soccer research paper week 50 2021

Pools RSK Papers – Soccer, Capital, Bob Morton

Pools rsk papers, rsk pool papers ,rsk pool papers this week, pool rsk paper for this week, soccer research pool paper, Bob Morton, rsk pool papers, rsk papers this week.

Week 39 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 39 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 38 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 38 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 37 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 37 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 36 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

  Week 36 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, ...

Week 35 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 35 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 34 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 34 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 33 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 33 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 32 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 32 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 31 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 31 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 30 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 30 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 29 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 29 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 28 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 28 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 27 Pool RSK Papers 2024: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 27 Pools RSK Papers 2024: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 26 Pool RSK Papers 2023: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 26 Pools RSK Papers 2023: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 25 Pool RSK Papers 2023: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 25 Pools RSK Papers 2023: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 24 Pool RSK Papers 2023: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 24 Pools RSK Papers 2023: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 23 Pool RSK Papers 2023: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 23 Pools RSK Papers 2023: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 22 Pool RSK Papers 2023: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 22 Pools RSK Papers 2023: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

Week 21 Pool RSK Papers 2023: Bob Morton, Capital Intl, Soccer X Research, BigWin

Week 21 Pools RSK Papers 2023: Soccer X Research, Bob Morton, Capital Intl, Winstar, BigWin ...

• Review Article
• Open access
• Published: 21 September 2020

Reducing Injuries in Soccer (Football): an Umbrella Review of Best Evidence Across the Epidemiological Framework for Prevention

• Oluwatoyosi B. A. Owoeye   ORCID: orcid.org/0000-0002-5984-9821 1 , 2 ,
• Mitchell J. VanderWey 1 &
• Ian Pike 3 , 4  

Sports Medicine - Open volume  6 , Article number:  46 ( 2020 ) Cite this article

38k Accesses

41 Citations

44 Altmetric

Metrics details

Soccer is the most popular sport in the world. Expectedly, the incidence of soccer-related injuries is high and these injuries exert a significant burden on individuals and families, including health and financial burdens, and on the socioeconomic and healthcare systems. Using established injury prevention frameworks, we present a concise synthesis of the most recent scientific evidence regarding injury rates, characteristics, mechanisms, risk and protective factors, interventions for prevention, and implementation of interventions in soccer. In this umbrella review, we elucidate the most recent available evidence gleaned primarily from systematic reviews and meta-analyses. Further, we express the exigent need to move current soccer injury prevention research evidence into action for improved player outcomes and widespread impact through increased attention to dissemination and implementation research. Additionally, we highlight the importance of an enabling context and effective implementation strategies for the successful integration of evidence-based injury prevention programs into real-world soccer settings. This narrative umbrella review provides guidance to inform future research, practice, and policy towards reducing injuries among soccer players.

This review provides a one-stop evidence reference regarding the prevention of soccer injuries, including evidence and perspectives on the implementation of proven interventions.

Overall evidence supports the use of the 11+ neuromuscular training warm-up and focused strength training, and there is emerging evidence for load management programs to mitigate injury risk among soccer players.

Theory-driven dissemination and implementation studies are needed to improve the adoption, adherence, appropriate adaptation, scale-up, and sustainment of evidence-based injury prevention interventions in soccer.

The findings from this review provide guidance to inform future research, practice, and policy towards reducing injuries among soccer players.

Soccer (football) is the most popular sport in the world [ 1 ], with some 270 million involved in the sport worldwide in 2006 [ 2 ]. For approximately 110,000, it is a profession and thus a source of income; for some 38 million registered players, it is a team game organized within leagues and competitions; and for about 226 million others, it is an enjoyable exercise surrogate for fitness and health [ 2 ]. The health benefits of soccer as “medicinal exercise” are well documented, for example, improved cardiovascular health, mental health, and bone health [ 3 ]. However, there is a paradoxical negative effect of soccer on health when players get injured (e.g., obesity or post-traumatic osteoarthritis after an anterior cruciate ligament injury) [ 4 , 5 ]. Furthermore, soccer injuries exert a significant burden on socioeconomic and healthcare systems [ 6 ]. Founded on established epidemiological frameworks describing the sequence of research steps to effective injury prevention practice [ 7 , 8 ]—from identifying injury rates to the implementation of effective interventions—we present a narrative umbrella review that articulates best available evidence to inform guidelines, practice, and policy towards mitigating the risk of injuries in soccer, and in turn maximizing the benefits of participation among individuals.

To achieve the above-mentioned purpose, we conducted methodical searches across five databases (MEDLINE, SPORTDiscus, PsycINFO, CINAHL, and Cochrane Database of Systematic Reviews) from January 2010 to January 2020 to identify all systematic reviews, meta-analyses, reviews, and original research (where limited or no reviews were available) across soccer injury studies that investigated injury incidence, characteristics, mechanisms, risk and protective factors, interventions for prevention, and implementation and evaluation of interventions. A summary of the search records for our primary source of data (systematic and narrative reviews) is presented in Table 1 , and details of the search terms used—key concepts and search words—are presented in an additional file ( Supplementary File ). Our search strategy involved the use of relevant search descriptors of “OR” and “AND” to combine search/key words and key concepts, respectively, after each search word was exploded (exp) to capture all literature possible. Search records were limited to articles with full text, written in the English language, and relating to humans. The same methodology was used to obtain primary research articles where no reviews were available.

Injury Rates

Injury incidence among soccer players differs across levels of participation, age, type of exposure, and sex. The incidence of injuries in soccer is mostly significant during games/matches, ranging from 9.5 to 48.7 injuries/1000 h among competitive male youth players, 2.5 to 8.7 injuries/1000 h among male professional players, and 12.5 to 30.3 injuries/1000 h among female players [ 9 , 10 , 11 , 12 ] (Table 2 ). The incidence of injuries appears higher among males vs. females, and injury incidence is higher during games/matches vs. practice/training for all participation categories, among both male and female players [ 10 , 11 , 12 ]. Soccer players younger than 12 years of age have a lower injury rate (1.0–1.6 injuries per 1000 h) compared to older players [ 9 ].

Injury Location and Type

Most soccer injuries occur in the lower limbs (60–90%), especially the ankle, knee, and thigh [ 10 , 11 , 12 , 13 , 14 ]. Among male players, the most common injuries affect the hamstring muscles followed by the ankle, knee, and groin [ 11 , 13 ]. Comparably, among female players, knee and ankle injuries are the most common, followed by thigh/hamstring injuries [ 10 , 13 ].

Thigh, Knee, and Ankle Injuries

Most thigh injuries result from strains with a high proportion of hamstring injuries, despite quadriceps injuries leading to longer absence from play [ 15 ]. The prevalence and history of hamstring injury is greater among adult professional players (40%) compared to under-20 players (18%) [ 16 ]. Up to 18% of severe soccer injuries presenting at hospital emergency departments involve the knee [ 17 ]. One such injury involves the anterior cruciate ligament (ACL). The ACL injury rate among females (2.0/10,000 athlete exposures) is 2.2 times higher than that of males (0.9/10,000 athlete exposures), independent of participation level [ 18 ]. Ankle injuries account for up to 20% of all soccer injuries with ankle sprains constituting 77% of all ankle injuries [ 14 , 19 ].

The prevalence of concussion in youth soccer appears to be relatively low with an incidence of 0.19 (95% CI 0.16–0.21) concussions per 1000 athletic exposures and 0.27 (95% CI 0.24–0.30) concussions per 1000 athletic exposures among male and female players, respectively [ 20 ]. A higher concussion incidence has been consistently reported among females [ 10 , 20 ].

Injury Mechanisms

Overall, about two-thirds of soccer injuries are traumatic and the other one-third (27–33%) are caused by overuse [ 11 , 12 , 21 ]. These findings are based on a medical attention/time-loss injury definition, and emerging evidence from studies using an all-complaint injury definition suggests that overuse onset injuries may be as prevalent as acute onset injuries [ 22 ]. About two-thirds of traumatic injuries are contact injuries, of which 12–28% are caused by foul play. Notably, non-contact injuries account for 26–58% of all injuries [ 13 , 21 ]. Injuries occur primarily during the initial or final 15 min of the match, indicating the significance of an appropriate warm-up and the effects of fatigue on players [ 23 ].

Risk and Protective Factors

Non-modifiable risk factors, player position.

Goalkeepers are at a lower overall risk of injury compared to outfield players in the male game [ 24 ]. Independent of goalkeepers, current evidence is inconsistent regarding the association between player position and injury risk; however, it appears that strikers may be at a greater risk as compared with other outfield players during matches [ 24 ].

Previous Injury

A history of previous injury continues to be the most consistent and strongest risk factor for future injury, and this also holds true for specific injuries [ 9 , 25 , 26 , 27 , 28 , 29 ]. For example, a history of previous hamstring injury is associated with future hamstring injury among male players [ 25 , 28 ], previous ACL injury is associated with risk of future ACL injury [ 29 ], and previous ankle sprain injury is related to the emergence of new ankle sprain injuries [ 27 ].

Current evidence regarding age as a risk factor for soccer injury is limited. One systematic review suggested that increasing age was a risk factor for future hamstring injury among male players [ 25 ]. Another systematic review concluded that existing literature was insufficient to infer any relationship between age and the risk of ACL injury among soccer players [ 29 ]. In a single prospective study, age > 14 years was a significant risk factor for future acute knee injury among female players [ 30 ].

Familial predisposition for ACL injury is associated with increased risk of ACL injury and acute knee injury [ 29 , 30 ].

Overall, the incidence of injuries is higher among males vs. females [ 10 , 11 ]; however, female sex is associated with increased ACL injury risk [ 29 ].

Competitive Setting

Game exposure demonstrated increased injury risk compared to practice for both male and female soccer players [ 29 , 31 ]. Furthermore, within the practice setting, the risk of injury is higher for scrimmage compared to normal practice and walk-through [ 29 ].

Shoe-Surface Interaction

Current research suggests there is an association between higher shoe-surface interaction and increased ACL injury risk [ 29 ].

Pre-season Knee Complaints

Females presenting with pre-season knee complaints appear to be at increased risk for acute knee injury during the season [ 30 ].

Early Sport Specialization

Though there is a lack of substantive evidence for soccer specifically, early sport specialization has been found to be associated with a greater risk for overuse injuries across multiple youth sports [ 9 ]. One study showed that female soccer players 12–15 years of age playing on more than one team had increased risk for lower extremity overuse injuries [ 32 ].

Growth and Leg Length

Elite male youth soccer players are at greater risk for traumatic injury in the year of peak height velocity [ 33 ]. A recent prospective study of male soccer players aged 10–12 years shows an association between an increase in leg length throughout the season and risk for overuse injury [ 34 ]. The same study suggests an association between longer leg length and risk of overuse injury among male soccer players aged 13–15 years. Additionally, they found a higher weight and a decreased growth rate to be associated with an increased risk of acute injury.

Modifiable Risk Factors

Evidence regarding load-injury relationships among soccer players is still emerging as reviews remain sparse in this area of inquiry. Current evidence across team sports indicates that load, in terms of player exposure and/or exertion, could either be an independent protective or risk factor for injury, depending on whether load administration is optimal and progressive or suboptimal (e.g., load spike), respectively, and that this relationship is likely moderated by other risk factors for injury [ 35 , 36 , 37 , 38 , 39 , 40 ]. Prospective studies showed that a high amount of absolute (accumulated or cumulative) load, based on different calculations of load measures (e.g., 1-weekly, 2-weekly), was associated with greater risk of injury among elite youth and professional soccer players [ 39 , 40 , 41 ]. These findings suggest that it may be expedient to have an absolute load threshold, for example, weekly load threshold, to further mitigate injury risk in soccer, especially youth soccer [ 39 , 40 ]. Altogether, available evidence suggests that avoiding a spike in load (e.g., the acute to chronic workload ratio) is associated with less soccer injuries [ 39 , 40 , 41 ].

Neuromuscular Factors

Hamstring/quadriceps strength ratio imbalance is a key risk factor for hamstring muscle injury; specifically, decreased hamstring strength relative to quadriceps strength is a risk factor for knee ligamentous injuries in both male and female youth soccer players [ 29 , 42 ]. Decreased single leg hop distance is also associated with increased hamstring injury risk [ 43 ]. While current evidence is inconclusive for muscle strength asymmetry (i.e., right vs. left) as a risk factor, eccentric hamstring strength asymmetry is specifically indicated as a key predictor of injury among male youth soccer players [ 26 ]. Furthermore, eccentric hamstring strength (< 256 N) and single leg hamstring bridge scores of less than 20 reps on the right leg are associated with increased risk of hamstring strain [ 43 ]. Poor landing mechanics, specifically, increased dynamic knee valgus, is associated with increased risk for lower limb injury, including ACL injury [ 9 , 42 , 43 ]. Leg dominance and leg asymmetry also relates to increased risk of injury; a difference of 15% or greater, between an individual’s dominant and non-dominant limb, has been shown to predict future injury [ 42 ]. An asymmetry of greater than 4 cm on the anterior reach portion of the Y-balance test places athletes at 2.5 times greater risk for injury among male youth soccer players [ 42 , 43 ]. Hip external rotation strength scores using handheld dynamometry of less than 18% of the individual’s body weight is associated with lower extremity and back injuries [ 43 ]. Additionally, the literature suggests that the risk of injury may increase with altered neuromuscular firing during dynamic movements like cutting or landing, and dynamic stability deficits may increase lower extremity injury risk for male youth soccer players [ 42 ].

Protective Factors

Although mention of protective factors in review level evidence did not exist at the time of this evidence review, findings from original research previously described (under modifiable risk factors) signify load management as a viable target for mitigating injury risk in soccer. For example, an in-season relative load measure of acute to chronic workload ratio of 1 to 1.25 significantly reduced injuries among youth players [ 40 ], and a reduced absolute load significantly reduced injuries among youth and adult professional players [ 39 , 40 ]. Additionally, current evidence suggests that improved neuromuscular capacity and control, including increased quadriceps, hamstring, hip flexor strength, and movement control are protective against injuries among soccer players [ 9 , 26 , 29 , 42 , 43 ].

Opportunities for Prevention

Effective interventions.

