken in the rest of the attacking and defensive half, an teams utilising backwards and lateral throw-ins more
interaction for direction * length for shot creation was often, in comparison to lower ranked teams favouring
shown F(4, 76) ¼ 3.230, p ¼ 0.029, gp2 ¼ .145. The main a forward throw-in.
effect of direction affected shot creation F(2, 38) ¼ To negate the effect increased skill level of higher
29.080, p < 0.001 gp2 ¼ .605. Post-hoc analysis showed ranked teams may have on throw-in outcome, we uti-
backwards (11.2%) and lateral (12.2%) throws were lised a repeated measure design to examine how specific
more likely to produce shots than forwards throws throw-in strategies influence success rates. From the
(6.6%). The main effect for length was not significant 16,154 throw-ins, 83% of throws resulted in a success-
F(2, 38) ¼ 3.054, p < 0.05, gp2 ¼ .138. ful first contact, 54% resulted in the team retaining
Stone et al. 837

possession for 7 seconds or longer, with a shot being opposition.1 In line with first contacts, throwing the
achieved 8.8% of the time from throw-ins after a suc- ball backwards had the highest association with retain-
cessful first contact. This is in line with previous data ing possession. In the rest of the defensive half throw-
analysing throw-ins from the MSL.11 The attention of ins that went backwards or laterally had the greatest
throw-in analysis has typically focused on long throw- success at retaining possession when thrown long and
ins within the attacking 18-yard box due to similarities decreased from medium to short length. Furthermore,
with corner kick set pieces.16 However, the results here examining the length of possession (of those throws
show 78.5% of throws come from the rest of the with possession retained for a minimum of 7 seconds)
attacking and defensive areas of the pitch. This high- shows mean time in possession was longest when
lights the importance throw-ins have on restarting, and throwing backwards (24 s) compared to forwards
then building a team’s possession in open play and (13 s). It’s suggested when throwing backwards or lat-
hence, we further explored these specific pitch locations erally, compared to throwing forwards, teams may not
in more detail. apply pressure high up the pitch allowing the receiver
When exploring the throw-in strategies used in the to secure possession with time and space to build an
rest of the attacking and defensive areas a clear pattern effective attack. From an opposition perspective, this
of findings emerged. The data here empirically sup- highlights the importance of applying pressure high up
ports anecdotal evidence that a common strategy is to the pitch, preventing the backwards or lateral throw-in.
throw the ball forwards and long15 with the most Therefore, reducing the likelihood of longer posses-
common direction of throw being forwards, at either sions and increasing the rate of turnovers from a for-
medium (10–20 yards) or long (20þ yards) distances. ward ‘fight ball’ throw-in.
However, the data suggested this throw-in strategy, In the rest of the attacking half, when throwing
although being the most common, is also the least backwards, the length of throw did not affect first con-
effective at both achieving a successful first contact tacts. Laterally, when throwing long, there was a reduc-
and retaining possession. As Siegle and Lames14 sug- tion in success rate. When throwing forwards, there
gested, the team not in ball possession frequently sees a was a reduction in success from (94.2%) a short
throw-in in the defensive area as an opportunity to length, compared to (59.3%) a long length. Hence, if
conquer the ball and create pressure. A possible expla- teams want to increase their chance of achieving a suc-
nation is when throwing the ball forwards, the opposi- cessful first contact, they should throw backwards, or
tion are set up in a compact shape, outnumbering the laterally, not forwards in the rest of the attacking half.
