Use of IoT technologies to improve shooting performance in basketball

i Use of IoT technologies to improve shooting performance in basketball André Salomão dos Santos de Almeida Antunes Dissertation report presented as partial requirement for obtaining the Master’s degree in Statistics and Information Management i Use of IoT technologies to improve shooting performance in basketball André Salomão dos Santos de Almeida Antunes MGI 2018 MGI i ii NOVA Information Management School Instituto Superior de Estatística e Gestão de Informação Universidade Nova de Lisboa USE OF IOT TECHNOLOGIES TO IMPROVE SHOOTING PERFORMANCE IN BASKETBALL by André Salomão dos Santos de Almeida Antunes Dissertation report presented as partial requirement for obtaining the Master’s degree in Information Management, with a specialization in Knowledge Management and Business Intelligence Advisor: Vítor Manuel Pereira Duarte dos Santos, PhD. May 2018 iii DEDICATION Nesta seção de agradecimentos e dedicações, escrevo em português, na minha lingua materna. Sendo esta etapa uma das mais importantes que tive até hoje, só posso dedicar este feito à minha familia. Se não fosse pelo constante suporte e apoio, não tenho a certeza se teria completado este passo. Por esse motivo o mérito de ter finalizado este estudo não é só meu, e quero fazer um agradecimento especial à minha madrinha, à minha mãe, ao meu pai e à minha avó. Um obrigado a todos! ACKNOWLEDGEMENT Agradeço o apoio e colaboração do Prof. Dr. Vítor Duarte dos Santos na orientação pedagógica e acompanhamento efetuado ao longo do desenvolvimento desta dissertação e à NOVA IMS, pelos fundamentos académicos que, no âmbito do Mestrado em Gestão de Informação, pude adquirir e se revelaram essenciais no decurso deste trabalho. iv ABSTRACT Technology is revolutionizing the world of sports in every way, from the experience of fans to the making and customising of training plans and even refereeing. Inertial sensors are now being used in many sports as they allow effective tracking of metrics that were previously not “within reach” without affecting the performance of players, due to the improvement of their size and “durability”. But it is not just the technological component that is evolving; new strategies and tactical displays are being increasingly seen in several sports, such as basketball. Indeed, in the NBA, in recent years a new trend has emerged as teams are shooting more 3pt shots, and the centre position is progressing as taller players are asked to be more skilled than ever. However, although the game of basketball is changing, a gain in efficiency is not being observed in jump shooting since the percentages of 3pt shots made in the NBA are not increasing as it would be expected from the increase in 3pt shot attempts. The purpose of this study was thus to analyse and make recommendations concerning the use of current technology for tracking shooting performance, as well as the use of new sensors. In order to do so, the main factors behind shooting success were taken into account, to guarantee that the recommendations were as well-founded as possible. The chosen methodology was design science research, where the proposed artifacts were submitted to validation through interviews, and according to received feedback, the proposed artifacts were updated. KEYWORDS Internet of Things; Sports; Basketball; Jump shot; Devices. v INDEX 1. Introduction .................................................................................................................. 1 1.1. Background ............................................................................................................ 1 1.2. Motivation ............................................................................................................. 2 1.3. Objectives .............................................................................................................. 3 2. Literature review .......................................................................................................... 4 2.1. IoT .......................................................................................................................... 4 2.1.1. Concepts ......................................................................................................... 4 2.1.2. Context for IoT ................................................................................................ 4 2.1.3. IoT in sports .................................................................................................... 5 2.2. Basketball .............................................................................................................. 7 2.2.1. Concept, rules and scoring ............................................................................. 8 2.2.2. Major competitions ...................................................................................... 10 2.2.3. Trends and challenges .................................................................................. 11 2.2.4. Shooting critical components ....................................................................... 13 2.2.5. Technologies in Basketball ........................................................................... 24 3. Methodology .............................................................................................................. 34 3.1. DSR ....................................................................................................................... 34 3.2. Strategy of the Research ..................................................................................... 34 4. Proposals, Results and discussion .............................................................................. 36 4.1. Assumptions of the proposed artifacts ............................................................... 36 4.2. Proposed Artifacts ............................................................................................... 42 4.3. Validation ............................................................................................................. 45 4.3.1. Target Audience ........................................................................................... 45 4.3.2. Sample .......................................................................................................... 45 4.3.3. Summary of interview guide ........................................................................ 46 4.3.4. Results .......................................................................................................... 47 4.4. Discussion ............................................................................................................ 50 5. Conclusions ................................................................................................................. 56 5.1. Synthesis of the developed work ........................................................................ 56 5.2. Limitations of the research .................................................................................. 57 5.3. Future work ......................................................................................................... 57 Bibliography..................................................................................................................... 58 Annexes ........................................................................................................................... 61 vi LIST OF FIGURES Figure 1. ITPD ring around IoTSport (from Ray, 2015) ............................................................... 6 Figure 2. 2016 NBA playoffs' bracket ....................................................................................... 11 Figure 3. Average number of 3 point shot attempts per game in each season ...................... 12 Figure 4. Average percentage of points resulting from 3 point shots per season ................... 12 Figure 5. Average percentage of success of 3 point shots per season .................................... 13 Figure 6. Variables that influence the basketball jump shot (from Okazaki, Rodacki, & Satern, 2015) ................................................................................................................................. 14 Figure 7. Virtual target of the basket rim as a function of the angle of entry of the basketball (from Okazaki et al., 2015) ............................................................................................... 14 Figure 8. Left: Distribution of the Left-Right values for all 3-point shots measured. Right: correlation between Left-Right consistency and 3-point shooting percentage for individual shooters (from Marty & Lucey, 2017). ............................................................ 17 Figure 9. Left: Distribution of Depth values for all 3-point shots measured. Right: correlation between Depth consistency and shooting percentage for individual shooters (from Marty & Lucey, 2017). ...................................................................................................... 17 Figure 10. Left: Distribution of Angle values for all 3-point shots measured. Right: Correlation between Angle consistency and shooting percentage for individual shooters (from Marty & Lucey, 2017). ...................................................................................................... 18 Figure 11. Left: Make percentage at each Depth value for straight shots with an Angle value of 45°. Right: Guaranteed Make Zone (GMZ) for straight shots across all Angle values (from Marty & Lucey, 2017). ............................................................................................ 18 Figure 12. Heat map showing players clustered by their proficiency at the four factors (from Marty & Lucey, 2017). ...................................................................................................... 19 Figure 13. Left: Current shot distribution of Player Y where Depth median is not centred in the GMZ. Right: potential shot distribution of Player Y if his median Depth is improved by centring in the GMZ (from Marty & Lucey, 2017). ...................................................... 20 Figure 14. Phases of the jump shot (from Okazaki et al., 2015) .............................................. 21 Figure 15. Evo one smart basketball ........................................................................................ 