Drawing from available evidence regarding modifiable risk factors and protective factors for soccer injuries, injury prevention experts have developed and tested interventions for reducing musculoskeletal injuries in soccer. There is extensive high-quality evidence (including two reviews of systematic reviews) showing the clinical effectiveness of exercise-based interventions in the form of neuromuscular training (NMT) warm-up programs in reducing all soccer-related injuries across sex, ages, and skill levels. Specifically, the 11+ (formerly called the FIFA 11+) warm-up program reduces overall injury rate (i.e., all injuries) by 30 to 47% [ 23 , 44 , 45 , 46 ], lower limb injury rate by 39 to 44% [ 44 , 45 ], overuse injury rate by 55%, and knee injury rate by 52% [ 47 ]. Emerging evidence also suggests that the 11+ Kids (a version for children under 12 years old) is efficacious (48% reduction for all injuries) for reducing injuries in younger players [ 48 ]. Additionally, the “Knee Injury Prevention Program” (KIPP) has the potential to significantly reduce non-contact lower limb injury and overuse injury among young female soccer players by 50% and 56%, respectively [ 47 ].

In a recent systematic review, the application of a variety of exercise-based injury prevention programs for youth players was found to reduce injury rates by up to 46% [ 49 ]. Furthermore, the risk of hamstring injuries can be reduced by up to 51% when the Nordic Hamstring exercise is implemented in isolation [ 50 ]. A recent meta-analysis showed that ankle injuries can be reduced by as much as 40% [ 51 ] and a meta-analysis of meta-analyses [ 52 ] demonstrated that a 50% reduction can be achieved for all ACL injuries in a heterogeneous sample of athletes, including soccer players, when NMT warm-up is implemented.

Specific instructions on how to perform aforementioned NMT warm-up programs can be found in the International Olympic Committee’s “Get Set” app, an innovative and accessible mobile app that provides continued access to illustrative and video information regarding effective sport- and body-specific NMT warm-up programs, including the 11+ program. The 11+ program can also be accessed from the following website: https://www.youtube.com/watch?v=RSJIp7e7fyY

Although concussions are not frequent in soccer, sustaining a concussion may present severe and lasting negative health consequences [ 53 ]. It is important for coaches, parents, and administrators to be aware of concussion signs and symptoms and know what to do if concussion is suspected. For concussion prevention, there is evidence that education about concussion among key stakeholders, e.g., coaches, referees, and parents, can reduce the incidence of concussion and facilitate improved outcomes [ 54 ]. Interventions for primary (e.g., rule change and avoiding a slippery playing surface) and secondary (e.g., concussion recognition and decision on return to playtime) prevention are mainly informational for coaches and parents/guardians. A popular evidence-based educational tool is the Concussion Awareness Training Tool, available at https://cattonline.com .

Cost-Effectiveness of Interventions

Literature regarding the cost-effectiveness of injury prevention interventions in soccer is limited. A reduction of 43% was reported in healthcare costs in the training group that underwent an NMT warm-up similar to the 11+ program with additional use of a wobble-board, when compared to a standard practice control group [ 55 ]. Similarly, the “11+ Kids” program showed a 51% reduction in healthcare costs when compared with a regular warm-up [ 56 ].

Implementation and Evaluation

Literature regarding the evaluation of the implementation of efficacious/effective interventions such as the 11+ and other NMT warm-up programs is advancing despite the lack of reviews [ 57 , 58 , 59 , 60 , 61 , 62 , 63 , 64 , 65 , 66 , 67 ]. However, of all the studies currently available, only two reported using an implementation framework to evaluate a preventative program. The Reach Effectiveness Adoption Implementation Maintenance Framework was used in both studies: one to evaluate an NMT warm-up program for knee/ACL prevention, and the other to evaluate the Adductor Strengthening Program for groin injury prevention [ 57 , 59 ]. Overall, the execution of NMT warm-up programs when implemented ranged between low and moderate [ 60 , 68 ].

To improve the spread and implementation of evidence-based injury prevention intervention in soccer, an understanding of implementation contexts is imperative. Although more rigorous theory-driven studies are needed to further understand potential contextual moderators of successful/unsuccessful implementation, a small number of studies have investigated perceived facilitators and barriers to NMT programs across levels of soccer participation (Table 3 ).

Current Best Practices for Implementation

Literature regarding best practices for onward translation of evidence-based injury prevention programs into routine practice in community and professional soccer remains sparse, and the urgent need for research in this field of inquiry has been identified [ 70 ]. The following conclusions have been reached in existing literature:

Preseason structured coaching workshops have the potential to effectively increase coach attitudes, perceived behavioral control, self-efficacy, and intention and subsequent implementation of NMT programs [ 64 , 71 , 72 ]. However, it remains unclear whether high levels of behavioral determinants, i.e., cognitive and psychosocial factors, would ultimately result in high levels of program adherence and maintenance over time [ 57 ].

Coach-led delivery of the 11+ appears to be relatively sufficient in implementing the program; evidence on the advantage of having additional support or supervision from research or team staff, e.g., strength and conditioning coach, an athletic trainer, or physiotherapist, is mixed [ 57 , 60 , 71 ].

For maximum effectiveness, coaches need to ensure quality delivery to their teams by performing NMT warm-up exercises with proper technique and adhering to the program guidelines, while adapting it to fit their local setting. A minimum of 2× weekly appears to be optimal and thereby recommended [ 58 , 61 ].

Quality implementation requires soccer associations and organizations at the federal, provincial, and community levels to enact policies that enforce injury prevention programs and education and policies that require coaches to use proven NMT warm-up programs such as the 11+ [ 60 , 73 , 74 ].

Conclusions and Call to Action

This review provides guidance to inform future research, policy, and practice towards reducing injuries among soccer players. It presents a one-stop evidence reference regarding the burden, etiology, and prevention of soccer injuries, including current opportunities for evidence-based interventions and their implementation. To achieve desired outcomes and population-level impact from injury prevention research evidence, evidence-based interventions need enabling contexts and effective implementation strategies for a successful integration into real-world settings. Consequently, innovators (e.g., researchers) and implementation actors at the organizational (e.g., football associations, government/public health agencies, non-profit organizations, football clubs) and individual (e.g., coaches, strength and conditioning personnel, medical staff) levels have critical roles to play and are urged to rise to the occasion.

Researchers need to acquire an appreciable level of proficiency in dissemination and implementation research designs to build upon current literature to advance dissemination and implementation science in soccer injury prevention. Specifically, theory-driven dissemination and implementation studies are needed to improve the adoption, adherence, appropriate adaptation, delivery, scale-up, and sustainment of evidence-based injury prevention interventions such as the 11+ in soccer. Researchers should move beyond randomized controlled trials evaluating efficacy in NMT programs (considering that there is extensive evidence supporting NMT efficacy ) to evaluating strategies for implementation in randomized controlled and pragmatic (e.g., quasi-experimental) trials. Further, researchers should use current information on implementation barriers to and facilitators of evidence-based interventions and knowledge from implementation science to conceptualize and test potential implementation strategies. In addition, soccer organizations and their staff, especially coaches, have the obligation of ensuring safety among their players. Collectively, researchers, knowledge brokers, policymakers, leaders, and administrators in soccer and other related organizations need to work collaboratively to move current injury prevention evidence into action in order to protect players’ current and future health.

Availability of data and materials

Data sharing is not applicable to this article as no datasets were generated during the current study.

Abbreviations

Anterior cruciate ligament

Neuromuscular training

Top-10 trending most popular sports in America 2019 | SportyTell [Internet].

FIFA Big Count 2006: 270 million people active in football [Internet]. 2007 [accessed 2020 Apr 1]. https://www.fifa.com/mm/document/fifafacts/bcoffsurv/bigcount.statspackage_7024.pdf .

Krustrup P, Dvorak J, Junge A, Bangsbo J. Executive summary: the health and fitness benefits of regular participation in small-sided football games. Scand J Med Sci Sports. 2010.

Lohmander LS, Englund PM, Dahl LL, Roos EM. The long-term consequence of anterior cruciate ligament and meniscus injuries: osteoarthritis. Am J Sports Med. 2007.

Toomey CM, Whittaker JL, Nettel-Aguirre A, Reimer RA, Woodhouse LJ, Ghali B, et al. Higher fat mass is associated with a history of knee injury in youth sport. J Orthop Sports Phys Ther. 2017;.

Fuller CW. Assessing the return on investment of injury prevention procedures in professional football. Sports Med. 2019.

van Mechelen W, Hlobil H, Kemper HCG. Incidence, severity, aetiology and prevention of sports injuries: a review of concepts. Vol. 14, Sports Medicine: An International Journal of Applied Medicine and Science in Sport and Exercise. 1992. p. 82–99.

Finch C. A new framework for research leading to sports injury prevention. J Sci Med Sport. 2006;9(1–2):3–9.

Article   Google Scholar  

Watson A, Mjaanes JM. Soccer injuries in children and adolescents. Pediatrics. 2019.

Junge A. Epidemiology in female football players. In: Football traumatology: new trends: Second Edition. 2015.

Pfirrmann D, Herbst M, Ingelfinger P, Simon P, Tug S. Analysis of injury incidences in male professional adult and elite youth soccer players: a systematic review. J Athl Train (Allen Press. 2016 May;51(5):410–424.

López-Valenciano A, Ruiz-Pérez I, Garcia-Gómez A, Vera-Garcia FJ, De Ste CM, Myer GD, et al. Epidemiology of injuries in professional football: a systematic review and meta-analysis. Br J Sports Med. 2020.

Junge A, Dvorak J. Soccer injuries: a review on incidence and prevention. Sports Med. 2004;34(13):929–38.

Feria-Arias E, Boukhemis K, Kreulen C, Giza E. Foot and ankle injuries in soccer. Am J Orthop (Belle Mead NJ). 2018;.

Pfirrmann D, Herbst M, Ingelfinger P, Simon P, Tug S. Analysis of injury incidences in male professional adult and elite youth soccer players: a systematic review. J Athl Train. 2016.

Ribeiro-Alvares JB, Dornelles MP, Fritsch CG, de Lima-e-Silva FX, Medeiros TM, Severo-Silveira L, et al. Prevalence of hamstring strain injury risk factors in professional and under-20 male football (soccer) players. J Sport Rehabil. 2019.

Roth TS, Osbahr DC. Knee injuries in elite level soccer players. American journal of orthopedics (Belle Mead, N.J.). 2018.

Montalvo AM, Schneider DK, Silva PL, Yut L, Webster KE, Riley MA, et al. “What’s my risk of sustaining an ACL injury while playing football (soccer)?” A systematic review with meta-analysis. British Journal of Sports Medicine. 2019.

Fong DT-P, Hong Y, Chan L-K, Yung PS-H, Chan K-M. A systematic review on ankle injury and ankle sprain in sports. Sports Med. 2007;37(1):73–94.

Pfister T, Pfister K, Hagel B, Ghali WA, Ronksley PE. The incidence of concussion in youth sports: a systematic review and meta-analysis. Br J Sports Med. 2016 Mar;50(5):292–7.

Bizzini M, Dvorak J. FIFA 11+: an effective programme to prevent football injuries in various player groups worldwide-a narrative review. Br J Sports Med. 2015;49(9).

Harøy J, Clarsen B, Thorborg K, Hölmich P, Bahr R, Andersen TE. Groin problems in male soccer players are more common than previously reported. Am J Sports Med. 2017;45(6):1304–8.

Sadigursky D, Braid JA, De Lira DNL, Machado BAB, Carneiro RJF, Colavolpe PO. The FIFA 11+ injury prevention program for soccer players: a systematic review. BMC Sports Sci Med Rehabil. 2017.

Della Villa F, Mandelbaum BR, Lemak LJ. The effect of playing position on injury risk in male soccer players: systematic review of the literature and risk considerations for each playing position. American journal of orthopedics (Belle Mead, N.J.). 2018.

Hughes T, Sergeant JC, Parkes MJ, Callaghan MJ. Prognostic factors for specific lower extremity and spinal musculoskeletal injuries identified through medical screening and training load monitoring in professional football (soccer): a systematic review. BMJ Open Sport Exerc Med. 2017.

McCall A, Carling C, Davison M, Nedelec M, Le Gall F, Berthoin S, et al. Injury risk factors, screening tests and preventative strategies: a systematic review of the evidence that underpins the perceptions and practices of 44 football (soccer) teams from various premier leagues. Br J Sports Med. 2015 May;49(9):583–9.

Owoeye OBA, Palacios-Derflingher LM, Emery CA. Prevention of ankle sprain injuries in youth soccer and basketball: effectiveness of a neuromuscular training program and examining risk factors. Clin J Sport Med. 2018;28(4):325–31.

van Beijsterveldt AMC, van de Port IGL, Vereijken AJ, Backx FJG. Risk factors for hamstring injuries in male soccer players: a systematic review of prospective studies. Scand J Med Sci Sports. 2013;23(3):253–62.

Volpi P, Bisciotti GN, Chamari K, Cena E, Carimati G, Bragazzi NL. Risk factors of anterior cruciate ligament injury in football players: a systematic review of the literature. Muscles, Ligaments and Tendons Journal. 2016.

Hägglund M, Waldén M. Risk factors for acute knee injury in female youth football. Knee Surgery, Sport Traumatol Arthrosc. 2016;24(3):737–46.

Emery CA, Meeuwisse WH, Hartmann SE. Evaluation of risk factors for injury in adolescent soccer: implementation and validation of an injury surveillance system. Am J Sports Med. 2005;33(12):1882–91.

O’Kane JW, Neradilek M, Polissar N, Sabado L, Tencer A, Schiff MA. Risk factors for lower extremity overuse injuries in female youth soccer players. Orthop J Sport Med. 2017.

Read P, Oliver JL, De Ste Croix MBA, Myer GD, Lloyd RS. Injury risk factors in male youth soccer players. Strength Cond J (Lippincott Williams Wilkins). 2015 Oct;37(5):1–7.

Google Scholar  

Rommers N, Rössler R, Goossens L, Vaeyens R, Lenoir M, Witvrouw E, et al. Risk of acute and overuse injuries in youth elite soccer players: body size and growth matter. J Sci Med Sport. 2020.

Gabbett TJ. The training-injury prevention paradox: should athletes be training smarter and harder? Br J Sports Med. 2016;.

Malone S, Hughes B, Doran DA, Collins K, Gabbett TJ. Can the workload–injury relationship be moderated by improved strength, speed and repeated-sprint qualities? J Sci Med Sport. 2019;22(1).

Gabbett TJ, Hulin BT, Blanch P, Whiteley R. High training workloads alone do not cause sports injuries: how you get there is the real issue. Br J Sports Med. 2016;50(8):444–5.

Owoeye OBA. Digging deep into the etiology of basketball injuries: a complex systems approach for risk mitigation. In: Lavar L, Kocaoglu B, Bytomski J, Cole B, Arundale A AN, editor. The basketball sports medicine and science book. Springer; 2020.

Watson A, Brickson S, Brooks A, Dunn W. Subjective well-being and training load predict in-season injury and illness risk in female youth soccer players. Br J Sports Med. 2017.