attacking team with defensive players. This results in a After first contact success, in the rest of the attacking
‘fight ball’ being thrown down the line into an unfav- half, there was also a significant association between
ourable situation and therefore in a loss of first contact the combined direction and length of throw on retain-
and ball possession retention. Furthermore, when ing possession. Throw-ins that went backwards had the
throwing forwards the aim might be for players to greatest success when thrown longer, however, length
head/flick the ball onto a teammate, however, as did not affect possession retained rates when throw-ins
Szczepa
nski and McHale22 demonstrate headed went laterally. Forward throw-ins again had the lowest
passes are less accurate and have a negative effect of success rates and decreased as the throw-in length
the following pass, in comparison to those passed from increased. There was also significant interaction of
the ground, hence might lead to a loss of possession. mean time in possession for direction and length. The
The results here, suggest the common coaching princi- direction showed throwing backwards (24 s) had a
ple of throwing the ball forwards and long away from longer mean possession than forwards (13 s) or laterally
the goal in the defensive half is an ineffective tactic. (18 s). With both first contact success and retaining of
In comparison when throwing backwards or lateral- possession demonstrating clear advantages for throw-
ly, the length of the throw did not affect first contact ing backwards or laterally, finally it was explored if
success rate. Results here demonstrate one way to these possessions resulted in more successful outcomes
relieve pressure in the rest of the defensive half is to (i.e. shots being created).
throw the ball long backwards with a 99.9% first con- When examining shot creation, after a successful
tact success rate, this is over double the success rate first contact, throw-ins which went backwards or later-
compared to throwing long and forwards (47.4%). ally had more chance of creating shots than throwing
A key element after a successful first contact, is the forwards. This provides further evidence to emphasise
team’s ability to retain possession within the central the importance of teams needing to show the compo-
areas of the pitch. This allows them to build either a sure to throw backwards and go against the common
successful attack or negate conceding possession and coached principles of throwing forwards down the line
defending a fast counter attacking situation from the towards the opposition’s half. Therefore, increasing
838 International Journal of Sports Science & Coaching 16(3)

both their time in possession but also a chance of cre- Acknowledgements

ating a shot. From a defensive perspective, the The authors would like to thank Dougie Wright for his help
common coached strategy has been to drop off and during the collection of the raw data presented within this
allow the opposition to throw the ball backwards or manuscript.
laterally to a position which is perceived to be a less
threatening area away from their own goal. However, Data availability statement
with the finding’s presented here, coaches should exam- The datasets generated during and/or analysed during the
ine their own strategy to consider if a different defen- current study are not publicly available but are available
sive strategy might be more effective. One possible from the corresponding author on reasonable request.
approach could look to force the opposition into
throwing the ball long and forwards which may result Declaration of conflicting interests
in regaining possession quicker allowing a counterat- The author(s) declared no potential conflicts of interest with
tack while also potentially conceding less shooting respect to the research, authorship, and/or publication of this
opportunities. article.
The findings here provide a starting point to support
the importance of coaches focusing on the use of Funding
throw-in strategy to increase possession and chance The author(s) received no financial support for the research,
creations within professional soccer. However, with authorship, and/or publication of this article.
limited published data, and one season’s data examined
here, the findings should be interpreted with caution ORCID iD
and there are many future areas of research that Joseph Antony Stone https://orcid.org/0000-0002-
should be examined to explore if similar patterns 9861-4443
emerge. Further comparison within and between
soccer leagues will enable a greater understanding on References
the importance throw-ins have on team performance. 1. Lago C and Mart
ın R. Determinants of possession of the
Furthermore, with backwards and lateral throw-ins ball in soccer. J Sports Sci 2007; 25: 969–974.
appearing to show an advantageous tactic in this data 2. Lago-Pe~ nas C, Lago-Ballesteros J, Dellal A, et al. Game-
set, further exploration to explain why throwing in related statistics that discriminated winning, drawing and
these directions has greater success rates should be losing teams from the Spanish soccer league. J Sports Sci
explored in more detail. For example, lower ranked Med 2010; 9: 288–293.
teams may be willing to concede possession against a 3. Yue Z, Broich H and Mester J. Statistical analysis for the
higher ranked team, so they remain in a compact defen- soccer matches of the first Bundesliga. Int J Sports Sci
Coach 2014; 9: 553–560.
sive shape and do not overcommit players with their set
4. Zhao YQ and Zhang H. Analysis of goals in the English
up on the throw-in. Hence, further evaluation on what Premier League. Int J Perform Anal Sport 2019; 19:
constitutes a successful throw-in needs investigating. 820–831.