26 Figure 16. Wilson X smart basketball ....................................................................................... 27 Figure 17. Wilson X application modes .................................................................................... 27 Figure 18. 94fifty gear .............................................................................................................. 28 Figure 19. ShotTracker gear ..................................................................................................... 29 vii Figure 20. Hoop tracker gear.................................................................................................... 30 Figure 21. Hoop tracker features ............................................................................................. 31 Figure 22. Catapult's ClearSky T6 ............................................................................................. 32 Figure 23. Noahlytics entry attributes ..................................................................................... 32 Figure 24. Design Science Research cycle (from Mendonça 2015) ......................................... 34 Figure 25. Design Science Research methodology adopted to current investigation ............. 35 Figure 26. Suggested use of sensors for the measurement of metrics ................................... 40 Figure 27. Summary of the interview's sections ...................................................................... 47 5 monitoring. On a global scale, it is being used to gather information and help us better understand our planet and its resources, enabling us to make better use of them and care for the environment (Evans, 2011). These are only a few of the potential benefits of IoT for society at large. Nevertheless, there are a number of challenges that must be overcome for IoT to succeed in the long run. The sheer variety and heterogeneity of IoT applications has made it extremely difficult to identify solutions that meet the requirements of all possible scenarios. Furthermore, the wide range of devices, link layer technologies and services involved in the IoT system, have made the task of creating a uniform architecture that would work in different situations very challenging. (Zanella et al., 2014) Other technical issues that must be addressed include existing problems with data transmission (particularly location data), irregular wireless coverage, and difficulties with instant analysis and conversion of large data streams into meaningful, real-time and personalised recommendations. More work is also needed to further reduce the cost of sensors and increase battery life (Swan, 2012). In this context, it should be noted that a commercially viable nanogenerator, which uses body movements to generate energy, has been developed (Evans, 2011) and constitutes a significant step towards much needed self-sustainability. Due to the complexity and novelty of IoT, an established best practice is also lacking. In addition, a clear and proven business model is necessary so as to attract investment for the deployment of innovative technologies that will allow further development (Somayya Madakam, R. Ramaswamy, 2016) (Zanella et al., 2014). Lastly, with the rising amounts of information available through IoT technology, more work needs to be done to ensure the privacy and security of personal and confidential data (Cisco Systems, 2013). The resolution of these and other identified problems is not insurmountable but will require the joint work of government, businesses, academia and other organisations. Moreover, in order to be welcomed by the public, future IoT applications should focus on adding tangible value to the lives of people (Evans, 2011). 2.1.3. IoT in sports Despite being in its early stages, IoT for sports is a rapidly growing area of research. The practice of sports is deeply ingrained in our society and the competitive nature of professional sport places a premium on any information that might improve the performance of the athlete. The aim of IoT for sports is to ultimately enable athletes to function at an optimum level by collecting and analysing relevant types of data regarding their health and athletic performance (Partha Pratim Ray, 2015). Ray, P.P. suggests that the IoT for sports architectural framework should be based upon the structure of the ITPD ring which stands for Interaction, Things, Processes and Data, see figure 1. The first step in this structure highlights the athlete’s need to become familiar with and learn to successfully interact with the data collection or measuring devices (here referred to as “things”). These devices connect the objects to the network making the collected information available to the user. The next step focuses on the business and/or technological processes that mainly deal with the tasks of accumulation, communication and analysis. Their goal is to expedite and automate the flow of data (the last concept in the ring structure) from the moment it is captured by the aforementioned 6 devices to the moment when it is displayed for analysis in a cloud-based or real-time platform (Partha Pratim Ray, 2015). Figure 1. ITPD ring around IoTSport (from Ray, 2015) Traditionally, and before IoT came into play, video recording was the only way of analysing the performance of athletes at an individual and collective level. Indeed, to this day, teams still film their training sessions and matches to gain a better understanding of what went wrong and why it went wrong; and find ways to ensure that mistakes are not repeated. For instance, in the NBA league, players and coaches in post-match interviews often refer to video analysis when speaking of adjustments that might need to be made in order for the team to succeed. Moreover, with the advances in technology, match footage can now be fed into software that provides new visualization tools as in the case of TenniVis - a new visualization system for tennis matches that only requires the input of non-spatial game data like score, point outcomes and length, service information and video that can be recorded by a single consumer-level camera. The main goal of this system is to allow non-professional players and coaches to learn about their swings and service techniques through economic and feasible technology which provides clear visualization tools, like the Pie Meter and Fish Grid. This enables the detection of patterns and the sharing of newly gained knowledge between player and coach (Polk, Yang, Hu, & Zhao, 2014). While video analysis remains a widely used tool in sport, recordings typically generate large amounts of data making the search for key events an extremely time-consuming process. To tackle this, the use of inertial sensors in combination with video has been attempted with a degree of success. In 2012, a general technique was proposed to show how inertial sensors could be used to index video. This was another tennis visualization system, but in this case the required data consisted of inertial data on strokes (collected by accelerometers or gyroscopes mounted in the racquet), a timestamp for these measurements and video data. The main advantage was that, by using this system, coaches were able to skip large amounts of irrelevant video data speeding up the process of analysis. In 7 addition, if only the relevant video segments were stored, significant savings would be made (Rowlands, McCarthy, & James, 2012). Inertial sensors are considered to be the present and the future of the classification of sports activities. In actual fact, it has been shown that it is possible to use accelerometer and gyroscope data to accurately classify sporting activities (Wundersitz et al., 2015). Nonetheless, one should be cautious when using sensor technology and ensure that both internal and external validations of devices and algorithms, as applicable, are satisfactorily carried out before considering the results accurate. The first microtechnology sensor to be validated both internally and externally was the minimaxX - a wearable microtechnology unit able to offer a valid method of quantifying the number and intensity of collisions performed in rugby leagues (Gabbett, 2013). Wearable technology enables athletes to efficiently collect data in the familiar surroundings of their usual training location, thus avoiding the physical limitations of the laboratory environment and facilitating an authentic assessment of performance. Furthermore, recent reductions in sensor size have been undertaken to ensure that freedom of movement is preserved and minimal disruption is caused (Espinosa et al., 2015). The use of sensors presents major advantages such as the measurement of physical activity and monitoring of physical fitness which can be used to predict performance and prevent injury. In addition, when combined with biophysical markers, sensor data can be used to evaluate muscle damage and establish the required time period for full muscle recovery. Unlike other time-motion analysis systems, integrated technology offers real-time collection of information on various parameters including impact load, directional movement and activity identification (Dellaserra, 2013). For instance, in 2015, a rugby study was conducted to determine whether the number of contact efforts had an impact on running intensity, for which data was collected using global positioning system units. The results showed that there was indeed an impact, with running intensity decreasing progressively as the number of contact efforts increased. Taking this into account, it is suggested that coaches should target defensive players in attack to affect their performance on offense (Johnston, Gabbett, Walker, Walker, & Jenkins, 2015). Overall, an increase in battery power and memory storage combined with further reductions in product size could not only widen the use of this technology but also improve its functionality by providing quicker access to new information, which in turn could lead to significant improvements in the preparation, training and recovery programmes (Dellaserra, 2013). 