Bowen L, Gross AS, Gimpel M, Li F-X. Accumulated workloads and the acute:chronic workload ratio relate to injury risk in elite youth football players. Br J Sports Med. 2016;bjsports-2015-095820.

Malone S, Owen A, Newton M, Mendes B, Collins KD, Gabbett TJ, et al. The acute:chonic workload ratio in relation to injury risk in professional soccer. J Sci Med Sport. 2016;0(0):646–8.

Read PJ, Oliver JL, De Ste Croix MBA, Myer GD, Lloyd RS. Neuromuscular risk factors for knee and ankle ligament injuries in male youth soccer players. Sports Medicine. 2016.

Read PJ, Oliver JL, De Ste Croix MBA, Myer GD, Lloyd RS. A review of field-based assessments of neuromuscular control and their utility in male youth soccer players. Journal of Strength and Conditioning Research. 2019.

Al Attar WSA, Alshehri MA. A meta-analysis of meta-analyses of the effectiveness of FIFA injury prevention programs in soccer. Scand J Med Sci Sports. 2019.

Al Attar WSA, Soomro N, Pappas E, Sinclair PJ, Sanders RH. How effective are F-MARC injury prevention programs for soccer players? A systematic review and meta-analysis. Sports Med. 2016;46(2):205–17.

Thorborg K, Krommes KK, Esteve E, Clausen MB, Bartels EM, Rathleff MS. Effect of specific exercise-based football injury prevention programmes on the overall injury rate in football: a systematic review and meta-analysis of the FIFA 11 and 11+ programmes. Br J Sports Med. 2017;bjsports-2016-097066.

Herman K, Barton C, Malliaras P, Morrissey D. The effectiveness of neuromuscular warm-up strategies, that require no additional equipment, for preventing lower limb injuries during sports participation: a systematic review. BMC Med. 2012.

Rössler R, Junge A, Bizzini M, Verhagen E, Chomiak J, aus der Fünten K, et al. A multinational cluster randomised controlled trial to assess the efficacy of ‘11+ Kids’: a warm-up programme to prevent injuries in children’s football. Sport Med. 2018;.

Hanlon C, Krzak JJ, Prodoehl J, Hall KD. Effect of injury prevention programs on lower extremity performance in youth athletes: a systematic review. Sport Heal A Multidiscip Approach. 2020;12(1):12–22.

Al Attar WSA, Soomro N, Sinclair PJ, Pappas E, Sanders RH. Effect of injury prevention programs that include the Nordic hamstring exercise on hamstring injury rates in soccer players: a systematic review and meta-analysis. Sports Med. 2017;47(5):907–16.

Grimm NL, Jacobs JC, Kim J, Amendola A, Shea KG. Ankle injury prevention programs for soccer athletes are protective: a level-I meta-analysis. J Bone Joint Surg (Am Vol). 2016.

Webster KE, Hewett TE. Meta-analysis of meta-analyses of anterior cruciate ligament injury reduction training programs. J Orthop Res. 2018.

Gardner RC, Yaffe K. Epidemiology of mild traumatic brain injury and neurodegenerative disease. Molecular and Cellular Neuroscience. 2015.

Tator CH. Sport concussion education and prevention. J Clin Sport Psychol. 2012.

Marshall DA, Lopatina E, Lacny S, Emery CA. Economic impact study: neuromuscular training reduces the burden of injuries and costs compared to standard warm-up in youth soccer. Br J Sports Med. 2016;bjsports-2015-095666-.

Rossler R, Verhagen E, Rommers N, Dvorak J, Junge A, Lichtenstein E, et al. Comparison of the “11+ Kids” injury prevention programme and a regular warmup in children’s football (soccer): a cost effectiveness analysis. Br J Sports Med. 2019.

Frank BS, Register-Mihalik J, Padua DA. High levels of coach intent to integrate a ACL injury prevention program into training does not translate to effective implementation. J Sci Med Sport. 2015;18(4):400–6.

Hägglund M, Atroshi I, Wagner P, Waldén M, Hagglund M, Atroshi I, et al. Superior compliance with a neuromuscular training programme is associated with fewer ACL injuries and fewer acute knee injuries in female adolescent football players: secondary analysis of an RCT. Br J Sports Med. 2013;47(15):974–9.

Harøy J, Wiger EG, Bahr R, Andersen TE. Implementation of the adductor strengthening programme: players primed for adoption but reluctant to maintain — a cross-sectional study. Scand J Med Sci Sports. 2019.

Joy EA, Taylor JR, Novak MA, Chen M, Fink BP, Porucznik CA. Factors influencing the implementation of anterior cruciate ligament injury prevention strategies by girls soccer coaches. J Strength Cond Res. 2013;27(8):2263–9.

Soligard T, Nilstad A, Steffen K, Myklebust G, Holme I, Dvorak J, et al. Compliance with a comprehensive warm-up programme to prevent injuries in youth football. Br J Sports Med. 2010;44(11):787–93.

Steffen K, Emery CA, Romiti M, Kang J, Bizzini M, Dvorak J, et al. High adherence to a neuromuscular injury prevention programme (FIFA 11+) improves functional balance and reduces injury risk in Canadian youth female football players: a cluster randomised trial. Br J Sports Med. 2013;47(12):794–802.

Lindblom H, Carlfjord S, Hägglund M. Adoption and use of an injury prevention exercise program in female football: a qualitative study among coaches. Scand J Med Sci Sports. 2018;28(3):1295–303.

Article   CAS   Google Scholar  

McKay CD, Merrett CK, Emery CA. Predictors of FIFA 11+ implementation intention in female adolescent soccer: an application of the health action process approach (HAPA) model. Int J Environ Res Public Health. 2016;13(7).

O’Brien J, Finch CF. Injury prevention exercise programs for professional soccer: understanding the perceptions of the end-users. Clin J Sport Med. 2017.

Norcross MF, Johnson ST, Bovbjerg VE, Koester MC, Hoffman Marc F.; ORCID: http://orcid.org/0000-0001-5329-3925 MAAI-O http://orcid. org/Norcros. Factors influencing high school coaches’ adoption of injury prevention programs. [Internet]. Ardern Comstock, Donaldson, Finch, Finch, Finch, Finch, Frank, Gilchrist, Glasgow, Glasgow, Hagglund, Joy, Kamath, LaBella, Lindblom, McKay, Morrissey, Orr, Pate, Rogers, Rogers, Sadoghi, Sawyer, Soligard, Twomey, Van Tiggelen, White B, editor. Vol. 19, Journal of Science and Medicine in Sport. Norcross, Marc F., [email protected] : Elsevier Science; 2016. p. 299–304.

Donaldson A, Callaghan A, Bizzini M, Jowett A, Keyzer P, Nicholson M. Awareness and use of the 11+ injury prevention program among coaches of adolescent female football teams. Int J Sports Sci Coach. 2018;13(6):929–38.

Junge A, Lamprecht M, Stamm H, Hasler H, Bizzini M, Tschopp M, et al. Countrywide campaign to prevent soccer injuries in Swiss amateur players. Am J Sports Med. 2011.

O’Brien J, Young W, Finch CF. The use and modification of injury prevention exercises by professional youth soccer teams. Scand J Med Sci Sports. 2017.

Owoeye OBA, McKay CD, Verhagen EALM, Emery CA. Advancing adherence research in sport injury prevention [Internet]. Vol. 52, British Journal of Sports Medicine. 2018. p. 1078–9.

Steffen K, Meeuwisse WH, Romiti M, Kang J, McKay C, Bizzini M, et al. Evaluation of how different implementation strategies of an injury prevention programme (FIFA 11+) impact team adherence and injury risk in Canadian female youth football players: a cluster-randomised trial. Br J Sports Med. 2013;47(8):480–7.

Pryor JL, Root HJ, Vandermark LW, Pryor RR, Martinez JC, Trojian TH, et al. Coach-led preventive training program in youth soccer players improves movement technique. J Sci Med Sport. 2017;.

Bizzini M, Dvorak J. FIFA 11+: an effective programme to prevent football injuries in various player groups worldwide-a narrative review. Br J Sports Med. 2015;49(9):577–9.

Bizzini M, Junge A, Dvorak J. Implementation of the FIFA 11+ football warm up program: How to approach and convince the Football associations to invest in prevention. Br J Sport Med. 2013;47(12).

Download references

Acknowledgements

The funding for this review was managed by Pike, I. and Babul, S. of the British Columbia (BC) Injury Research and Prevention Unit and BC Children’s Hospital Research Institute and coordinated by Richmond, S. of the Canadian Injury Prevention Trainee Network.

Provided by the British Columbia Alliance for Healthy Living Society, Canada, and supported by the Saint Louis University, MO, USA.

Author information

Authors and affiliations.

Department of Physical Therapy and Athletic Training, Doisy College of Health Sciences, Saint Louis University, Allied Health Professions Building, 3437 Caroline Street, St. Louis, MO, 63104, USA

Oluwatoyosi B. A. Owoeye & Mitchell J. VanderWey

Sport Injury Prevention Research Centre, Faculty of Kinesiology, University of Calgary, Calgary, AB, Canada

Oluwatoyosi B. A. Owoeye

Department of Paediatrics, Faculty of Medicine, University of British Columbia, Vancouver, BC, Canada

BC Injury Research and Prevention Unit, BC Children’s Hospital Research Institute, Vancouver, BC, Canada

You can also search for this author in PubMed   Google Scholar

Contributions

OO conceived the study and did the initial systematic literature search. MV did the updated literature search. OO and MV drafted the manuscript, and OO and IP substantially revised it. The authors read and approved the final manuscript.

Corresponding author

Correspondence to Oluwatoyosi B. A. Owoeye .

Ethics declarations

Ethics approval and consent to participate.

Not applicable

Consent for publication

Competing interests.

The authors, Oluwatoyosi Owoeye, Mitchell VanderWey, and Ian Pike, declare that they have no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Additional file 1:..

Search terms

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Owoeye, O.B.A., VanderWey, M.J. & Pike, I. Reducing Injuries in Soccer (Football): an Umbrella Review of Best Evidence Across the Epidemiological Framework for Prevention. Sports Med - Open 6 , 46 (2020). https://doi.org/10.1186/s40798-020-00274-7

Download citation

Received : 24 April 2020

Accepted : 19 August 2020

Published : 21 September 2020

DOI : https://doi.org/10.1186/s40798-020-00274-7

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Open supplemental data
• Reference Manager
• Simple TEXT file

People also looked at

Original research article, a goal scoring probability model for shots based on synchronized positional and event data in football (soccer).

• 1 Sportec Solutions AG, Subsidiary of the Deutsche Fußball Liga (DFL), Munich, Germany
• 2 Institute of Sports Science, University of Tübingen, Tübingen, Germany
• 3 DFB-Akademie, Deutscher Fußball-Bund e.V., Frankfurt am Main, Germany

Due to the low scoring nature of football (soccer), shots are often used as a proxy to evaluate team and player performances. However, not all shots are created equally and their quality differs significantly depending on the situation. The aim of this study is to objectively quantify the quality of any given shot by introducing a so-called expected goals (xG) model. This model is validated statistically and with professional match analysts. The best performing model uses an extreme gradient boosting algorithm and is based on hand-crafted features from synchronized positional and event data of 105, 627 shots in the German Bundesliga. With a ranked probability score (RPS) of 0.197, it is more accurate than any previously published expected goals model. This approach allows us to assess team and player performances far more accurately than is possible with traditional metrics by focusing on process rather than results.

1. Introduction

In professional football (soccer), only 1% of all attacking plays and only around 10% of all shots taken end up in a goal ( Pollard and Reep, 1997 ; Tenga et al., 2010 ; Lucey et al., 2014 ). However, goals alone decide the outcome of a game and are the most common metric to judge both a team's and individual player's performance. For example, both the best goal scorers 1 and the players with the most assists 2 receive a lot of attention from experts and the media. Nevertheless, judging performances solely based on this binary metric ( goal or no goal ) loses a lot of information and places results over process. For example, the performance from an outstanding creative player could be made void by strikers missing all their chances.

For this reason, in football as well as in other sports, it has become typical to consider more granular process-based metrics. In baseball, scouts and experts focused their attention on homeruns or hits for decades until more complex evaluation metrics changed the assessment procedure of hitters' performance significantly ( James, 1985 ). Another famous example is basketball: By calculating scoring probabilities of different shot locations ( Reich et al., 2006 ; Chang et al., 2014 ; Harmon et al., 2016 ; Jagacinski et al., 2019 ), the NBA's shooting behavior changed significantly 3 . The high scoring nature of basketball enables clubs to go even further and to apply individual shooting efficiency models ( Beshai, 2014 ). Similar shot prediction models were also developed for ice hockey ( Macdonald, 2012 ) as well as for return plays in tennis ( Wei et al., 2016 ) and table tennis ( Draschkowitz et al., 2015 ).

The fact that football is the lowest scoring game of the above-mentioned sports, makes it harder to develop such models, because of the scarcity of data. Consequently, the rareness and therefore importance of goals makes such a metric even more relevant when assessing teams and players. As another consequence of this low-scoring nature, the role of shots as a success proxy within several studies in football is fortified ( Spearman et al., 2017 ). However, assessing shots just by being successful or not is a too rough abstraction that warps reality. An expected goals model (hereafter xG model ) tries to estimate the probability of any given shot being converted to a goal based on various different factors describing the shot. These probabilities can then be added up per team and yield a “result-agnostic” description of the teams' performance. The xG metric is well-established in the football analytics community (see Davis and Robberechts, 2020 ) 4 , 5 , 6 , 7 . Although to the best of our knowledge, no peer-reviewed journal publication has introduced a positional data-driven xG model, valuable work has been done in “gray literature” like master theses ( Hedar, 2020 ; Rowlinson, 2020 ) and conference proceedings ( Lucey et al., 2014 ). Rathke (2017) analyzed in total around 18, 000 shots from one season of Bundesliga and Premier League based on manually acquired shot annotations. Differentiating between four different shooting types (open play footed shot, header, freekick, or penalty shot) , Ruiz et al. (2017) built a multi-layer perceptron to predict shot outcomes based on roughly 10, 000 shots. Using a similar approach, Fairchild et al. (2018) tried to predict the goal scoring probabilities of 1, 115 non-penalty shots from 99 Major League Soccer matches, again solely based on event data.