Finally, investigating if score line, or individual 5. Yiannakos A and Armatas V. Evaluation of the goal
match outcome is influenced by throw-in strategy will scoring patterns in European Championship in Portugal
aid future coaching practice. 2004. Int J Perform Anal Sport 2006; 6: 178–188.
6. Strafford BW, Smith A, North JS, et al. Comparative
analysis of the top six and bottom six teams’ corner
Conclusion kick strategies in the 2015/2016 English Premier
In conclusion, results here suggest throw-in success League. Int J Perform Anal Sport 2019; 19: 904–918.
may be associated with final league performance. 7. Casal CA, Maneiro R, Ardá T, et al. Effectiveness of
This data highlights to coaches how throw-in tactics indirect free kicks in elite soccer. Int J Perform Anal
might affect first contact success rates, possession Sport 2014; 14: 744–760.
8. Link D, Kolbinger O, Weber H, et al. A topography of
retention and shot creation in professional soccer.
free kicks in soccer. J Sport Sci 2016; 34: 2312–2320.
Findings demonstrate, throwing the ball laterally or
9. Almeida CH, Volossovitch A and Duarte R. Penalty kick
backwards can increase throw-in success rates in com- outcomes in UEFA club competitions (2010-2015): the
parison to throwing the ball forwards. Furthermore, roles of situational, individual and performance factors.
higher ranked teams utilised this strategy more often Int J Perform Anal Sport 2016; 16: 508–522.
and coaches could examine their current throw-in strat- 10. International Football Association Board. Laws of the
egies to see if implementing changes may link to an game 2019/20, https://www.theifab.com (2019, accessed
overall improved team performance. 25 November 2020).
Stone et al. 839

11. McKinley E. Game of throw-ins. American soccer anal- 17. Stanculescu G, Melenco I and Popa C. A comparative
ysis, https://www.americansocceranalysis.com/home/ study on the evolution of the parameters in professional
2018/11/27/game-of-throw-ins (2018, accessed 3 May soccer matches. Procedia Soc Behav Sci 2014; 127: 63–67.
2020). 18. Wallace JL and Norton KI. Evolution of World Cup
12. Casal CA, Maneiro R, Arda T, et al. Analysis of corner soccer final games 1966–2010: game structure, speed
kick success in Elite Soccer. Int J Perform Anal Sport and play patterns. J Sci Med Sport 2014; 17: 223–228.
2015; 15: 430–451. 19. Bascombe C. Jurgen Klopp appoints ‘throw-in’ coach at
13. Statsbomb. Database for the English Premier League Liverpool. The Telegraph, 9 June 2018, https://www.tele
during the 2018-2019 season, https://statsbomb.com graph.co.uk/football/2018/08/26/jurgen-klopp-appoints-
(2020, accessed 20 October 2020). throw-in-coach-liverpool
14. Siegle M and Lames M. Game interruptions in elite 20. Fleiss JL. Statistical methods for rates and proportions
soccer. J Sport Sci 2012; 30: 619–624. (2nd ed.). New York, NY: John Wiley, 1981.
15. Linthorne NP and Thomas JM. The effect of ball spin 21. Statsbomb. StatsBomb Specification, https://statsbomb.
rate on distance achieved in a long soccer throw-in. Pro com (2019, accessed 11 January 2021).
Eng 2016; 147: 677–682. 22. Szczepa
nski Ł and McHale I. Beyond completion rate:
16. Kline LE and Samonisky M. The soccer throw-in. J Phys evaluating the passing ability of footballers. J R Stat Soc
Educ Rec 1981; 52: 57–59. A 2016; 179: 513–533.