2.2. BASKETBALL Sports can be generally divided into two main categories: individual sports and team sports. As the name suggests, the level of success achieved in individual sports depends solely on the performance of the individual athlete. In contrast, the success attained in team sports relies on the performance of the whole group, even if some players arguably contribute to it more than others. In 1891, James Naismith, an instructor at the YMCA International Training School in Springfield, was tasked with the creation of a new indoor team sport to keep his students in shape during the harsh weather conditions of the Massachusetts’ winters. With the use of a basket of fruit, one soccer ball, and thirteen basic rules, basketball was born. As time passed, changes were made regarding the way 8 the game was played, and the backboard was eventually added. The year of 1936 remains a historical year for basketball since it was entered in the Olympic Games as an official event for the first time. 2.2.1. Concept, rules and scoring Concept Basketball is a team sport in which two teams compete against each other. Each team is usually composed of twelve rostered players, with five on the court at any one time and up to seven on the bench. The objective of the game is to beat the opposing team by making goals and scoring more points than them. Goals are made by inserting the ball into the opponent’s basket and preventing the opponent from doing the same as much as possible. The winner of the match is the team with the highest score at the end of regular time. In case of a draw, additional periods of five minutes (known as overtime) are added to game time, until there is a winner. There are two main basketball associations - FIBA and NBA. A summary of the rules by which FIBA is governed is given below, which is followed by a table 1 that outlines the main differences between the rules that apply to both associations. Characteristics of the game Basketball is played with a spherical ball and two baskets with backboards at either end of the court. The standard for a ball is approximately 75 centimetres in circumference and 600 to 650 grams in weight. The hoop is positioned 3.05 metres above the floor and has a diameter of 45 centimetres. The court itself has a number of lines and markings that are crucial to the game. The side-lines and baselines make up the perimeter of the playing area, and the midcourt line divides it in symmetrical halves. The free throw or foul line, as the name indicates, is used for free throws when fouls are committed. The three-point line surrounds the basket and helps determine how many points a shot is worth. The key, also known as the free throw lane (NBA) or restricted area (FIBA) is the area closest to the basket usually painted in a different colour to the rest of the court. The duration of a basketball game is forty minutes (known as regulation in the NBA) split into four periods or quarters of ten minutes each. There is a fifteen-minute halftime between the second and third periods and a two-minute break between the first and second, and the third and fourth periods. Each team is allowed two timeouts during the first two periods, and three during the last two periods. Timeouts are one-minute breaks that can only be requested by the coach. Basketball places no limit on the number of player substitutions that can be made during a game. However, teams can only switch players when the clock is stopped or the ball is declared “dead” by officials, such as when a foul is committed, a player is injured, the ball is out-of-bounds, following a timeout or when the opposite team scores a basket in the last two minutes of regulation or overtime. Rules General rules Below is a summary of the main rules that guide the game. Breaking them results in immediate loss of ball possession and may have additional consequences such as in the instance of goaltending 9 where the attempted shot that was interfered with is automatically considered made regardless of whether or not the ball made it through the hoop. • The ball carrier cannot take more than two steps without dribbling, otherwise a travelling violation is called. • Upon receiving the ball, a player can either pass, shoot or dribble but once dribbling is stopped, it cannot be reinitiated as this would constitute double dribble. • A defensive player cannot interfere with the ball when it is in downward flight towards the basket or has touched the backboard in the process. Doing so would be considered goaltending. • When a team has ball possession, a player from the offensive team cannot be more than three seconds in the key area, or a three-second violation is called. • Upon gaining possession, a team has eight seconds to make the ball pass the half court, or an eight second violation is called. • The offensive team has twenty-four seconds to take a shot; if the ball touches the rim and the team regains possession they have another twenty-four seconds. However, if no shot is taken within the allotted time, the team gets a shot clock violation. • When the offensive team passes the half court line, it constitutes a backcourt violation to pass or dribble the ball back to the backcourt. Fouls Fouls are called when players break the rules and result in free throws for the other team. There are different types of fouls in basketball: personal fouls which can be offensive or defensive, and technical fouls. Personal fouls involve illegal physical contact with an opponent either through holding, pushing or charging into the opponent. If a player was fouled in the act of shooting, he/she will be permitted to shoot as many free-throws as the points that the shot would have been worth, had it gone in successfully. If the foul is committed by a player of the team in possession of the ball, e.g. by charging into a defensive player, it is called an offensive foul and results in loss of ball possession for the team of the offending player. In contrast, defensive fouls are those committed by players of the team that does not have possession of the ball and include intentionally moving in front of and making undue physical contact with the ball carrier to hinder his/her progress. Technical fouls are umbrella terms that cover a wide range of infractions such as hanging on the basket rim for any reason other than preventing injury to another player, excessive timeouts and undue delays of game. Nonetheless, many if not most, are related to the conduct of players including unsportsmanlike acts such as disrespectfully addressing or making physical contact with a referee, overreacting concerning a made call, and using profanity or taunting. An unsportsmanlike foul is called when a referee interprets the contact as unnecessary, and it is equivalent to a flagrant foul type 1 in the NBA, resulting in two free throws and ball possession for the opposing team. A disqualifying foul is called when the contact is deemed unnecessary and excessive, and results in the 10 granting of two free throws and ball possession to the opposing team as well as the ejection of the offender. A player will be ejected if he/she commits any of the following: five personal fouls, two technical fouls, two unsportsmanlike fouls, one technical foul and one unsportsmanlike foul, or one disqualifying foul. Scoring A shot made inside the three-point line is worth two points, and one made behind it is worth three points. If the player is fouled in the act of shooting and makes the shot, he/she will receive the points and take one free throw as a bonus. Furthermore, a made shot will still count if it left the player's hand and time expired while the ball was still in the air. Table 1. Rules comparison (from FIBA 2014) 2.2.2. Major competitions The main basketball leagues are the NBA (North American league), the Euroleague (European league) and the Liga Endesa (Spanish professional league). Contrary to sports such as football, in basketball the champion is not the team that accumulated the highest number of points at the end of regular season, but the one that wins the playoffs. To go to the playoffs, a team has to ensure a certain position in regular season. In the NBA there are 30 teams, 15 per conference. In each conference, 8 teams go to the playoffs where the first seed will face the eighth, the second will face the seventh and so on. Each round is the best of seven games and whoever wins 4, goes on to the next round. In figure 2 is an example of last year’s NBA playoffs’ bracket. 11 Figure 2. 2016 NBA playoffs' bracket The biggest event at national team level is the Olympic Games, where teams are distributed into two groups of six teams each, according to their ranking. The top four in each group advance to the quarter-finals, where they will face a team from the other group. The rationale remains the same, with the first in group A playing the fourth in group B, the second in group A playing the third in group B and so on. Unlike the NBA, the rounds are decided with only one game; the team that loses is eliminated. If a team loses in the semi-finals, it will be given the opportunity to compete for the bronze medal. The team that has come first in the Spanish league and the Euroleague more often than any other is Real Madrid with 33 and 10 wins, respectively. Regarding the Olympic games, not surprisingly, the team that won the gold medal most often was the United States national team with 15 victories. 2.2.3. Trends and challenges Basketball is changing. Today, it is a much faster game than it was 20 years ago, with more possessions and more 3pt shot attempts. To demonstrate this evolution, all regular seasons from 2000-01 to the current one were analysed in terms of 3pt shot attempts, 3pt shots made, 3pt field goal percentage and the percentage of points obtained from 3pt shots. With regard to 3pt shot attempts, whereas in 2000-01, on average, a team took 13.7 3pt shots per game; in the current season this average increased to 29 shots per game. The latter figure represents two more shots than in the previous season (2016-17) and more than double the figure for 2000-01. This increase is also shown by the fact that in the current season the team with the lowest number of 3pt shot attempts per game (22.5 attempts) still surpasses the team with the highest number of 3pt shot attempts per game (19.9 attempts) in 2000-01. As expected, the number of 3pt shots made also increased from 4.84 to 10.49 in the time frame considered. In terms of percentages, from 2000-01 to 2017-18 the number of 3pt shot attempts and that of 3pt shots made rose by 111% and 116%, respectively. 