Recent developments in technology allows us not only to make use of manually annotated event data (shots, passes, goals with a manually assigned location) but also accurate positions of all 22 players and the ball at up to 25 times a second. It is quite intuitive that the positioning of the defensive team, especially of the goalkeeper, has a crucial influence on the shot outcome ( Lucey et al., 2014 ; Schulze et al., 2018 ). Figure 1 displays the positioning of relevant players during two shots occurring at similar spots. In the left figure, both a defender and the goalkeeper are in good position to block the shot, while in the right figure the attacker has already dribbled past the goalkeeper (#38) and defenders, and faces an easy tap-in into an empty goal 8 . However, this information is not covered in event data and thus not taken into consideration in the previously listed xG models. Lucey et al. (2014) were the first to estimate goal probabilities using event and positional data in their model. They used 10, 000 shots of the English Premier League.

Figure 1 . Player positions of two shots from roughly the same location, but different surrounding environments. In both cases, the blue team is playing from left to right.

In this paper, we will introduce a shot prediction model, utilizing event and positional data. The accuracy of this model is evaluated both statistically and based on the discussion with professional match analysts. We also incorporate their expertise both when defining the model's features and when interpreting their influence on the prediction. Additionally, we show how our model can support coaching staffs by introducing various use cases and applying them on one season worth of Bundesliga data.

The remainder of this paper is structured as follows. In section 2, we introduce the data and definitions. How event and tracking data are synchronized is described in section 3. Section 4 describes how the supervised prediction model is build, and finally, section 5 consists of two parts: practical applications (5.1) of our approach based on a season of German Bundesliga and a critical discussion of the results (5.2).

2. Data and Definitions

Like in most other professional football competitions, the German Bundesliga systematically collects positional and event data on a league-wide level in a pre-defined and thus consistent format. Positional data —often also referred to as tracking or movement data ( Stein et al., 2015 )—provides the positions of all players, referees, and the ball related to the pitch boundaries with a frequency of 25 Hz. These data are gathered by an optical tracking system, which captures high-resolution video footage from different camera perspectives. On the other hand, event data are manually acquired by trained operators live during the match. Among other things, this event data contain many details about basic events, such as passes, shots, fouls, saves, and so on including the involved players or special characteristics.

Since shots are an important statistic in football, the event data in the Bundesliga describe them with more than 20 attributes. For example, the collector differentiates between three basic shot types (leg, header, other) or six different scenarios how a player controlled the ball before taking a shot (direct, volley, two touches, dribbling > 10 m, dribbling < 10 m, set-piece) .

In this investigation, we make use of 105, 627 shots from German Bundesliga and 2 nd Bundesliga of the seasons 2013/2014 until 2019/2020. The event data were collected according to the official Bundesliga match-data catalog 9 , and the optical tracking data were provided by Chyronhego's TRACAB system 10 .

Due to a growing availability of optical tracking systems in football, several studies have been conducted to evaluate their accuracy ( Redwood-Brown et al., 2012 ; Linke et al., 2018 , 2020 ; Linke, 2019 ; Taberner et al., 2019 ). In Linke et al. (2020) , the two versions of the TRACAB system ( Gen 4 / Gen 5 ) 11 were compared to an accurate ground truth measurement 12 . Both systems achieved a diversion of < 10 cm from the ground truth system (RMSE Gen 4: 0.09 cm, Gen 5: 0.08 cm). A non-peer reviewed study confirmed these results 13 . All above-mentioned evaluation studies focused on player detection, whereas the detection of the ball—probably the hardest challenge for optical tracking systems—is not covered.

To the best of our knowledge, no scientific study evaluated the quality of event data. However, in the German Bundesliga the acquisition follows an elaborate quality assurance process. Critical information is double-checked manually live (e.g., goals and red cards). Finally, an independent person inspects and adds additional information (e.g., event locations) to all acquired event data after the match.

3. Making Use of Both Positional and Event Data

3.1. synchronizing shots with tracking data.

A major challenge when attempting to use both tracking and event data is that they are generally not aligned. This is due to the fact that they come from different data providers and/or acquisition methods, one specialized in logging events manually according to catalog of set definitions (i.e., what is considered a shot or a tackling) and the other focusing on extracting player positions through, for example, computer vision algorithms. This leads to two potential issues when synchronizing the data:

(a) The manual collected event time stamps are prone to human errors, e.g., reaction time, distractions, and decision time, leading to time offsets of up to 20 s based on our investigations.

(b) The two systems use their own clock, causing systematic offsets between the two sources.

For these reasons, a “naive” synchronization—using the time stamp from the event data—to identify player positions at the time of an event leads to large inaccuracies. The upper plots in Figure 2 display the coordinates of the players and the ball at the different moments of the scenario from Figure 1 (right plot). The scene describes Kevin Volland's (Bayer Leverkusen) 1:0 against Borussia Dortmund (BVB) at the 14th matchday in the 2017/2018 season: 14 The upper right plot in Figure 2 displays the shot time stamp tagged in the event data, which is roughly 2 s after the time stamp our synchronization suggests the shot took place (upper middle plot). The upper left plot in Figure 2 shows the positioning of the players 2 s prior to that. As one can see, the situations are drastically different ranging from a distant dribble to a player celebrating his goal. The figure underpins that a shift of a few seconds in the synchronization can have a massive impact on the features used for the xG calculation, like the shot location or the goalkeeper position.

Figure 2 . Relevant metrics for the synchronization over time. The green points highlight the time interval where we detect a potential individual ball possession sequence. The orange point indicates where the shot event was finally detected.

Therefore, we developed a synchronization algorithm tackling both issues. As a first step, we shift all tracking time stamps by the time difference between the kick-offs in both data sets. This resolves issue (b) and furthermore reduces a potential systematic delay in the manual event collection. In order to tackle issue (a), we compute several features that help to determine when a particular shot could have happened in the tracking data. First, we determine when the shooting player was in ball possession. We define potential individual ball possession sequences as the time interval when the player is in close proximity to the ball—our subject experts suggested 2 m as a cut-off, which is in line with Linke et al. (2018) . Next, within each possession window, we identify the frame with the maximum acceleration of the Euclidean distance between player and ball. This aims to identify the exact moment where a shot occurred. Lastly, since there are potentially many situations that fulfill the above-mentioned criteria, we identify which best matches the event description. For that we compute Euclidean distances between the player and ball, the player and the manual collected event location as well as between the ball and the manual collected event location. Additionally, we compute the time difference between the (shifted) tracking time stamps and the manual collected event time stamp. We compute a weighted sum of these features, and the one frame out of the solution space that minimizes this weighted sum is chosen. The weights were obtained by performing a grid-based search that aimed to optimize accuracy of the synchronization on a manual labeled test set. The lower part of Figure 2 shows how these features behave in the 20 s before and after the exemplary shot described above. When we applied this synchronization algorithm on the full data set of six seasons, the event shot times had an average absolute offset of 2.3 s (≈57 frames) from the synchronized frame. Figure 3 displays histograms of the differences in timing (left) and locations (right) of each shot.

Figure 3 . Time stamp (left) and shot location (right) differences between event and synchronized time stamps.

3.2. Evaluation of the Synchronization

In order to evaluate the accuracy of the synchronization, we manually annotated the timing of total 219 shots of the nine matches from matchday one of Bundesliga season 2018/2019. First, a full 90 min video animation of the 2D tracking data was created for each match. As a ground truth, we used a tactical video feed, which is filmed manually with an angle to capture all outfield player (and the most relevant goalkeeper). Additionally, for each match a xml-file 15 containing all shot-events, and the kick-off was produced. Next, we used the kick-offs in all three data sources to synchronize them manually as accurately as possible using Hudl Sportscode 16 —a dedicated tool for football video analysis with functionalities to combine different video sources and data sources (i.e., event data can be imported via xml-files). For each shot, we stop the video at the exact moment the shot occurred—defined as the first frame when the ball left the shooter—and extract this time point using Sportscode functionalities.

We now use these labeled shot timestamps as the ground truth and compare them with both, the results from our synchronization, and the event timestamps. Our synchronization displays an average absolute offset of 0.23 (±0.49) s, while the event timestamps differ by 1.82 (±4.06) s. Out of the 219 shots, we were able to synchronize 218, and 210 (95.9%) of these shots were < 0.3 s apart from the ground truth 17 . In contrast, only 63 (28.8%) of the event timestamps were within 0.3 s of the ground truth. It is evident that generally this synchronization is far superior to event timestamps. Two exemplary situations for a successful and an unsuccessful shot synchronization can be found here 18 , 19 .

When a shot cannot be synchronized, it is typically due to either tracking data quality issues (e.g., the ball is poorly tracked, and never gets close to the player taking the shot, or two players were swapped in the tracking data) or event data quality issues (e.g., the wrong shooter is identified). To ensure that the quality of the input data is as high as possible, all shots that could not be synchronized at all were excluded from further analysis. Over the entire data set, this was the case in 3.4% of the shots.

All together, the synchronization of positional and event data presents a tremendous improvement for the analysis of shots, and could potentially be extended, using a similar algorithm, to other event types, like passes or tacklings. As we have seen above, misidentifying the shot time just slightly can cause a stark misrepresentation of its surrounding circumstances, and consequently affect the xG value significantly.

4. Expected Goals Modeling

4.1. hand-crafted feature extraction.

To feed the supervised machine learning model, features influencing the goal scoring opportunity were defined together with professional match analysts from Bundesliga clubs and the German national team. A description of all features can be found in Table 1 . In order to make full use of the synchronization of our two data sources, the features are based on both event and tracking data. The goalkeeper positioning is included in two features: We check whether they are in the line of shot, defined as the triangle between the shot location and the two posts, which is also the baseline for our shot angle calculation. Second, the distance between the goalkeeper and the goal is used as features in our model. The defending players' positions, either threatening to block the shot or applying pressure on the shooter, are also taken into consideration. Similarly to the goalkeeper feature, we count the number of defenders in the line of shot. Based on the logic from Andrienko et al. (2017) , we calculate the total amount of pressure on the shooter aggregated over all defending players, as well as the maximum individual pressure on the shot-taking player. For both pressure metrics, we additionally compute the differences to the expected pressures given the shot location. Furthermore, the speed of the shooter, while taking the shot, is integrated in our model.

Table 1 . Features derived from synchronized positional and event data used to train our model.

4.2. Predict the Scoring Probability as a Supervised Machine Learning Task

For a total of 105, 627 shots, all features from Table 1 were calculated based on the synchronized positional and event data. Since the features shot type and freekick significantly influence the contribution of all other features, we split our problem into three subtasks: the prediction of goal scoring probabilities of open play leg-shots, headers, and direct freekicks. Per subtask, the optimal set of features was explored. Consequently, for all three subtasks we trained several supervised machine learning models based on 81, 462 open play leg-shots, 18, 748 headers and 5, 417 direct freekicks, respectively, labeled by the information whether the shot ended up in a goal (1) or not (0). For each subtask, the shots were randomly split into 60% training, 20% validation, and 20% test data sets. To avoid over representing teams or scores, this split was conducted for every match separately. The final model, shown in Table 1 (row 5), describes the combination of our three submodels. To investigate the efficiency of the division into the three subgroups, another model is trained based on all 105, 627 shots taking all features from Table 2 including the information whether the shot was a header, a leg-shot from open play or a direct freekick.

Table 2 . Statistical evaluation of the expected goal model outcome.

Various standard supervised machine learning models were trained on the training data set, hyperparameters were optimized on the validation data set and the models' accuracy's were evaluated on the test data set. Naturally, the necessary hyperparameters depend on the machine learning algorithm. In the case of the extreme gradient boosting model (hereafter referred to as XGBoost), the parameters we optimized are as follows: Learning rate : controls the step size used per update; Max depth : limits the depth of the tree; Subsample : controls number samples applied to the tree; Min child weight : controls instance weight of a node. For the optimization, we applied Bayesian tree-structured Parzen Estimator hyperparameter optimization approaches for the gradient boosting model ( Bergstra et al., 2011 ; Dewnacker et al., 2016 ; Wang, 2019 ).

For several models in Table 2 , we calculated SHAP values per feature ( Roth and Thomson, 1988 ; Lundberg and Lee, 2017 ; Rodríguez-Pérez and Bajorath, 2020 ). In several applications, using SHAP values 20 instead of standard gain values has proven to be beneficial ( Antipov and Pokryshevskaya, 2020 ; Ibrahim et al., 2020 ; Meng et al., 2020 ).

In order to get a better understanding of the resulting model's accuracy, we implemented two simple models as a baseline models. The first one uses an attribute that is collected for every shot ( chance quality ). This manually collected attribute can contain one of the following two values: sitter or chance. The very simple model now assigns each shot the average conversion rate of the corresponding class. So all shots labeled as chances are assigned a value of 0.063, while the remaining shots labeled as sitters receive a value of 0.548. The second baseline model uses all the event data based features from Table 1 (namely Shot location, Type of shot, Taker ball-control, After freekick , and Freekick ), and train a XGBoost model using these features.

4.3. Statistical Evaluation of the Shot Prediction Model

The first two validation metrics (precision and recall) presented in Table 2 evaluate the outcome of a classification problem. A goal classified with an xG above 50% is classified as a true positive, whereas an unsuccessful shot with an xG below that threshold is defined as a true negative. Thereafter, a recall of 1 could simply be achieved by assigning each shot an xG value above 50%. To incorporate both the true positive and the false positive rate depending on the threshold into our evaluation, we also use the area under the receiving operator curve ( AUC ) as an error function ( Daskivich et al., 2018 ). However, it is our objective to assess the accuracy of the underlying goal scoring probabilities and not just of a binary classification (goal or no goal). While this is possible with the AUC, using the ranked probability score ( RPS ), as presented in Murphy (1970) , fulfills this purpose better, especially for imbalanced data sets.

By splitting up the shots into two groups (chances and sitters), the chance evaluation model ( Table 2 , row 6) achieves a good balance between precision and recall. While this relatively simple model already achieves a somewhat satisfactory RPS of 0.170, the human-made classifications are possibly biased by the shot outcomes. This label is therefore not used as a feature for the remaining prediction models. For the event data based model, the extremely low recall can be interpreted as follows: The model predicts xG value below 50% for most of the shots that actually end up as goals. However, the AUC shows that the event-based model yields more granular predictions than the chance evaluation model. In the direct freekick submodel, no xG prediction exceeds 50%, and therefore its precision is undefined.

Shots are non-deterministic, at the time of the shot, meaning that no model can have a 100% accuracy predicting whether any given shot will score. But what we can expect from our model predictions is that they converge over a large sample. To verify this, we looked at the first 54 matches (matchday one through three) of the 2020/2021 season in Bundesliga and 2nd Bundesliga. Out of the 1, 357 shots, 150 found the back of the net and our model predicted an aggregated xG value of 151.6.