12 Figure 3. Average number of 3 point shot attempts per game in each season These findings are directly correlated with the percentage of points obtained by each team from 3pt shots. On average, this season, nearly 30% of the points scored by a team in a single game are from 3pt shots. This is the highest percentage of the past 17 years and is almost twice that of 2000-01 (15.28%). Figure 4. Average percentage of points resulting from 3 point shots per season However, when it comes to the percentage of 3pt field goals, there is not such a marked upward trend. In fact, from 2000-01 to the current season, it only increased by 2 percent points. The 2008-09 season remains the one with the highest percentage (36.54%) in the last 17 years. 13 Figure 5. Average percentage of success of 3 point shots per season These statistics help us understand how the game is evolving. For instance, they clearly show that while teams are taking more shots from behind the 3pt line, they are not being more efficient in scoring from that distance. In other words, the trend is the increasing use of 3pt shots and the challenge is to shoot more efficiently from that distance. Data supporting figures 3, 4 and 5 can be found in Annexes. 2.2.4. Shooting critical components In 2015, a literature review was made by Okazaki to identify the factors behind a successful jump shot. In order to do this, the authors divided the potential factors into three categories: ball trajectory, segmental movement organisation and variables that influence shooting performance. Below is a summary of the authors’ discussion regarding each of these categories. Ball trajectory Upon examination of ball trajectory, three components stand out as decisive for a successful shot: release angle, velocity and height. 14 Figure 6. Variables that influence the basketball jump shot (from Okazaki, Rodacki, & Satern, 2015) The angle of entry of the ball into the basket is one of the most important factors for shooting success. This is due to the fact that by increasing the angle, one instantly increases the width of the basket, giving the ball a larger area to go in. There are three variables that, together, determine this angle: vertical displacement, horizontal displacement and velocity (Victor H A Okazaki et al., 2015). Figure 7. Virtual target of the basket rim as a function of the angle of entry of the basketball (from Okazaki et al., 2015) The vertical displacement of the ball is negatively correlated with the release height and positively correlated with the release angle. In other words, the lower the release height and/or the wider the 21 Figure 14. Phases of the jump shot (from Okazaki et al., 2015) Preparation phase This phase is the starting point in which the player should have his/her elbow and wrist semi-flexed and hands positioning the ball at waist level with fingers spread out to facilitate the slide through at release. The feet must be spread shoulder width apart and the body weight evenly distributed between both feet. Some players opt to have the foot, on the same side as the shooting hand, slightly ahead of the other to increase stability and reduce upper body rotation during the act of shooting. If a player follows these steps prior to receiving the ball, he/she is prepared to shoot as soon as he/she catches the ball without having to make any adjustment. The main advantage in this is that the defensive player, more often than not, does not have time to contest the shot giving the shooter less opposition. This phase ends when the player starts to elevate the ball (Victor H A Okazaki et al., 2015). Ball elevation phase In this phase, the player starts to flex his/her shoulder and elbow as he/she positions the ball for release. The wrist does not show signs of variation, rather it is the elbow joint that starts to decelerate as the shoulder flexion increases. As the ball is lifted up, it should be close to the player’s body so as to reduce the body’s horizontal displacement. In addition, the athlete must have his/her shoulder, elbow and wrist aligned during the process and make sure that the elbow is below the ball at all times. This phase ends as the elbow becomes steady and ceases its flexion (Victor H A Okazaki et al., 2015). Stability phase The stability phase takes place when the elbow stops flexing and begins its extension. This movement is in sync with the deceleration of the shoulder joint’s extension, enabling the upper body to be balanced while the ball is being positioned for its release. While in the beginning of this phase, the wrist joint is still accelerating as it extends, as the movement progresses it starts to slowdown in line 22 with the other upper body joints (elbow and shoulder). It is not just the upper body that is in motion in this phase, the lower limbs extend at the same time. Players that need to generate more force during their shot, might shorten or even skip this phase to benefit from the energy created by the elbow’s extension (Victor H A Okazaki et al., 2015). Release phase The release phase begins with the extension of the elbow followed by wrist flexion and ends when the ball is no longer in contact with the shooting hand. Supported by the stability that was gained in the previous phase, at this stage joints will accelerate their movement once again. Elbow extension is identified as the most important moment of the shot, with the full extension of this joint being commonly observed among expert players. Concerning wrist flexion, some authors argue that it should take place shortly after the elbow is extended, while others propose that it should occur before the elbow’s full extension. When the shooting hand launches the ball, it must be done through finger and wrist flexion in order to apply the desired trajectory as well as backward rotation of the ball during flight. As Okazaki explained, “greater wrist flexion, coupled with the application of rotation to the ball at release, has been observed as a characteristic of the performance of expert players” (Victor H A Okazaki et al., 2015). It is very important that the ball is released at the peak of the jump because at that moment the player is in his/her highest and most stable point. When the ball is released before this brief period, the athlete is passing the vertical velocity of his/her body into the shot (thus skipping the stability phase as seen before) and while it enables him/her to shoot deeper, it also affects his/her balance and consequently shot accuracy. If the ball is released after the peak of the jump, the amount of strength that the upper body has to generate is higher, resulting in greater velocity from the joints at release and therefore less movement accuracy and consistency (Victor H A Okazaki et al., 2015). Inertial phase Finally, the inertial phase commences when the ball is released and is no longer in contact with the shooting hand; and ends when the player lands on the floor. It represents the gesture taken during release and is defined by shoulder flexion and complete flexion of the wrist, elbow extension and the hand parallel to the floor with fingers pointing towards the basket (Victor H A Okazaki et al., 2015). Additional variables In Okazaki’s literature review on the basketball jump shot, a number of variables that have impact on shooting mechanics were identified and, therefore, should be considered when acknowledging an attempt. Physical characteristics Physical characteristics that influence the performance of the basketball jump shot include the player’s height, limb length, and ability to generate force and velocity. While player height and wingspan influence ball release height in that taller players tend to shoot from higher heights, if taller players do not use their lower limbs as expected, by jumping, they will not benefit from their natural advantage and smaller players might actually release the ball higher as proven in (Miller & Bartlett, 1996). Players that are not able to generate the required velocity at release have to produce more 23 force to compensate for it. However, as previously mentioned, it should be borne in mind that by increasing the force or velocity at release, an athlete is also increasing movement variability, consequently affecting the shot’s success (Victor H A Okazaki et al., 2015). Fatigue According to Allen fatigue results in the decline of muscle performance caused by its intense activity. Its main consequences are reduced force production, decreased velocity of shortening and slowed relaxation. The combination of these elements is associated with great reductions in performance, especially when it comes to rapidly repeated movements (Allen & Westerblad, 2001). Muscle fatigue causes a decrease in the muscle activation pattern, affecting the sense of joint position which influences balance (Abd-Elfattah, Abdelazeim, & Elshennawy, 2015). A number of studies were carried out in order to assess whether or not fatigue had an impact in jump shooting mechanics and accuracy. The methodology used in most of these studies consisted in having the player perform jump shot series as the level of fatigue increased. To ensure this, between each series, the player engaged in physical activity like jumping and running. In what regards shooting motion, some important observations were made. Firstly, the elbow joint angle decreased resulting in a decrease of the wrist and elbow’s height at release. In the last series of one of the studies, there was a decrease of as much as 21 degrees in elbow joint angle that resulted in a decrease of a shoulder average height of 11 cm and a decrease of wrist average height of 16 cm, when compared to the first series (Frane & Matej, 2009) Other observations include the fact that shoulder joint angle increased; jump height differences between series were statistically significant since the athlete jumped increasingly lower as the series progressed, and hip joint angle decreased (Frane & Matej, 2009) (Slawinski, 2015). Even though it was proven that fatigue had an impact on the kinematics of jump shooting, and in one of the studies the number of made field goals was at its lowest in the last series (maximum fatigue), it was not possible to establish a relationship between fatigue and shot accuracy (Victor H A Okazaki et al., 2015). Shooting distance Shooting distance has an impact on the act of shooting. When it changes, the shooter makes adjustments to his/her gesture, and changes in release height, angle and velocity are observed. As the distance increases, the player has to achieve greater velocity at release which causes greater variability of movements and affects performance (Victor H A Okazaki et al., 2015). In 2012, a study was conducted to evaluate the impact of shooting distance. In this study, 10 expert players performed jump shots from three different locations: close distance (2.8 m), intermediate distance (4.6 m) and far distance (6.4 m) from the basket. Researchers collected information on kinematics using a digital camcorder to capture the act of shooting followed by standard 2D analysis. The results showed that shot accuracy decreased from 59% to 37%, from close to long range. Ball release height decreased from 2.46 m to 2.38 m, when comparing close to intermediate range, and reached a minimum of 2.33 m when players shot from long range. As expected, release angle decreased from 78.92 degrees (close range) to 65.60 degrees (intermediate range). In contrast, ball release velocity increased from 4.39 m/s (close range) to 5.75 m/s (intermediate range), reaching a maximum value of 6.89 m/s from long distance (Victor Hugo Alves Okazaki & Rodacki, 2012). 24 The changes in release height, angle and velocity of the ball caused by movement adaptations were suggested as the main factors that influence jump shot accuracy when distance varies (Victor H A Okazaki et al., 2015) (Victor Hugo Alves Okazaki & Rodacki, 2012). Presence of an opponent In basketball, there is a high probability that the player in possession of the ball will have to perform a jump shot while being guarded by a defensive player. When facing such opposition, and to increase his/her chances of success, the player has to make adjustments in the act of shooting which mostly include: higher release angle of the ball, greater elbow extension, higher release height, quicker release and lower vertical displacement from the centre of gravity (Rojas, Cepero, Oña, & Gutierrez, 2000) (Victor H A Okazaki et al., 2015). In this scenario, players are focused on releasing the ball before facing opposition from their opponent in order to reduce the chances of the shot being blocked. To do so, athletes tend to have a more upright position at the beginning of the upward movement of the ball, enabling them to initiate the shot from a greater height. Players also have a more stable base for generating greater initial velocity of the ball (Rojas et al., 2000). In this case, the knees are not flexed as usual, translating into a shorter but quicker jump. Even though players opt for a quick and high release, it is also observed that when they release quick enough, the release height can be lower since there is an absence of contest given that the opponent does not have time to challenge the shot (Victor H A Okazaki et al., 2015). In conclusion, since it has been proven that opposition influences the mechanics of the shot and consequently performance (Victor H A Okazaki et al., 2015), it stands to reason that training would benefit from practice with an opponent to prepare players to successfully deal with the demands that they are most likely to face in a game situation (Rojas et al., 2000). 2.2.5. Technologies in Basketball 2.2.5.1. Current situation Today, the NBA is considered one of the most technologically-advanced sports leagues in the world. Like in NFL and other major North American leagues, technology plays a key role both during training sessions and during the course of the actual games. The use of wearable technology, such as ClearSky and Optimeye by Catapult, Zebra wearable tags or Zephyr Bioharness systems, is already an everyday practice for NBA teams, as pointed out by Jen Booton (Booton, 2017). In her post, Jen identifies the main technology storylines of 2017. Regarding wearable technology, due to the increase in its use, the NBA has imposed limits on the scope of data provided by such equipment. Data can only be used to monitor players’ health and performance, and it is prohibited to use it for commercial purposes or contract bargain and to make it public. Another added technological innovation was the emergence of VR. VR was intended to improve the experience of fans and refereeing. The NBA broadcasts to more than 200 countries and VR allows fans to watch games from the first row or the most spectacular angles, enabling every fan to feel like he/she is in the arena. In 2017, once a week, a regular season game would be made available in VR to NBA League Pass users. 25 But the use of VR goes beyond the fan’s experience; it is also aimed at refereeing. In this context, the NBA has decided to use VR and data-driven analytics to recruit and train referees, since it is capable of monitoring the accuracy of calls made by referees and, consequently, tracking the amount of errors made (Booton, 2017). In the NFL, referees have a VR training platform. However, the role of technology with regard to refereeing does not end there. In the 2014-15 NBA season, the NBA Replay Centre was launched. The Replay Centre provides the officials with the best angles to inform their calls, the objective being to improve the accuracy and response time of the calls. According to the NBA’s official website, the Replay Centre is equipped with 94 HD monitors and bandwidth capacity to download the entire digitized Library of Congress (more than 158 million documents) in over 30 minutes. In the Replay Centre, there are 17 replay operator stations and 3 replay manager stations making up a total of 20 work stations, not including the referees. To make this possible, all 29 NBA arenas are connected to the Replay Centre, which allows every fan in the arena to see the same replays that the officials are watching and contributes to a better understanding of why the call was made. Last season, the NBA decided to expand the role of this facility by enabling active referees in the Replay Centre to make the calls after review in every case except for player altercations or flagrant fouls, as stated in the NBA’s official website. With greater responsibility in making judgement calls and an optimized process to display the replays, the average review time dropped from 42 to 31.9 seconds. Currently, all in-game statistics displayed on the NBA official website are provided by SportVU, a player tracking system. Each arena used by the NBA has six cameras that capture the exact position of every player and ball at a rate of 25 frames per second. The data provided by this system is analysed and stored within specific software, enabling data to be shared in real time. With the use of this tracking system, not only is it possible to provide precise data that is not accessible in other ways, but it is also shared with teams and some media outlets like ESPN or Bleacher Report. Some examples of complex data recorded by SportVU are the amount of touches per 100 possessions by each player, the shooting percentage of the opposite team while a certain player is on the court, or even the percentage of rebounds won by a player when facing opposition. These statistics provide more detail than ever and enable a greater understanding of the game instead of the usual number of points, rebounds and assists by each player. The main advantage of data being provided in real time is that it feeds information to the coaching staff that can lead to possible changes in the line-up, as a result of what is happening in the game, which supports decision-making during the game. 2.2.5.2. IoT in Basketball Evo One Evo one is a smart basketball that is meant to give feedback to the user upon the act of shooting. When the user shoots the basketball, he will hear a beep if the ball’s backspin rate is 2-3 rotations/second (the ideal rate). The purpose of this feature is to let the user know that his attempt was successful and create shot consistency through muscle memory. It measures the ball’s backspin using only one sensor located inside the ball. While there are already a number of reviews on this device, the performance product review website Weartesters provides possibly the most succinct and accurate summary of its pros and cons. 26 According to them, apart from the fact that the ball has been made according to regulation size and weight making it no different to a common basketball, the main pro identified by the consumers is its leather cover which gives the user a good grip and touch. Having said this, a leather cover makes it more of an indoor ball than an all-surface ball. Regarding the cons, there are some considerable ones such as the dead spot on the ball where the sensor is inserted, which affects dribbling capacity and turns it into a simple catch and shoot ball, preventing it from being used in a game. However, for the purpose of shooting this is not a major concern. Another important issue is the inconsistency of the sound feedback. Indeed, the ball does not provide feedback unless the player’s fingers are aligned with its ribs even if the shot is perfect in terms of backspin, which poses serious questions regarding the feasibility of its use in a real game situation where quick catch and shoot is fundamental and players do not have time to adjust their grip. In addition, it is unable to differentiate between a pass and a shot attempt, i.e. should the fingers be aligned, it will beep whenever it reaches the ideal backspin, even if it was a pass. Furthermore, it has been noted that the lack of consistent and accurate feedback can promote some bad habits in shooting motion (Weartesters, 2017) (Evo1, 2017). Figure 15. Evo one smart basketball Wilson X The main purpose of the Wilson X smart basketball is to track field goal accuracy. It is made to regulation size and weight with a solid grip and durability, suitable for both indoor and outdoor use. The sensor is embedded within the ball and does not require charging (lasting for 100,000 shots). In addition, the product comes with a mobile app which enables the user to track his/her performance on his/her mobile phone via Bluetooth connection. However, there are some limitations. For instance, in order to ensure tracking accuracy, the ball must go through a hoop at least 