Estimating a team's true strength or its future performances is a crucial unsolved problem in football with many potential use cases ( Goes et al., 2019 ). Both shots on target, two well-established metrics in the literature, have been used for this context ( Lamas et al., 2014 ). Figure 4 displays in which scenarios our xG values fulfills this task better than traditional approaches. It looks at how well you can predict a team's future rest of the season goal ratio (defined as the difference between goals scored and goals conceded) after a certain matchday, by only taking into account one aggregated metric before said matchday. On the y -axis, the correlation between the future goal ratio and the respective metrics (see legend) before that matchday ( x -axis) is shown. Consistently, over all considered seasons a team's historic xG values are able to predict future results better than traditional metrics, especially between matchday 10 and 20. Additionally, we found that in 73.3% of all matches (excluding draws), the winner had a higher xG value 21 , while only in 56.2% of these games, the winning team had more shots, than its opponent.

Figure 4 . Correlation between a team's future goal ratio after a certain matchday and an aggregated metric before said matchday (average of all seasons 2013/2014–2019/2020).

Next, we analyze the features' influence on the predicted goal scoring probability. In the following, we discuss the overall feature importance of our gradient-boosting model trained on all shots with the subcategories as features ( Table 2 , row 1). Figure 5 displays the overall influence according the respective SHAP values per feature on the right, which can be interpreted as an aggregated quantification of the feature's influence. The SHAP values show that the most crucial factors are the shot location (Goal Distance, Angle) and the goalkeeper position (Distance Goalkeeper to Goal) . Maximum Individual Pressure Diff , defined as the difference between the actual pressure and the average pressure given the shot location, has the third highest influence on the predicted values. In Figure 5 (left plot), the x -value of each colored dot displays how a feature influences the model, whereas the color scaling describes the value of the respective feature. Both a flat line and a smooth change of colors (from left to right or vice versa) indicates a roughly linear correlation between the feature value and the model outcome. In Figure 6 , this relationship between the feature values ( x -axis) and influence on the model ( y -axis) is shown more granularly. Although the red line shows a regression, the dispersion of the blue dots provide a deeper insight. Both the left plot in Figure 5 (smooth decrease of the colored dots from left to right) and Figure 6 (red line) shows that the goal distance has an almost linear impact on the predicted values. However, if the distance to the goal is very high, influence relies more on other features, as can be seen by the growing dispersion of the blue dots. The importance of the number defenders in the line of the shot (here Defenders ) underpins the relevance of using positional data, including all opposing players' positions. Looking deeper into the SHAP distributions of this feature, Figure 6 shows an almost linear decrease of the average SHAP value over all shots from zero to four defenders in the line of shot. For more defenders in the line of shot, the average SHAP value—describing a proxy for the features influence—remains mostly constant. In Figure 6 , the feature Goalkeeper in the goal underpins our practitioners' intuitive assumption and can be interpreted as follows: If the goalkeeper is not in the line of shot, it increases the xG value significantly.

Figure 5 . Feature importance according to Shapley values displayed as a SHAP summary plot (left) and global feature contributions by the mean SHAP value across all samples (right) .

Figure 6 . SHAP dependence Plot. For each shot, the respective feature value is plotted on the x-axis vs. the corresponding Shapley values on the y-axis (distance is displayed in meter, and speed is shown in meter per hour).

Again, most of this information would not be available in event data, which highlights the benefit of using both event and positional data once more.

4.4. Evaluation by Subject Matter Expertise

In several workshops with match analysts from Bundesliga clubs and the German national team, the features were defined and ranked according to the estimated influence. These estimations were compared with the above calculated feature importance. Additionally, the SHAP value dispersions and interpretations were discussed in detail. Besides a lot of agreement from practitioners, some statistical results—, e.g., the influence of 4–10 players in the line of shot—were discussed intensively among experts. To evaluate the plausibility of our model from a practitioners perspective, a workshop with selected (assistant) coaches of Bundesliga and 2nd Bundesliga clubs was conducted. For the recently concluded season, the coaches were asked to classify their matches into four categories: deserved or undeserved victories, draws, or losses as in Figure 8 . Afterwards, we compared their labels to the ones produced from our xG model. With a category-accordance of more than 85% (in total 102 matches with 293 goals), practitioners characterized our approach as a helpful tool to assess individual shot qualities and the overall performance of a team.

5. Application and Discussion

5.1. applications.

For the following section, we consider the 2019/2020 season of the German Bundesliga, with in total 306 matches, 954 goals, and 5, 450 shots. We describe how the goal scoring probability xG ( S ) model for a given shot S is aggregated over a season to evaluate teams and players further:

Own goals are not a subtype of a shot event, but rather a separate event type with different attributes. Therefore, they are excluded from our xG calculation. Penalties are assigned an xG value of 0.766, which is the average conversion rate in the Bundesliga history. In the case of so-called double-chance, situations in which a first shot is blocked, but is immediately followed up by a rebound shot, we calculate xG values for each shot. But when we aggregate the team level xG values, we do not want to simply add them up, because it could lead to situations where a teams xG value for small time-window could exceed 1. Therefore, given a double-chance S , defined as two shots within 5 s, we compute the overall probability as:

5.1.1. Teams

Figure 7 displays how many goals each team scored and conceded in comparison to the aggregated xG values our model computed. Consequently, for the 2019/2020 season, BVB (sixth place in the left ranking of Figure 7 ) scored roughly 30 more goals than the sum of all the respective shots' xG values would suggest. Figure 8 provides a closer look at BVB efficiency on a match level. Comparing actual goal differences to the xG differences, the upper right quadrant could be interpreted as deserved wins, where BVB created more promising shot opportunities than their opponents. Matches on the lower right could be interpreted as lucky wins, e.g., the return match 22 against Borussia Mönchengladbach (black and white hatched diamond logo in the bottom right of the left figure).

Figure 7 . Bundesliga 2019/2020 season ranking with aggregated xG and the actual number of goals (xG red, actual goals gray).

Figure 8 . Season report of BVB in season 2019/2020 showing efficiency of BVB matches according to the underlying xG values.

Another match, where our model would have predicted a different result is displayed in Figure 9 23 . The graph shows the aggregated xG values per team over the course of a match. Although SC Freiburg displayed an extraordinary shooting efficiency, by scoring three goals out of three difficult situations, Eintracht Frankfurt created several high quality chances but only converted three of them.

Figure 9 . xG match report of a Bundesliga match between SC Freiburg and Eintracht Frankfurt in season 2019/2020.

Furthermore, our model can help match analysts examine a teams' shooting behavior. Figure 10 presents the number of shots taken vs. the average xG-value per team (left) and for the most scoring strikers (right). Although Fortuna Düsseldorf (red/white logo furthest left in Figure 10 ) had an average xG value (∅(xG)) of 0.08 in the 2019/2020 season, Borussia Mnchengladbach seems to take their shots only in cases of a clear scoring opportunity (∅(xG) = 0.14). FC Bayern Munich (red/blue/white logo top right in Figure 10 ), takes by far the most shots per game. However, with around four less shots per match, Borussia Mönchengladbach has a higher quality of attempts according to our xG model. Comparing FC Augsburg (red/white/green logo with FCA inscription) to Werder Bremen (green diamond logo with a white W as an inscription) shows two distinct patterns. While both teams had a similar number of aggregated xGs over the whole season (see Figure 7 ), Bremen tends to take more shots in less promising situations, while FC Augsburg emphasizes more on taking their shots in situations with a higher goal scoring probability. Having this information for the next opponent prior to each match can help teams to adapt their defending strategy depending on the opponent's shooting preferences.

Figure 10 . Quality vs. quantity of shots taken per team (left) and player (right). The total number of goals scored over the whole season per team and player is displayed in black.

5.1.2. Players

Additionally, we can use player aggregated xG values, both for individual player performance analysis as well as scouting. Comparing Jadon Sancho to Serge Gnabry shows that both players—playing in similar positions and both with very successful teams—have strongly differing shooting patterns. Although Serge Gnabry (top left in Figure 10 ) takes the second most shots per match, Jadon Sancho (lowest in Figure 10 ) takes the fewest shots out of the top 10 scorers, but often in more promising situations according to the xG-values. Besides an overview of strikers shooting behavior in Figure 10 , xG provides a lot more applications to quantify a player's offensive contribution more granularly than traditional metrics.

Since our xG model can be seen as an average across all Bundesliga players' shot efficiency, it can also be used to find players that convert shots at an above average rate. Using this approach, we see that Robert Lewandowski (upper right in Figure 10 ) outscored his aggregated xG value (29.6) by about four goals, scoring a total of 34 in the season out of his 140 shots ( Table 3 , row 12). While this is already an impressive feat, there were in total 11 players, outscoring their xG totals by a larger margin. Jadon Sancho (17 goals/53 shots/8.49 xG agg ) and Erling Haaland (13 goals/34 shots/7.59 xG agg ) lead this category and showed an extraordinary scoring efficiency.

Table 3 . Players with the highest scoring efficiency in the German Bundesliga 2019/2020 season.

5.2. Discussion

We present an xG model that performs better than any of the approaches discussed in the introduction. Rathke (2017) split the pitch into eight zones and trained a logistic regression on each, indirectly taking shot location and angle into consideration. However, their analysis was neither tested on unseen data nor took the positions of defenders and goalkeepers into consideration. By contrast, Lucey et al. (2014) did not only make use of positional data, but also displayed the improvements of the model accuracy. They split all shots into six different game-context situations (open play, counterattack, corner, penalties, freekicks, set pieces) and also learned a regressor for each. Their average error across all shots and scenes is 0.1439. In our final combined model ( Table 2 , row 5), this average error is 0.0928. As a combination of the larger data set (more than 100, 000 shots), our novel synchronization approach (see section 3) and the expert crafted features (see section 4.1) are possible reasons for this improvement.

However, xG models in football are not without flaws. An often criticized point is that they are not evaluating dangerous situations where no shot took place. While this criticism certainly has merits, most offensive actions end up in shots. The official Bundesliga event data include an event type chance without a resulting shot , describing situations, where a team was in a scoring position, but failed to attempt a shot. In our data set, this event occurs on average only 0.93 times per match, underlining that the impact non-shot opportunities have for measuring team performance is rather small. Additionally, as seen in section 5.1, evaluating team strength is not the only application of xG. Shot conversion on team/player level, average shot quality or even on a goalkeeper analysis are insightful use cases that only depend on actual shots taken. Nevertheless, several studies aim to tackle this problem, of noteworthy goal-scoring opportunities without shots, by computing so-called expected possession values ( Link et al., 2016 ; Spearman, 2018 ; Fernández et al., 2019 ), but even these concepts are often build upon a well-calibrated xG model.

Following the logic of expected possession values, it is definitely a potential next step to break the contribution to a goal scored further down to the participating players and their actions. For instance, in the situation described in Figure 2 by assuming shots at several time-points, a simple rule-based approach using our xG model can quantify how much xG Volland added through his dribbling. Another popular extension of xG are expected assists (xA), which measure the likelihood that a pass leading to a shot becomes an assist, by assigning it the resulting xG value. This allows to quantify a player's shot assisting qualities independent of the final shooter's ability to score.

Both the synchronization and the inputs for the xG model heavily rely on the quality of the underlying data. Even for purely event data based xG models, Robberechts (2019) showed that their usefulness strongly depends on the event data quality. One of the parameters causing the biggest inaccuracy in the current model is the ball height. Small objects—like a ball—are hard to track based on video footage, especially due to confusion with replacement balls or other small white objects occurring in the stadium. For header shots, little differences in the ball height have a large impact on the ability of a player to control the placement of a shot causing inaccuracies for our current header model ( Table 2 , row 7). With a steady increase of video camera resolutions and object detection algorithms, we expect a significant improvement for ball tracking. This increase in data quality would likely improve shot synchronization results even further (see section 3.2) and consequently result in even more accurate xG models. Nevertheless, both for tracking data (including ball tracking) and for event data additional evaluation studies to ensure a high data quality for similar projects is essential. Although latest positional and event data provide accurate and detailed information about players, their body orientation and limb tracking could further improve the model's accuracy. For the header model in particular, heights and jumping altitude capacities could be taken into consideration as well.

The harmonization of tracking and event data is not a problem unique to football, which has been barely explored in the literature. In basketball, for instance, the two data sources 24 are mainly used independently of one another ( Tian et al., 2020 ), but as Manisera et al. (2019) noted the combination of both data sources is a crucial future issue. While our algorithm is optimized for football events, it could be adapted and applied to several other sports where both data sources are available.

An accurate expected goals model provides tremendous decision-making support for clubs: Creating many high-quality shooting situations is a crucial indicator of a good performance. To which extent these situations actually end up in goals often depend on random factors or luck. Consequently, a single final match result may not represent the actual team performance accurately. By quantifying a team's conversion rate (goals vs. xG) separately from their aggregated offensive contribution (created xG), clubs can evaluate the performance of their players, teams, and coaches objectively. Future research could even go one step further and explore how this work could affect the way the game is played. One could use our goal probabilities to determine numerically in which situations it is beneficial to shoot, and when one is better of risking an additional dribble or pass. Another area where the use of xG could be explored further are media applications: Recently, media and broadcasting have included xG values in their match coverage. For each goal occurring in German Bundesliga, different broadcasters have chosen to display our xG value seconds after the goal occurred 25 .

Now that the amount of data-driven approaches to support tactical analysis in football is increasing ( Goes et al., 2020 ), more qualitative studies might help to underpin the statistical evaluation of models like xG. Although we present a first attempt toward an expert-based evaluation of our approaches (see sections 3.2 and 4.4), there is a lot of potential for further investigations, which could also serve to establish data-driven methods in the sport science and football community.

6. Conclusion

We present a meaningful proxy for goals scored in football, which helps to evaluate players' and teams' performance more accurately and objectively. Our xG model is based on a huge data set of cutting-edge and consistently acquired positional and event data that we combined using our own synchronization algorithm.

It exceeds traditional metrics significantly when evaluating strikers' ( Table 3 ) and teams' ( Figure 7 ) scoring efficiency, when evaluating single match performances (i.e., teams with higher xG win 73.3% of all not-drawn matches) and even when predicting future match results ( Figure 3 ). It also allows us to evaluate assist performances of players independent of the striker's final touch. Additionally, several future potentials are shown for sport and data science research.

Data Availability Statement

The data analyzed in this study is subject to the following licenses/restrictions: data are property of DFL/DFB e.V. and thus can not be shared publicly. Requests to access these datasets should be directed to Sportec Solutions AG, DFL e.V., DFB e.V.