10ft tall, with a tight net and rigid backboard, and it must hit the floor. A number of consumer reviews highlight two important issues. The first one is related to inaccurate tracking of field goal percentage and attempts. In actual fact, the ball has been reported to record some made shots as missed and viceversa, and to not acknowledge some made attempts at all. While Wilson estimates that the ball has an accuracy of 97%, the American technology news and media network – The Verge argues that it is only around 80%. This is not great news since the main purpose of the ball is to monitor field goal percentage. The second main concern is related to the synchronization between the ball and the user’s mobile phone. Users have described the process as more difficult than advertised and plagued with interruptions throughout. Another criticism has to do with the fact that the app only provides a basic shot chart showing the player’s accuracy rate from specific shooting ranges, but not from 27 different shot angles or locations within the same range which could be very useful especially for players who shoot better from different positions (O’Kane 2015). Figure 16. Wilson X smart basketball Figure 17. Wilson X application modes On a more positive note, the app provides four interesting modes that will keep the user engaged while practicing. The first mode is called free range and is essentially a shoot-around game tracker that shows shooting percentage as well as distance from the hoop. The free throw mode tracks made/missed shots in real time from the free throw line. In the buzzer beater game mode, the clock keeps ticking and the player is required to repeatedly shoot under pressure as every shot made adds seconds to the clock extending the game experience. Finally, the game time mode recreates a real game environment including actual crowd sounds and commentary. It tracks both field goal percentage and points, and lets the player know whether his/her effort was sufficient to secure the victory. Although these game modes attempt to make the user more familiar with playing under pressure, without another actual player guarding and blocking the user in real life, their help in improving game performance remains limited. 28 94fifty The 94fifty smart basketball enables the user to check his/her shot arc, rotation and release speed. It is a ruggedized ball of regulation size and weight, with a good grip suitable for both indoor and outdoor use. It does have a dead spot that can be found while dribbling but, as for Evo 1, this is not a major concern for shooting practice alone. While, at first, the ball was unable to track made or missed shots, with the increment of the smart net, this function has now been made possible. Furthermore, unlike the previous basketballs, this one can be charged wirelessly and has approximately 8 hours of battery life with continuous use. Following performance analysis of some of the best basketball shooters, the developers concluded the optimal range for shot arc to be between 42 and 48 degrees, ideal backspin between 130 and 150 rotations per minute and best release speed under 0.7 seconds, and they calibrated the ball accordingly. When shot, the ball provides instant feedback regarding what is being measured. Thus, if it is arc, there will be positive feedback if the player’s shot arc falls within the optimal range, and an alert message if not. In addition, the app offers a wide variety of drills designed to improve performance in specific categories, such as dribbling and shooting. In the latter category, on which this report focuses, there are various drills available with different levels of difficulty - from playground all the way to professional, and there can be combinations of shot accuracy with shot arc, release speed or shot rotation. However, there are limitations to what the ball can measure at any given moment. For instance, the ball can only measure and display one indicator at a time and is not equipped with certain useful tools such as GPS for determining shooting location. Hence, before performing a shooting drill, the player needs to select what indicator to measure, identify his/her shooting location and whether it is a free throw or not (e.g. jump shot). Nonetheless, overall the range of workouts offered by the app does facilitate skill improvement and player engagement (Fritz 2014). Figure 18. 94fifty gear In studies that were carried out to assess the reliability of the 94fifty, it was concluded that it is not only possible to use it for practical purposes but also for scientific reasons given the accuracy of the obtained results. It was also considered an important instrument for the overall improvement of the basketball training process since this device allows the user to receive valuable feedback (Tomislav Rupčić, Ljubomir Antekolović, Damir Knjaz, Bojan Matković & Cigrovski, 2016). 29 ShotTracker ShotTracker is a shooting tracking system that registers missed and made shot attempts. For it to work, the user has to have a wrist sensor that captures shot attempts and a net sensor that registers whether or not the ball went in. Since there is no need for basketball sensors or smart balls, the player can choose his/her basketball of preference, which is an advantage as smart basketballs have been criticized for having different grips and not always having the touch of a regular ball. However, the fact that the net sensor has to be attached to the net which typically is 10ft above the ground may become a major inconvenience, especially in public places, since a ladder would be needed to reach the net and attach the sensor. Furthermore, if there are multiple players shooting at the hoop with the net sensor, all the shots will be taken into account, and this presents a problem for measuring individual performance. Nevertheless, ShotTracker remains user friendly in other ways, coming with a wrist band, a sleeve and even a shirt giving the user different options to correctly wear the sensor. In addition, both sensors can be charged at the same time and are weather proof. With this equipment, the user is able to track his/her field goal percentage and shooting location. Figure 19. ShotTracker gear In addition, the app enables the user to look at his/her results in real time and assess progress over time by checking personal daily, weekly and monthly stats. Another great advantage of the app is that coaches can monitor the players’ results and, based on the feedback from the drills, identify areas of weakness and customise workouts to the needs of each player (ShotTracker 2017). Hoop tracker The Hoop Tracker is a basketball shot tracking smartwatch which provides real time feedback on the player’s shot attempts. It detects shooting location at all times and whether the shot was made or missed. 30 Figure 20. Hoop tracker gear In order to do this, only two pieces of equipment are required: a wireless shot detector and a smartwatch. The shot detector is held in place by a powerful magnet designed to not impact the outcome of the shot which is a key feature. It also comes with a mounting pole which enables the user to place it on the rim safely from the ground, giving it an advantage over the Shot Tracker. The wristwatch, although lightweight and intended to be worn on the non-shooting hand to minimise the chance of damage and obstruction to the shot, is still an accessory that is not used in a game situation and, therefore, not ideal from this perspective. However, it can be quite useful as real time results are only a quick glance away during workout. When the ball goes through the hoop, it activates the sensor, which subsequently sends a signal to the watch that the shot was made. When a shot is missed, the vibration created by hitting the backboard or the rim is detected by the accelerometer sensors which send a signal to the watch that the shot was missed. Moreover, the developers incorporated a delay in the signal to account for the shots that bounce around the rim before going in. The only shot it cannot automatically detect is the air ball, for which there is a button on the watch that can be pressed to record it as a miss. With access to shooting percentages from different locations, players can evaluate their success rate from different positions on the court e.g. free throws, three-pointers, etc. and take advantage of the training modes available for those positions. Once their data is uploaded, they can access the Hoop Tracker dashboard on their own computer and analyse their stats, track their long-term progress and identify their strengths and areas for improvement. There is also a coach mode which allows coaches to monitor their players’ progress and customize training accordingly (Isom 2014). 37 Table 4. Present Artifact Matrix 38 Knowing shot location and accuracy helps to obtain a better understanding of where players are performing at a higher level, nevertheless, it does not show what was done wrongly and how it can be improved. Given that there has already been a large focus on ball trajectory, the following suggestions will rely heavily on the physical movements of the player upon the act of shooting (shooting mechanics) and the factors that can lead to differences in this behavior. In this endeavor, the findings from the literature review on the jump shot have been taken into account. Wristband sensor From the studied devices, only the ShotTracker used a wristband to count the number of shot attempts made by a player. But while this already helpful, it could be explored further. If extra sensors were incorporated into the wristband, which is feasible due to the increasingly smaller sizes of these devices, important data like release height and wrist angle at release could be obtained. As previously stated, release height in combination with release angle, are key factors that help to explain the angle of entry of the ball into the basket. Another valuable usage of the wristband might be to evaluate how the player performs when facing an opponent. By determining the closeness between the player’s shooting hand and the hand of the defensive player as well as the height of the blocking attempt, it will be possible to gain a greater understanding of how shooting mechanics are influenced by opposition and its consequences on ball trajectory. Metrics: release height, wrist angle, closeness of contest, height of contest, shoulder/elbow/wrist plane Vert sensor The Vert sensor has the ability to track the jumping activity of an athlete as it retrieves the maximum height achieved during the jump. It would be very useful to have communication between the vert sensor and the wristband at release; this way it would be possible to know the exact jump height reached by the player at release and how close it is to the desired value which is the maximum, as the player should release the ball at the peak of the jump. The reliability of this sensor has already been tested, and it was considered fit to measure jump height in in-game situations for both basketball and volleyball. It also has the ability to provide real time feedback (Mahmoud, Othman, Abdelrasoul, Stergiou, & Katz, 2015) (MacDonald, Bahr, Baltich, Whittaker, & Meeuwisse, 2017). Metrics: maximum jump height, jump height at release Sleeve sensor Nowadays, it is possible to measure joint angle with high precision using low-cost wearable inertial sensors (El-Gohary, 2013). Elbow extension plays a major role in release height and release angle. Since the sleeve is a common accessory in basketball with no scientific use, it would be helpful to integrate an inertial sensor into the sleeve that could correctly measure the elbow’s angle at release, by communicating with the wristband sensor. Metric: elbow angle, shoulder/elbow/wrist plane 39 Jersey sensor Currently, jerseys are also only being used as an accessory without any scientific advantage. As suggested for the sleeve, if a sensor was integrated in the shoulder joint, it would allow the tracking of the shoulder angle at release. Trunk inclination should be close to vertical at release, as this is a performance characteristic seen in expert players. Another good practice drawn from the observation of expert players in action is having the shoulder, elbow and wrist aligned in the same plane of motion (Victor H A Okazaki et al., 2015). It is important to note that metrics like trunk inclination and shoulder joint angles have already been studied with the use of sensors (Bergamini et al., 2013; Gert S. Faber, 2005; Yoon, 2017) and therefore it would be theoretically feasible to measure them through sensor integration into the jersey as suggested above. Metrics: shoulder angle, trunk inclination, shoulder/elbow/wrist plane In this report, two artifact matrices will be presented, one for amateur level and one for professional level. The main difference between them is that for professional level it is assumed that players train in indoor facilities that are fully capable of having technology like catapult or Noahlytics installed. On the other hand, amateur level recommendations do not require facilities to perform measurements. Smart ball From the list of analyzed devices, the smart ball that offers the most comprehensive variety of metrics measurements is the 94fify. The ability of the ball to measure release speed, shot arc and ball rotation at once make this product very useful to amateur players that do not have access to the tools and technology made available to professional athletes. Furthermore, its ability to autonomously identify if the shot was successfully made, coupled with the metrics previously mentioned, provide a great level of detail when merged with metrics seen for previous wearable sensors. Metrics: shot arc, ball rotation, made or miss Noahlytics Noahlytics has revolutionized shooting tracking in basketball. The detail of the metrics it measures is so great that new factors that influence shooting performance have been identified (Marty & Lucey, 2017). By merging the metrics of this system with shooting metrics captured by wearable inertial sensors, it is possible to establish the relation between ball trajectory and shooting motion. For the first time, analysts will be able to have detailed input on the end to end process of the jump shot. By running appropriate statistical tests, patterns can be discovered based on statistical evidence of how shooting parameters explain trajectory elements. Metrics: shot location, entry angle, shot depth, left-right position Catapult One factor that was shown to influence shooting mechanics is fatigue. It was demonstrated that fatigue has an impact on the ability of muscles to respond to physical exertion, resulting in a decrease of the elbow joint angle that causes lower angle and wrist height at release and lower jumping ability. This being said, researchers failed to identify a correlation between fatigue and 40 shooting accuracy. Since the metrics used were heart rate and other such factors, it would be interesting to see if a more complex metric like PlayerLoad that is an acceleration-based metric unlike the regular distance-based ones, could have different results in this regard. Metrics: PlayerLoad, heart rate Figure 26 and table 5 show at what point each shooting mechanic metric will be measured by the corresponding sensor. Figure 26. Suggested use of sensors for the measurement of metrics 41 Table 5. Suggested measurement of metrics by phase and sensor Feedback By comparing the devices under analysis in what concerns audio feedback, it is possible to conclude that audio feedback is given based on the outcome of the shot. It is either provided in the form of ‘beeps’, as in Evo One, to stimulate muscle memory or by giving the exact metrics concerning how the ball went through the rim, as in Noahlytics. Even though this type of feedback allows players to create muscle memory, it is not the most advised. This is because the majority of users that are training to improve their jump shot are not expert players, so the possibility that they do not possess a good understanding of how each movement impacts the outcome of the shot must not be ignored. Taking into account that today’s technology enables the monitoring of players’ shooting mechanics, valuable feedback can be provided in this regard. In 1993, a study was conducted to demonstrate how to provide appropriate feedback concerning the jump shot technique. Six key technique points were identified as critical in the shooting motion and adequate feedback was determined for each based on understandable terms. As an example of what was proposed, for the two critical points “optimize height of release” and “ball rotation”, the suggested terms to use in the feedback were “extend at the top of the jump” and “flip the fingers” under the ball at release, respectively (Knudson, 1993). Comprehensible feedback as illustrated above can improve the understanding of the player and generate knowledge on the best practices of the jump shot. Table 6, shows the suggested audio feedback for different metrics. 42 Table 6. Audio feedback suggestions 4.2. PROPOSED ARTIFACTS In the following tables the two proposed artifact matrices, one for the amateur level and one for the professional level, are presented in summary form. 43 Table 7. Future artifact matrix for amateur level 44 Table 8. Future artifact matrix for professional level 45 4.3. VALIDATION The validation of the proposed artifacts in this report was conducted via interview. Since the technology topic is specific, instead of using a questionnaire it was decided that the interview method would be more suitable as the interviewer would be able to give context and collaborate with the interviewee during the process. 4.3.1. Target Audience The target audience for the interviews was current or former basketball players, coaches or journalists. The probability of these groups having a better understanding of what technology could benefit the players’ shooting mechanics is greater than the common sports fan. In principle, by interviewing players and coaches with different levels of experience, conclusions might be drawn on how the level of expertise is related to the openness to the adoption of technology for training performance. 4.3.2. Sample As mentioned before, the sample was composed of former players and coaches, basketball personas that are related to the sport. However, due to the absence of response when contacting professional teams and network stations, the only participants in the interview process were of an amateur level. Although this does not mean that the interviewees were not qualified to participate in the study or able to provide adequate feedback, it would nonetheless have been extremely useful and relevant to compare and contrast the points of view of amateur and professional athletes and coaches. This is therefore seen as a limitation in this report. The volunteers that participated in the interviews were two former players with different levels of experience, one former coach with significant experience as a player and one current coach with a high level of experience. Table 9. Overview of personal information and basketball experience Variables used in table 9: • Participant: ID of the participant • Gender: M for male, F for female • Age: age of the interviewee in years • College player: number of years played in college • Club player: number of years played in a club 46 • Coach: number of years in a coaching capacity 4.3.3. Summary of interview guide The purpose of the interviews was to validate the recommendations about how current and new technology can improve jump shooting performance. The first page was designed to welcome the interviewee, give some context on the objective of the study and collect personal information such as the age and gender of the participant as well as the level of expertise/experience in the area. The following question introduced the devices under analysis in the literature review and asked the interviewee to identify which devices he had knowledge of. The next question required the participant to examine tables 3 and 4 which show every metric calculated by each sensor currently in use and, upon examination of the data, state whether he believed the metrics were relevant to the improvement of shooting mechanics and justify the answer. In the following question, the interviewee was presented with another table (table 10), where the metrics considered relevant in this literature review were displayed. He was then asked to assess each metric according to how relevant it is to the success of the jump shot. To do so, the following Likert scale was provided: 1 – Strongly