Ethics Statement

Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

Author Contributions

GA was responsible for the implementation of the approach and involved in all discussions with practitioners. PB focused on the practical evaluation and the communication with practitioners. Both authors conducted the scientific studies together and were equally involved in the writing process of the manuscript.

Conflict of Interest

GA was employed by the company Sportec Solutions AG and PB was employed by the company DFB-Akademie (Deutscher Fußball-Bund e.V).

Acknowledgments

This work would not have been possible without the perspective of professional match analysts from world class teams who helped us to define relevant features and spend much time evaluating (intermediate) results. We would cordially like to thank Dr. Stephan Nopp and Christofer Clemens (match analysts of the German National team), Jannis Scheibe (head match-analyst of the German U21 mens national team) as well as Sebastian Geißler (former match-analyst of Borussia Mönchengladbach). Additionally, the authors would like to thank Dr. Hendrik Weber and Deutsche Fußball Liga (DFL)/Sportec Solutions AG for providing the positional and event data.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fspor.2021.624475/full#supplementary-material

We also provide an example of the media live application in German Bundesliga 26 .

1. ^ https://www.goal.com/en-us/lists/cristiano-ronaldo-lionel-messi-pele-who-are-the-top-goal-scorers-/ynctx2o9fa371vi1x0dsgr0np (accessed July 10, 2020).

2. ^ https://www.givemesport.com/1534019-the-top-10-players-with-the-most-assists-in-europes-top-five-leagues-this-decade (accessed July 8, 2020).

3. ^ https://fivethirtyeight.com/features/how-mapping-shots-in-the-nba-changed-it-forever/ (accessed July 10, 2020).

4. ^ https://www.americansocceranalysis.com/home/2017/3/6/validating-the-asa-xgoals-model (accessed October 24, 2020).

5. ^ http://www.northyardanalytics.com/blog/2015/08/22/pitfalls-of-measuring-shooting-and-saving-skill/ (accessed October 24, 2020).

6. ^ https://www.optasports.com/services/analytics/advanced-metrics/ (accessed October 24, 2020).

7. ^ https://differentgame.wordpress.com/2014/05/19/a-shooting-model-an-expglanation-and-application/ (accessed October 24, 2020).

8. ^ The situation in the right plot is also displayed in Figure 2 . The respective video can be found here: https://www.youtube.com/watch?v=UdvrKfsJISY&feature=onebox&t=1m08s (accessed October 24, 2020).

9. ^ https://s.bundesliga.com/assets/doc/10000/2189_original.pdf (accessed September 10, 2020).

10. ^ https://chyronhego.com/wp-content/uploads/2019/01/TRACAB-PI-sheet.pdf (accessed September 10, 2020).

11. ^ Note that the Gen 5 system has been in use since season 2019/2020, while all prior Bundesliga seasons were tracked using the Gen 4 TRACAB version.

12. ^ The ground truth was measured by a VICON system, using an optoelectronic motion capture system based on markers placed on the tracked objects. Further details about this system can be found here: https://www.vicon.com/ . An evaluation study of that system can be found in Merriaux et al. (2017) .

13. ^ The study was conducted by the Fédération Internationale de Football Association (FIFA) in close cooperation with the Victoria University (Melbourne, Australia). An overview of the study can be found here: https://football-technology.fifa.com/en/media-tiles/fifa-quality-performance-reports-for-epts/ , the report of the Gen 5 system can be found here: https://football-technology.fifa.com/media/172171/chyronhegoopt-fifa-epts-report-nov2018.pdf (accessed December 26, 2020).

14. ^ https://www.youtube.com/watch?v=UdvrKfsJISY&feature=onebox&t=1m08s (accessed September 10, 2020).

15. ^ Xml stands for e X tensible M arkup L anguage and is an established format to transfer complex data files.

16. ^ https://www.hudl.com/products/sportscode (accessed June 20, 2020).

17. ^ We use a range here, because both, harmonizing the different video and data sources and the manual selection of the shot timestamp, may cause slight time discrepancies.

18. ^ In the first sequence, actual match-footage of a scene is shown. The second shows a 2D animation of the same scene, with a frame-counter on top. This frame counter counts down till 0 where the shot happened and increases afterwards again. The third sequence combines both video sources together (see Supplementary Video 1 ).

19. ^ See Supplementary Video 2 .

20. ^ The abbreviation SHAP stands for SH apley A dditive ex P lanation.

21. ^ On a match and team level the overall xG balance between the two teams is considered here. For both teams, we sum up the xG values per team of all their shots.

22. ^ https://www.youtube.com/watch?v=RUaORAinaoc&feature=onebox (accessed October 2, 2020).

23. ^ https://www.youtube.com/watch?v=jl1C0KsIqaQ (accessed October 2, 2020).

24. ^ In basketball, event level data are often referred to as play-by-play data.

25. ^ https://www.dfl.de/en/news/bundesliga-and-amazon-web-services-to-develop-next-generation-football-viewing-experience/ (accessed September 10, 2020).

26. ^ https://www.youtube.com/watch?v=5flVB9ef0uM .

Andrienko, G., Andrienko, N., Budziak, G., Dykes, J., Fuchs, G., von Landesberger, T., et al. (2017). Visual analysis of pressure in football. Data Mining Knowl. Discov . 31, 1793–1839. doi: 10.1007/s10618-017-0513-2

CrossRef Full Text | Google Scholar

Antipov, E. A., and Pokryshevskaya, E. B. (2020). Interpretable machine learning for demand modeling with high-dimensional data using gradient boosting machines and shapley values. J. Rev. Pricing Manage . 19, 355–364. doi: 10.1057/s41272-020-00236-4

Bergstra, J., Bardenet, R., Bengio, Y., and Kégl, B. (2011). Algorithms for hyper-parameter optimization, in Advances in Neural Information Processing Systems 24: 25th Annual Conference on Neural Information Processing Systems 2011, NIPS 2011 (Granada), 1–9.

Google Scholar

Beshai, P. (2014). Buckets: Basketball Shot Visualization . Semantic Scholar Preprint, 1–14.

Chang, Y. H., Maheswaran, R., Su, J., Kwok, S., Levy, T., Wexler, A., et al. (2014). Quantifying shot quality in the NBA, in MIT Sloan Sports Analytics Conference (Boston, MA), 1–8.

Daskivich, T., Luu, M., Noah, B., Fuller, G., Anger, J., and Spiegel, B. (2018). Differences in online consumer ratings of health care providers across medical, surgical, and allied health specialties: observational study of 212,933 providers. J. Med. Internet Res . 20, 29–36. doi: 10.2196/jmir.9160

PubMed Abstract | CrossRef Full Text | Google Scholar

Davis, J., and Robberechts, P. (2020). How data availability affects the ability to learn good xG models, in 7th International Workshop of Machine Learning and Data Mining for Sports Analytics (Ghent). doi: 10.1007/978-3-030-64912-8_2

Dewnacker, I., McCourt, M., and Clark, S. (2016). Bayesian optimization for machine learning. a practical guidebook. arXiv 2–5.

Draschkowitz, L., Draschkowitz, C., and Hlavacs, H. (2015). Using video analysis and machine learning for predicting shot success in table tennis. EAI Endorsed Trans. Creat. Technol . 2:150096. doi: 10.4108/eai.20-10-2015.150096

Fairchild, A., Pelechrinis, K., and Kokkodis, M. (2018). Spatial analysis of shots in MLS: a model for expected goals and fractal dimensionality. J. Sports Anal . 4, 165–174. doi: 10.3233/JSA-170207

Fernández, J., Bornn, L., and Cervone, D. (2019). Decomposing the Immeasurable Sport: a deep learning expected possession value framework for soccer, in MIT Sloan Sports Analytics Conference , 1–18.

Goes, F., Kempe, M., and Lemmink, K. (2019). Predicting match outcome in professional Dutch football using tactical performance metrics computed from position tracking data, in MathSport International Conference (Athens), 4–5. doi: 10.29007/4jjb

Goes, F. R., Meerhoff, L. A., Bueno, M. J. O., Rodrigues, D. M., Moura, F. A., Brink, M. S., et al. (2020). Unlocking the potential of big data to support tactical performance analysis in professional soccer: a systematic review. Eur. J. Sport Sci . doi: 10.1080/17461391.2020.1747552. [Epub ahead of print].

Harmon, M., Lucey, P., and Klabjan, D. (2016). Predicting shot making in basketball learnt from adversarial multiagent trajectories. arXiv .

Hedar, S. (2020). Applying machine learning methods to predict the outcome of shots in football outcome of shots in football (Thesis), Uppsala University, Uppsala, Sweden.

Ibrahim, L., Mesinovic, M., Yang, K.-W., and Eid, M. A. (2020). Explainable prediction of acute myocardial infarction using machine learning and shapley values. IEEE Access 8, 210410–210417. doi: 10.1109/ACCESS.2020.3040166

Jagacinski, R. J., Newel, K. M., and Isaac, P. D. (2019). Predicting the success of a basketball shot at various stages of execution. J. Sport Psychol . 1, 301–310. doi: 10.1123/jsp.1.4.301

James, B. (1985). The Historical Baseball Abstract .

Lamas, L., Barrera, J., Otranto, G., and Ugrinowitsch, C. (2014). Invasion team sports: strategy and match modeling. Int. J. Perform. Anal. Sport 14, 307–329. doi: 10.1080/24748668.2014.11868723

Link, D., Lang, S., and Seidenschwarz, P. (2016). Real time quantification of dangerousity in football using spatiotemporal tracking data. PLoS ONE 11:e0168768. doi: 10.1371/journal.pone.0168768

Linke, D., Link, D., and Lames, M. (2018). Validation of electronic performance and tracking systems EPTS under field conditions. PLoS ONE 13:e0199519. doi: 10.1371/journal.pone.0199519

Linke, D., Link, D., and Lames, M. (2020). Football-specific validity of TRACAB's optical video tracking systems. PLoS ONE 15:e0230179. doi: 10.1371/journal.pone.0230179

Linke, D. M. (2019). Validation of methodology, design & applications (Ph.D. thesis), Technische Universität München, Munich, Germany.

Lucey, P., Bialkowski, A., Monfort, M., Carr, P., and Matthews, I. (2014). “Quality vs Quantity”: improved shot prediction in soccer using strategic features from spatiotemporal data, in MIT Sloan Sports Analytics Conference , 1–9.

Lundberg, S. M., and Lee, S. I. (2017). Consistent feature attribution for tree ensembles, in Proceedings of the 34th International Conference on Machine Learning (Sydney), 1–9.

Macdonald, B. (2012). An expected goals model for evaluating NHL teams and players, in MIT Sloan Sports Analytics Conference 2012 (Boston, MA), 1–8. doi: 10.1515/1559-0410.1447

Manisera, M., Metulini, R., and Zuccolotto, P. (2019). Basketball analytics using spatial tracking data. Springer Proc. Math. Stat . 288, 305–318. doi: 10.1007/978-3-030-21158-5_23

Meng, Y., Yang, N., Qian, Z., and Zhang, G. (2020). What makes an online review more helpful: an interpretation framework using XGBoost and SHAP values. J. Theor. Appl. Electron. Comm. Res . 16, 466–490. doi: 10.3390/jtaer16030029

Merriaux, P., Dupuis, Y., Boutteau, R., Vasseur, P., and Savatier, X. (2017). A study of vicon system positioning performance. Sensors 17, 1–18. doi: 10.3390/s17071591

Murphy, A. H. (1970). The ranked probability score and the probability score: a comparison. Mon. Weather Rev . 98, 917–924. doi: 10.1175/1520-0493(1970)098<0917:TRPSAT>2.3.CO;2

Pollard, R., and Reep, C. (1997). Measuring the effectiveness of playing strategies at soccer. J. R. Stat. Soc. D Stat . 46, 541–550. doi: 10.1111/1467-9884.00108

Rathke, A. (2017). An examination of expected goals and shot efficiency in soccer. J. Hum. Sport Exerc . 12, S514–S529. doi: 10.14198/jhse.2017.12.Proc2.05

Redwood-Brown, A., Cranton, W., and Sunderland, C. (2012). Validation of a real-time video analysis system for soccer. Int. J. Sports Med . 33, 635–640. doi: 10.1055/s-0032-1306326

Reich, B. J., Hodges, J. S., Carlin, B. P., and Reich, A. M. (2006). A spatial analysis of basketball shot chart data. Am. Stat . 60, 3–12. doi: 10.1198/000313006X90305

Robberechts, P. (2019). Valuing the art of pressing, in StatsBomb Innovation in Football Conference 2019 (London), 11.

Rodríguez-Pérez, R., and Bajorath, J. (2020). Interpretation of machine learning models using shapley values: application to compound potency and multi-target activity predictions. J. Comput. Aided Mol. Des . 34, 1013–1026. doi: 10.1007/s10822-020-00314-0

Roth, A. E., and Thomson, W. (1988). The Shapley Value: Essays in Honor of Lloyd S. Shapley. Cambridge University Press . Available online at: https://www.hbs.edu/faculty/Pages/item.aspx?num=6946

Rowlinson, A. (2020). Football shot quality (Master thesis), Aalto University, Espoo, Finland.

Ruiz, H., Power, P., Wei, X., and Lucey, P. (2017). “The Leicester City Fairytale?”: utilizing new soccer analytics tools to compare performance in the 15/16 & 16/17 EPL seasons, in Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (Halifax, NS), 1991–2000. doi: 10.1145/3097983.3098121

Schulze, E., Mendes, B., Maurício, N., Furtado, B., Cesário, N., Carriço, S., et al. (2018). Effects of positional variables on shooting outcome in elite football. Sci. Med. Football 2, 93–100. doi: 10.1080/24733938.2017.1383628

Spearman, W. (2018). Beyond expected goals, in MIT Sloan Sports Analytics Conference (Boston, MA), 1–17.

Spearman, W., Basye, A., Dick, G., Hotovy, R., and Pop, P. (2017). Physics-based modeling of pass probabilities in soccer, in MIT Sloan Sports Analytics Conferece (Boston, MA), 1–14.

Stein, M., Häußler, J., Jäckle, D., Janetzko, H., Schreck, T., and Keim, D. A. (2015). Visual soccer analytics: understanding the characteristics of collective team movement based on feature-driven analysis and abstraction. ISPRS Int. J. Geoinform . 4, 2159–2184. doi: 10.3390/ijgi4042159

Taberner, M., O'Keefe, J., Flower, D., Phillips, J., Close, G., Cohen, D. D., et al. (2019). Interchangeability of position tracking technologies; can we merge the data? Sci. Med. Football 4, 76–81. doi: 10.1080/24733938.2019.1634279

Tenga, A., Ronglan, L. T., and Bahr, R. (2010). Measuring the effectiveness of offensive match-play in professional soccer. Eur. J. Sport Sci . 10, 269–277. doi: 10.1080/17461390903515170

Tian, C., De Silva, V., Caine, M., and Swanson, S. (2020). Use of machine learning to automate the identification of basketball strategies using whole team player tracking data. Appl. Sci . 10:24. doi: 10.3390/app10010024

Wang, Y. (2019). A Xgboost risk model via feature selection and bayesian hyper-parameter optimization. arXiv . doi: 10.5121/ijdms.2019.11101

Wei, X., Lucey, P., Morgan, S., Reid, M., and Sridharan, S. (2016). The Thin Edge of the Wedge: accurately predicting shot outcomes in tennis using style and context priors, in MIT Sloan Sports Analytics Conference (Boston, MA), 1–11. doi: 10.1145/2783258.2788598

Keywords: expected goals, XG, positional data, event data, applied machine learning, football, soccer, sports analytics

Citation: Anzer G and Bauer P (2021) A Goal Scoring Probability Model for Shots Based on Synchronized Positional and Event Data in Football (Soccer). Front. Sports Act. Living 3:624475. doi: 10.3389/fspor.2021.624475

Received: 31 October 2020; Accepted: 15 February 2021; Published: 29 March 2021.

Reviewed by:

Copyright © 2021 Anzer and Bauer. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Gabriel Anzer, gabrielanzer@gmail.com

† ORCID: Gabriel Anzer orcid.org/0000-0003-3129-8359 Pascal Bauer orcid.org/0000-0001-8613-6635

This article is part of the Research Topic

Using Artificial Intelligence to Enhance Sport Performance

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Front Sports Act Living

The perception of injury risk and prevention among football players: A systematic review

Beatriz cardoso-marinho.

1 Portugal Football School, Portuguese Football Federation, Oeiras, Portugal

2 Research Center in Sports Sciences, Health Sciences and Human Development, CIDESD, University of Maia, Maia, Portugal

3 Portuguese Institute of Sports and Youth, IPDJ, Sports Medicine Center, Porto, Portugal

4 Armed Forces Hospital, Porto, Portugal

Ana Barbosa

5 EPIUnit - Instituto de Saúde Pública, Universidade do Porto, Porto, Portugal

6 Laboratório Para a Investigação Integrativa e Translacional em Saúde Populacional (ITR), Porto, Portugal

Caroline Bolling

7 Amsterdam Collaboration on Health & Safety in Sports, Department of Orthopaedic Surgery and Sports Medicine, Amsterdam Movement Science, Amsterdam UMC, Amsterdam, the Netherlands

José Pedro Marques

8 Hospital da Luz, Lisboa, Portugal

Pedro Figueiredo

9 Physical Education Department, College of Education, United Arab Emirates University, Al Ain, Abu Dhabi, United Arab Emirates

10 Research Center in Sports Sciences, Health Sciences and Human Development, CIDESD, Vila Real, Portugal

João Brito

Associated data.

The original contributions presented in the study are included in the article/ Supplementary Material , further inquiries can be directed to the corresponding author/s.

Football is associated with a certain risk of injury, leading to short- and long-term health consequences. However, the perception of football players about injury risk and prevention strategies is poorly documented. The present article reviewed the literature about perceptions, beliefs, attitudes and knowledge toward injury risk and prevention strategies in football players. An electronic search was performed in PubMed, Scopus, Web of Science, and APA PsychINFO until July 2022. Studies were eligible if they included the perceptions, beliefs, attitudes, and knowledge about injury risk and prevention in football players from any competitive level. The risk of bias was assessed in included studies using the Joanna Briggs Institute critical appraisal checklist. A total of 14 studies were included. Most football players agreed that their risk of injury is high and prevention strategies are important, however they do not intend to use some of these strategies. The most frequent perceived injury risk factors were low muscle strength, lack of physical fitness, fatigue, excessive training and type and condition of surfaces. The most frequent perceived injury prevention factors were warm-up, workload monitoring and strength and conditioning training. It is essential to acknowledge perceived injury risk factors, as well as a better understanding of how coaching and medical departments' perceptions match with players' perceptions, and a modification in the perceptions of the several stakeholders at different levels of action.

Introduction

In football (soccer), given the popularity of the sport, the large number of players and the levels of competition, injuries are frequent ( 1 ).

Overall, the incidence of injuries in professional football players is 8.1 injuries/1,000 h of exposure for male players ( 2 ), and 6.1 injuries/1,000 h of exposure for female players ( 3 ). At the amateur level, generally, the incidence is 12.5/1,000 h of exposure ( 4 ). For both groups and competitive levels, the incidence of injuries is higher in matches than in training sessions, and lower extremity injuries have the highest incidence rates ( 2 – 4 ). Also, most time-loss injuries in professional football players led to an absence of up to four weeks ( 5 ).

With such numbers, football-related injuries may have a major negative impact on physical, mental and financial burden on the players and their clubs. For example, the mean cost of an injured player in a professional team has been reported for 500.000€/month ( 6 ). Also, injuries may have a significant impact on socioeconomic systems worldwide ( 7 ), thus requiring socio-ecological perspectives that consider the specific contexts and integrate comprehensive analyses at multiple levels ( 8 ).

There is a growing call to advance the translation of evidence-based sports injury prevention programs into sustained use in practice. Studies suggest that successful incidents with risk-taking may clue to a reduction in injury risk perception in sports. Conversely, an overestimation of the risk may increase the risk of injury due to the inability to make no-risk decisions ( 9 ). To implement effective and meaningful interventions toward injury risk and prevention, it is essential to acknowledge the players' perspectives on this topic.

Thus, the present systematic review aims to provide an overview of football players' perceptions, beliefs, attitudes, and knowledge regarding injury risk and prevention, based on epidemiological definitions.

We adopted the Preferred Reporting Items for Systematic Reviews and Meta-Analysis criteria (PRISMA) to conduct this systematic review ( 10 ). The PRISMA 2020 checklist is available for further consultation in Supplementary Material S1 . The protocol has been registered in PROSPERO (reference number CRD42021270395).

Eligibility criteria

Studies were considered eligible considering the following inclusion criteria: (i) the participants were male or female football (i.e., soccer) players; (ii) the intervention included any form of football practice as competitive or amateur; and (ii) outcome measures included perceptions, beliefs, attitudes, or knowledge towards injury risk and prevention.

Articles were excluded according to the following criteria: studies not in humans or other agents involved in sports such as coaches, medical or science departments; type of studies such as editorials, comments, case reports, guidelines, reviews or conference abstracts; studies that did not apply only to football but rather other sports, including studies with other football codes; studies that did not guide or have the primary outcome of interest.

Information sources

An electronic search was conducted in PubMed, Scopus, Web of Science and APA PsychINFO, from inception until July 2022, to identify articles assessing the perception of injury risk and injury prevention among football players.

Search strategy

For database search, we used the following keywords: (“injury risk” OR “injury prevention”) AND (perception OR beliefs OR knowledge OR attitude) AND (football OR soccer). We did not apply limits to the search. The full search strategy is available in Supplementary Material S2 .

Selection process

According to predefined steps, two authors (BCM and AB) independently reviewed the search results and screened publications retrieved from databases. First, articles were screened by the information outlined in the title and abstract. Then, articles potentially relevant were retrieved for full-text reading and determined eligibility for the review. Disagreements between authors were solved by consensus.

Data collection process

Two authors (BCM and AB) independently evaluated each selected article to extract information from the eligible studies. Data were compared and discussed in case of discrepancies. If necessary, the study authors were contacted to provide further explanation.

From the eligible studies, data were extracted regarding (1) study characteristics (first author, year of publication, country, objectives, design, instrument of data collection); (2) study participants (including age, sex, level of competition); (3) team type; and (4) outcomes (perceptions, attitudes, beliefs and knowledge about injury risk and prevention strategies).

Risk of bias assessment

To assess the risk of bias, two researchers (BCM and AB) independently applied the Joanna Briggs Institute (JBI) critical appraisal checklist according to the study design: randomized controlled trial (RCT) ( 11 ), cohort ( 12 ), cross-sectional ( 12 ), and qualitative and mixed methods' studies ( 13 ). If disagreements occurred, authors discussed them until they reached a consensus.

Synthesis methods

We conducted a narrative synthesis of included studies. We analysed and computed the outcomes of interest concerning football players' perceptions, beliefs, knowledge and attitudes about injury risk and injury prevention for each study.

Study selection

A total of 800 references were identified in the initial search from electronic databases search. After removing duplicated studies ( n  = 313), 487 studies remained. In step 1, 444 articles were excluded, and 43 studies were eligible for full-text reading, from which 29 were removed. Thus, 14 studies were included for qualitative synthesis ( Figure 1 ).

PRISMA 2020 flow diagram of included studies.

Study characteristics

Studies included were conducted in Qatar ( 14 ), Canada ( 15 , 16 ), Norway ( 17 ), United Kingdom ( 18 , 19 ), Germany ( 20 , 21 ), United States of America ( 22 ), Ireland ( 23 ), Brazil ( 24 ), Brunei ( 25 ), Austria ( 26 ), and Hong Kong Special Administrative Region of the People's Republic of China ( 27 ) ( Table 1 ).

Characteristics of included studies.

CG, control group; F, female; IG, intervention group; M, male; NR, Not reported; RCT, randomized controlled trial.

All studies were published in English between 1998 and 2022. Regarding the study design, eight studies were cross-sectional ( 14 , 15 , 19 , 21 , 23 – 25 , 27 ), two were cluster-randomized controlled trials ( 16 , 17 ), two were cohort ( 20 , 22 ), one was qualitative ( 18 ), and one used mixed methods ( 26 ). The sample sizes ranged from 30 to 1,129 participants. The participants’ age ranged from 11 to 40 years old. The studies comprised elite and professional players ( 14 , 17 – 20 , 24 , 26 , 27 ), youth and amateur players ( 15 , 16 , 22 , 23 , 25 ), and one study involved both competitive levels ( 21 ). Five studies included both genders ( 17 , 18 , 22 , 25 , 27 ), four studies were conducted on female players ( 14 – 16 , 23 ), and five studies involved male players ( 19 – 21 , 24 , 26 ) ( Table 1 ).

Seven studies used a questionnaire ( 15 , 16 , 18 – 20 , 24 , 25 ), five studies used a survey ( 14 , 17 , 21 , 23 , 27 ), one study used a scale to analyse the perception of injury risk and injury prevention among football players ( 22 ), and one study used semi-structured interviews and focus groups ( 26 ) ( Table 2 ).

Findings of included studies.

Risk of bias in studies

The included studies presented several methodological limitations. In RCT studies ( 11 ), the most common issues identified were the lack of information on randomization, allocation concealment and blinding. In cohort studies ( 12 ), most of the concerns were related to the report of factors and strategies to deal with confounding, the follow-up of participants and statistical analysis. In cross-sectional studies ( 12 ), the majority of issues identified were linked to the report of factors and strategies to deal with confounding, and the validity of the instruments used to measure the outcomes. In qualitative ( 13 ) and mixed-methods studies ( 13 ), studies lacked the statement of the influence of the researcher on the research. The assessment of the risk of bias in the studies is available in Supplementary Material S3 . There was an agreement between authors when assessing the risk of bias.

Results of individual studies

The detailed findings regarding the perceptions, beliefs, attitudes and knowledge of injury risk and prevention among football players are described in Table 2 .

Gender differences

Deeming to the perceptions between genders, McKay et al. ( 16 ) reported that 27.8% of female players believe that male and female soccer players have the same overall risk of injury. On the other hand, Kontos et al. ( 22 ) reported that boys (11–14 years) described significantly higher levels of risk-taking and lower levels of perceived risk than girls ( Table 2 ).

Injury risk factors

Concerning the perceptions about injury risk, most football players agreed that their risk of sustaining an injury was moderate to high and players expected to sustain at least one injury during the following season ( 14 , 15 , 17 ). Most football players believed injuries are a severe problem ( 20 ). Regarding overuse injuries, half of the players considered to be at high risk, and 10% of the players thought football players have an increased risk of illnesses ( 17 ). Considering injury severity, 50% and 40% of the players believed fractures and concussions to be very serious injuries, respectively ( 15 ). Also, Kontos et al. ( 22 ) noted that lower levels of perceived risk increased injury risk.

Intrinsic factors

The most frequently cited intrinsic injury risk factors were poor muscle strength ( 14 , 24 ), lack of physical fitness ( 20 , 25 ) and fatigue ( 21 , 23 ) ( Table 3 ). In the study of Alahmad et al. ( 23 ), joint mobility and menstrual regulation were not reasons to increase the risk of injuries in female athletes.

Summary of injury risk and prevention factors.

Extrinsic factors

The most commonly mentioned extrinsic injury risk factors for injury risk were excessive training ( 24 ), the type or condition of a playing surface and specifically the artificial surface ( 14 , 18 ) ( Table 3 ). Concerning the equipment, the study of Hawkins ( 19 ) found that in training, 51 players (out of 55) never wore shin pads, even though 30 of these players agreed that wearing shin pads reduced the risk of a lower leg injury.

Injury prevention factors

Despite players knowing that the risk of injury was high, their beliefs surrounding the perception of injury prevention and actual practices were low ( 14 , 15 ). Players seem to be interested in injury prevention strategies and consider them very important ( 20 , 21 ).

However, for example, a study of McKay et al. ( 15 ) showed that, besides the expectation to reduce the injury risk with the FIFA 11 + program, players reported a limited intention to use it. Having enough time, making the programme a routine, and having someone taking responsibility for leading the prevention programme were facilitators perceived as necessary for players ( 15 ). Moreover, 60% of the players reported poor team support of the program, finding the programme too complex, and scheduling changes by club officials, soccer federations and team staff as key barriers to the injury prevention program's implementation ( 15 , 19 , 26 ) ( Table 2 ).

Motivation ( 14 ), diet control ( 24 ) and knowledge about the cause of injury ( 23 ) were the intrinsic aspects identified to injury prevention ( Table 3 ).

The most frequent players' features for preventing injuries were a warm-up ( 16 , 19 , 23 – 25 ), strength and conditioning training ( 19 , 24 , 25 , 27 ), and workload monitoring ( 17 , 24 , 26 ) ( Table 3 ).

In the present systematic review, we summarized the evidence regarding football players' perceptions, beliefs, attitudes and knowledge toward injury risk and prevention. The injury risk is multifactorial: it entangles a match between the intrinsic and extrinsic factors. On this road, we should be aware of biomechanical, anatomical, hormonal, physiological, psychological, social, and neuromuscular factors that involve and evolves the athlete ( 28 ). Therefore, it is urgent to reflect that intrinsic and extrinsic risk factors should be studied and monitored by the technical and medical teams.

Verhagen et al. ( 29 ), explain the leadership and communication to enhance health and performance in elite sports: a multidisciplinary team is required to follow the road around all of these factors that are perceived by the athlete as the risk of injury.

Lack of muscle regeneration, densely packed games in a season, inadequate workload management, and inadequate warm-up were commonly cited extrinsic risk factors for injury ( 16 , 17 , 19 , 21 , 23 , 24 ). Players perceived artificial turf or uneven terrains as a possible extrinsic factor for injury, and players considered wearing shinpads to reduce the risk of injury ( 15 , 19 , 20 , 22 ). Since the extrinsic factors were important perceived risk factors for injury, they can be monitored during the season. These factors could be used in athlete screening to target preventive interventions ( 30 ).

Overall, one of the most frequently cited intrinsic injury risk factors was poor muscle strength ( 14 , 24 ), despite the already known evidence of the effectiveness of neuromuscular training strategies in reducing injury in football players ( 31 ). For example, one study conducted in female football players showed that a 15-minute neuromuscular exercise programme reduced the rate of ACL injuries, severe knee injuries and overall injuries ( 32 ).

In this systematic review, female players reported significantly lower levels of risk-taking and higher levels of perceived risk than male players. Nevertheless, gender-related risk factors show female populations to have a higher predisposition to ACL injury than males ( 33 ). However, there are no evidence-based guidelines around hormonal regulation and the injury risk for female athletes for practitioners to apply ( 34 , 35 ). Particularly, in the area of injury risk, further studies are needed.

However, the lack of uptake and current maintenance of such programs is an ongoing concern. For instance, high compliance with the 11+, an injury prevention programme developed by FIFA targeting to reduce the sway of intrinsic injury risk factors in football, led to decreases in injury rates and time loss in football players ( 36 ). Also, players with high compliance with neuromuscular training programs significantly reduced ACL injury rates compared with players with low compliance ( 37 ). This highlights the importance of consistency and compliance with injury prevention training.

In this study, although most studies reveal that football players have the perception of their higher injury risk, their motivation and attitude to pursue prevention programs were limited. Lack of personal, coach and team motivation were cited as the main reasons for that ( 15 ).

Currently, significant deterioration in team and player compliance may occur throughout the season ( 31 , 38 ). If injury awareness was given a similar weighting in elite sports as in any other highly physical occupation, the potential benefits and long-term health improvement could be significant ( 39 ). Therefore, the focus on implementation is critical to influencing knowledge, behaviour change, and sustainability of evidence-informed injury prevention practice. In future dissemination of injury prevention programs, players' reluctance to sustain exercise protocols should be addressed as a potential barrier to implementation ( 40 ).

Van der Horst et al. ( 41 ) studied the key issues in motivating football players to adhere to the Nordic hamstring training programme to decrease the risk of hamstring injuries. The issues were knowledge of the programme and personal motivation. Coaches and medical departments also cited personal enthusiasm and consensus with team staff to encourage adherence to the programme ( 41 ).

Moreover, the main enablers for players to implement a load management approach were scientific evidence for improved performance (88%) and mitigation of injuries and illnesses (84%), and a positive attitude of the coach towards it (86%) ( 17 ). This aligns with Andersson et al . ( 37 ), which established a link between player motivation and coach motivation.

Though injury prevention programs might be effective, there is a need to ensure the real vision of all stakeholders for failing to adhere to them. Players reported their motivation and the cooperation of the coach as facilitators ( 14 , 15 ).

Therefore, if players have a high interest in injury prevention ( 22 , 23 , 25 , 27 ) and firm beliefs about the warm-up ( 18 , 21 , 36 , 38 ), workload management ( 19 , 38 ), flexibility training ( 19 , 24 ) and strength training ( 19 , 24 , 27 ) as well as diet ( 38 ), a multidisciplinary team can address the need of the injury prevention programs.

However, there is still a lack of knowledge regarding the quantity and quality of coach-led injury prevention plans and the relative impact on players' performance. Therefore, injury prevention efforts need to be built around athletes' behaviours to be effective. Consequently, there is a need to know and understand more about the behavioural aspects related to injury occurrence.

Also, it is vital to identify reasons and perceptions for increasing adherence to injury prevention programs in real-world settings by including the limitations and barriers throw players' vision. It is now understood that sports injury interventions will not have a significant public health impact if they are not widely accepted and adopted by all stakeholders ( 42 ). Moreover, real-world implementation of injury prevention interventions and evaluation of their effectiveness needs to consider the wide-ranging environmental context in which they are introduced.

The most frequently quoted injury risk issue was the lack of muscle regeneration with a short break between matches and a high number of matches in a season ( 22 , 38 ). Though, players strongly believed that workload management could support reducing injury risk.

Alongside the lack of medical support ( 22 , 36 ), it's important to reflect on the periods of over-scheduling during the season, in which the player's training load increases, cutting player recovery time. Furthermore, all stakeholders should help in the schedule and consider the moments of injury risk reduction, as well as awareness about the management of training conditions, such as artificial turf and field conditions, since they were perceived as a common cause of injuries ( 16 , 20 ).

Concerning the agenda, players described scheduling changes as a significant hurdle for the injury prevention session ( 15 , 17 , 19 , 26 ). Although at the same time, the agenda is problematic for the strength and conditioning training programs, the planning and integration of soccer practices that could promote injury prevention, such as small-sided games, is a worthwhile opportunity ( 43 ).

Implications for practice

Sports injuries can result in significant setbacks, pain, social isolation, depression, disability and loss of income being some of them ( 44 ). It can also predispose athletes to degenerative disorders, such as osteoarthritis ( 44 ). A preventive approach is paramount, and exercise can be an effective tool to prevent sports-related injuries. Thus, the factors hindering athletes' injury awareness from achieving occupational health standards can be discussed from organizational, societal and individual safety management perspectives ( 45 ). Multi-level engagement strategies are required to maximize athlete adherence to the programs. Hence, future studies should focus on enhancing performance programs to catch athletes' engagement in prevention strategies.

There is also a lack of studies regarding the epidemiology of football-related injuries, especially in women's football, and at both professional and amateur levels. Specific evidence is necessary to investigate at these levels, such as intrinsic (e.g., hormonal regulation, motivation for prevention programs' implementation) and extrinsic factors (e.g., climate, playing position, periods of fixed match congestion, number of matches and breaks per season), with the multiple stakeholders involved in football, so that the effectiveness of interventions can be tested.

The evidence regarding the use of preventive programs and measures are still scarce, therefore no definite conclusions for football stakeholders (e.g., coaches, medical or science departments) can be made. However, low levels of perceived risk may increase injury risk ( 22 ). Therefore, studies are needed to design and implement behavioural interventions to educate players, coaches, and other stakeholders about injury risk, aiming to explain the risks and consequences of injuries, and elucidate long-term costs for health and socio-economic systems.

Limitations

There were small sample sizes across studies, the geographical areas of the studies were limited, the participants included in the studies were generally young, and studies with both sex ( 17 , 18 , 22 , 25 ) did not include representative samples of both sexes, thus affecting the generalization of the findings.

Some of the questionnaires were not available in the athlete's native language and did not include open questions about their opinions. Also, in some studies, the questionnaires were completed in a team setting and might have been subject to social desirability bias in the team ambience.

Official medical records did not back up most studies about injury risk, and retrospective study designs may have increased the bias regarding the assessment of accurate injury history.

Moreover, a limitation of the current review is the possibility of a high risk of bias, as assessed in the JBI critical appraisal checklists, with studies lacking essential items of the studies’ methodology, and therefore results should be interpreted with caution.

In conclusion, this systematic review explored football players' perceptions of injury risk and prevention. Most football players agreed that their risk of injury is high and that prevention strategies are important, still, they do not intend to use some of these strategies. It is crucial to acknowledge perceived injury risk factors, namely low muscle strength, lack of physical fitness, fatigue, excessive training and type and condition of surfaces, as well as injury prevention factors, such as the warm-up, workload monitoring and strength and conditioning training.

A better understanding of how the coach and medical departments match with players' perceptions may help inform delivery strategies, leading to better compliance with injury prevention programs. In addition, there is a need to alter the perceptions through education, rule changes, economic measures, and changes in the governance of the sport. Questioning more stakeholders and policymakers can shed light on such potential interventions.

Data availability statement

Author contributions.

BCM and JB: conceptualization. BCM: writing—original draft preparation. BCM and AB: methodology. BCM and AB: formal analysis. BCM and AB: interpretation of data for the work. CB, JPM, PF and JB: writing—review and editing and supervision. All authors contributed to the article and approved the submitted version.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fspor.2022.1018752/full#supplementary-material .

Studying professional and recreational female footballers: A bibliometric exercise

Affiliations.

• 1 James R. Urbaniak, Sports Sciences Institute, Duke Health, Durham, NC, USA.
• 2 Department of Sports Science and Clinical Biomechanics, SDU Sport and Health Sciences Cluster (SHSC), Faculty of Health Sciences, University of Southern Denmark, Odense, Denmark.
• 3 Danish Institute for Advanced Study (DIAS), University of Southern Denmark, Odense, Denmark.
• 4 Sport and Health Sciences, College of Life and Environmental Sciences, University of Exeter, Exeter, UK.
• PMID: 34363241
• DOI: 10.1111/sms.14019

Objectives: Research directed at soccer has seen dramatic growth in the last decade. While published research on soccer has shown exponential growth, the proportion of articles addressing females is lagging behind research addressing males. The purpose of this paper is to explore how the literature on soccer, female soccer, and professional female soccer has changed over time.

Methods: The Web of Science (WoS) was queried for all "articles" about soccer and association football from 1970 to 2019. This set of records was then queried to collect subsets of papers about females, professional/elite, and female professional/elite. Each of these data subsets was then queried for a number of characteristics and topics. The results were submitted to bibliometric analysis.

Results: WoS returned 16,822 "articles" about soccer from 1970 to 2019, 3242 of which addressed females. A total of 5924 "articles" about professional players was found, of which 919 had a female focus. Articles about anterior cruciate ligament injuries and concussion were the topics with the highest proportion of papers involving females. Articles directed at selective areas of training and performance were relatively infrequent. Prominent journals, authors, affiliations, and influential papers are presented.

Conclusions: A bibliometric analysis of the published research presents a high-level overview of trends in soccer research. Overall, studies about women accounted for around 20% of all soccer research and about 15% of studies on professional players. There were a number of topics where studies on females account for less than 10%-15% of the research on all professionals, and opens opportunities for future study.

Keywords: VOSviewer; association football; bibliometrics; females; professional players; soccer.

© 2021 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.

• Anterior Cruciate Ligament Injuries*
• Bibliometrics
• Soccer* / injuries

Finished Papers

The various domains to be covered for my essay writing.

If you are looking for reliable and dedicated writing service professionals to write for you, who will increase the value of the entire draft, then you are at the right place. The writers of PenMyPaper have got a vast knowledge about various academic domains along with years of work experience in the field of academic writing. Thus, be it any kind of write-up, with multiple requirements to write with, the essay writer for me is sure to go beyond your expectations. Some most explored domains by them are:

• Project management

What if I can’t write my essay?

Week 50 Pools RSK Papers 2022: Soccer, Bob Morton, Capital Intl, Winstar, BigWin

June 12, 2022 Obinna FOOTBALL POOLS , POOL RSK PAPERS 1

Week 50 pools RSK papers page.

Here, we furnish you with weekly and current pool rsk papers for your forecast and winning pleasure.  Click on the images to view more clearly. Enjoy

RSK PAPERS – Soccer X Research, Bob Morton, Capital International

Unfortunately we cannot upload rsk papers yet due to the legal threat.

However, for those that stake their coupons through us hand via our online pool agent office, the papers will be dully delivered to you in the whatsapp group.

Special Advance Fixtures

Right On Fixtures

Temple of Draw

Capital Internationa

Pools Telegraph

Soccer X Researc h

Page Two will be uploaded later.

Soccer Percentage

Week 50 Pools RSK Papers…. Enjoy your Forecast

Also check –  Week 50 Pools SURE banker room

TO VIEW PREVIOUS WEEKS,GO TO WEEKLY POOLS RSK PAPERS ARCHIVE

For Livescores update see – Livescoresupdate.com

You are simply the best @ admin.

Leave a Reply Cancel reply

Your email address will not be published.

Save my name, email, and website in this browser for the next time I comment.

Copyright © 2024 | Pools Betting Blog by SolutionTipster Web Services

• Advertise Here
• Pool Code For This Week
• Football Pools Draws
• Half-Time Results
• Bigwin, Telegraph Papers
• Calculate Pools Winning
• Cookies Policy
• Terms of Services

Week 30 Pool RSK Papers 2021: Bob Morton, Capital Intl, Soccer X Research, Winstar, BigWin

Week 30 pool rsk papers 2021, pool rsk papers week 30, week 30 rsk pool papers 2021, week 30 bob morton, week 30 capital international, week 30 soccer x research, week 30 special advance fixtures, rsk papers week 30 2021.

Welcome to Fortune Soccer we are  provide you with football pools papers from RSK and other publishers such as Bob Morton, Capital International, Soccer ‘X’ Research and WinStar, Bigwin Soccer, Special Advance Fixtures, Right On Fixtures, Weekly Pools Telegraph, Pools Telegraph, Temple of Draws, Dream International Research, Fortune Soccer Research and Fortune 1.X.2 Matrix papers.

To Download RSK papers: Click on the papers!

SPECIAL ADVANCE FIXTURES

RIGHT ON FIXTURES

CAPITAL INTERNATIONAL

SOCCER ‘X’ RESEARCH

BIGWIN SOCCER

POOLS TELEGRAPH

Related :  Week 29 Pool RSK Papers 2021: Bob Morton, Capital Intl, Soccer X Research, Winstar, BigWin

BIGWIN SOCCER PAPER

Week 31 Bigwin Soccer and Pools Telegraph Pool Late News 2021

[PUBLISHED EVERY THURSDAY]

powered by FortuneSoccer.com

Tags : Pool RSK Paper This Week

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Save my name, email, and website in this browser for the next time I comment.

Recommended Websites

Quick links.

• Banker Room | Pool Draw This Week
• Football Pools Draws This Week
• Football Pools Fixtures For This Week
• Late News Papers: Bigwin Soccer, Pools Telegraph
• Livescore Football Pools Results and Fixtures
• Pool Discussion Room For This Week
• Pool RSK Paper This Week
• Sports News