Disagree, 2 – Disagree, 3 – Agree, 4 – Strongly Agree. This particular version of the scale was chosen because it was even and did not offer a neutral answer. Since there is sometimes a tendency of respondents to pick this answer to avoid further justification, this option was removed so that the interviewer could receive as much feedback as possible from the participants. Table 10. Interviewees' views on the impact of suggested metrics on the success of the jump shot 53 The updated artifacts matrices are shown below, as well as the audio feedback table, with feedback resultant from the validation method. Table 14. Updated artifacts' matrix for amateur level 54 Table 15. Updated artifacts' matrix for professional level 55 Table 16. Updated table of audio feedback 56 5. CONCLUSIONS 5.1. SYNTHESIS OF THE DEVELOPED WORK The objective of this study was to recommend new uses for current sensors and new sensors that would result in improved shooting performance in basketball. Given that this is a theoretical study, the chosen method was design science research, a methodology that relies on four major premises: the search for flaws, the making of suggestions to correct the identified flaws, the validation of the suggestions made, and lastly the update of the initial proposals taking into account the feedback received during validation. In the first phase, the shooting factors behind a successful jump shot were identified, after which an investigation was conducted to determine what gadgets with sensor technology are currently being used and what metrics are being measured. After comparing the indicators that are being measured by these devices with the factors behind a successful jump shot, it became evident that there is a big gap to be filled as the shooting technique metrics remain unmeasured, which is of no help whatsoever to the user when it comes to understanding how he/she can improve his/her performance. By taking into account the metrics that need to be captured, the measuring capabilities of currently used devices and the potential uses for new ones, two artifacts were proposed, one for the amateur level and another for the professional level. In order to comprehend if the suggestions that were put forward made a real contribution to the resolution of the problem, a set of interviews was conducted. Analysis of the results of the interviews indicates that the metrics provided by current devices enlarge the users’ understanding of the locations on the court where they are more efficient and allow them to track their evolution through time; however, they provide no valuable feedback that would enable the users to improve their shooting technique. Despite the small sample size, the results from the interviews add to the already mounting evidence that the use of IoT, having a set of sensors connected and registering specific metrics, in this context would benefit the user as he/she could have access to useful data explaining exactly how he/she is performing and what adjustments might be made to improve his/her performance. It was also mentioned that by combining outcome variables with shooting factors, relationships can be found and possibly predict the outcome of the shot based on shooting metrics. In addition, it was concluded that audio feedback can contribute to the improvement of the shooting technique, as it enables players to correct their motion through easily understandable commands instead of hearing what they can observe themselves  the outcome of their shot. Audio feedback was considered to be particularly important to adjust wrist flexion, elbow extension and the height of the jump that led to ball release. Finally, release speed, ball rotation and emotional status were the metrics suggested by the respondents to complement the proposed artifacts which were updated accordingly with suitable devices. 57 5.2. LIMITATIONS OF THE RESEARCH As this is an academic research report, some limitations were encountered and aspects that could be improved were identified. The major limitation concerning this study was related to the sample used for the interviews. Even though the coaches were experienced which contrasted with the players’ level of experience, it would have been very useful to have participants from both amateur and professional levels as well as journalists. This would have brought in different perspectives and provided valuable feedback about diverse aspects. Furthermore, having representatives from all levels within the sport, amateur, professional and academic world could have greatly enriched the conclusions. The invitations for the interviews were sent to professional teams and sports television networks, but no answer was obtained. Another limitation about the sample was that the respondents had no prior knowledge of the studied devices and, although the interviewer endeavoured to give them objective information in a clear and concise way about each device, participants still seemed overwhelmed and were not able to suggest new uses for current or new technology. It would have been interesting to analyse the feedback from a person with more experience in the field. 5.3. FUTURE WORK As for the next steps, I would consider repeating the interview process with professional athletes or people used to working with technology in sports in order to compare their feedback. The following step would be the implementation in a smaller scale of the proposed network of sensors. This would allow the collection of specific data that could help determine whether the shot outcome can be statistically predicted by mixing the proposed shooting metrics. 58 BIBLIOGRAPHY Abd-Elfattah, H. M., Abdelazeim, F. H., & Elshennawy, S. (2015). Physical and cognitive consequences of fatigue: A review. Journal of Advanced Research, 6(3), 351–358. https://doi.org/10.1016/j.jare.2015.01.011 Allen, D. G., & Westerblad, H. K. (2001). Role of phosphate and calcium stores in muscle fatigue. The Journal of Physiology, 536(3), 657–65. https://doi.org/10.1111/j.1469-7793.2001.t01-1-00657.x Bergamini, E., Guillon, P., Camomilla, V., Pillet, H., Skalli, W., & Cappozzo, A. (2013). Trunk inclination estimate during the sprint start using an inertial measurement unit: A validation study. Journal of Applied Biomechanics, 29(5), 622–627. https://doi.org/10.1123/jab.29.5.622 Booton, J. (2017) The 10 Most Important Technology Storylines From The 2017 NBA Season. https://www.sporttechie.com/the-10-most-important-technology-storylines-from-the-nba2017-season/ 2017 Cisco Systems. (2013). The Internet of Everything and the Connected Athlete : This Changes … Everything, 1–10. Retrieved from http://www.cisco.com/c/en/us/solutions/collateral/serviceprovider/mobile-internet/white_paper_c11-711705.pdf D., Slawinski J., Poli J., Karganovic S., Khazoom C., Dinu (2015). Effect of fatigue on basketball three points shot kinematics. 33rd International Conference on Biomechanics in Sports, (EA 2931), 4– 6. Dellaserra, C. L. (2013). Use of Integrated Technology in Team Sports: A Review of Opportunities, Challenges, and Future Directions for Athletes. Journal of Strength and Conditioning Research. El-Gohary, M. A. (2013). Joint Angle Tracking with Inertial Sensors. https://doi.org/10.15760/etd.661 Erculj, Frane; Supej, M. (2009). Impact of fatigue on the position of the release arm and shoulder girdle over a longer shooting distance for an elite basketball player. Journal of Strength and Conditioning Research, 23(3), 20. Espinosa, H. G., Lee, J., & James, D. A. (2015). The Inertial Sensor: a Base Platform for Wider Adoption in Sports Science Applications. Journal of Fitness Research, 4(1), 13–20. Evans, D. (2011). The Internet of Things - How the Next Evolution of the Internet is Changing Everything. CISCO White Paper, (April), 1–11. https://doi.org/10.1109/IEEESTD.2007.373646 Evo1 (2017). https://evo1sports.com/ FIBA Central Board (2014). Official basketball rules, FIBA Fritz, N. (2014). 94Fifty Smart Sensor Basketball Review https://www.livescience.com/43410-94fifty-smart-sensor-basketball-review.html Gabbett, T. J. (2013). Quantifying the Physical Demands of Collision Sports: Does Microsensor Technology Measure What it Claims to Measure?, 1–14. Gert S. Faber, I. K. and J. H. van D. (2005). Optimal inertial sensor placement for determining trunk inclination. Hevner, B. A. R., March, S. T., Park, J., & Ram, S. (2004). Design science in information systems research. MIS Quarterly, 28(1), 75–105. https://doi.org/10.2307/25148869 59 Hooptracker (2017) http://www.hooptracker.com/ Isom, J. (2014). Basketball Wearables: Comparing the ShotTracker and Hoop Tracker https://www.sporttechie.com/basketball-wearables-comparing-the-shottracker-and-hooptracker/ Johnston, R. D., Gabbett, T. J., Walker, S., Walker, B., & Jenkins, D. G. (2015). Are three contact efforts really reflective of a repeated high-intensity effort bout? Journal of Strength and Conditioning Research, 29(3), 816–821. https://doi.org/10.1519/JSC.0000000000000679 Knudson, D. (1993). Biomechanics of the Basketball Jump Shot—Six Key Teaching Points. Journal of Physical Education, Recreation & Dance, 64(2), 67–73. https://doi.org/10.1080/07303084.1993.10606710 Lacerda, D. P., Dresch, A., Proença, A., & Antunes Junior, J. A. V. (2013). Design Science Research: Método de pesquisa para a engenharia de produção. Gestao e Producao, 20(4), 1–21. https://doi.org/10.1590/S0104-530X2013005000014 Luteberget, L. S., Spencer, M., & Gilgien, M. (2018). Validity of the Catapult ClearSky T6 local positioning system for team sports specific drills, in indoor conditions. Frontiers in Physiology, 9(APR), 1–10. https://doi.org/10.3389/fphys.2018.00115 MacDonald, K., Bahr, R., Baltich, J., Whittaker, J. L., & Meeuwisse, W. H. (2017). Validation of an inertial measurement unit for the measurement of jump count and height. Physical Therapy in Sport, 25, 15–19. https://doi.org/10.1016/j.ptsp.2016.12.001 Mahmoud, I., Othman, A. A. A., Abdelrasoul, E., Stergiou, P., & Katz, L. (2015). The reliability of a real time wearable sensing device to measure vertical jump. Procedia Engineering, 112, 467–472. https://doi.org/10.1016/j.proeng.2015.07.226 Marty, R., & Lucey, S. (2017). A data-driven method for understanding and increasing 3-point shooting percentage. MIT Sloan Sports Analytics Conference, 1–14. Miller, S., & Bartlett, R. (1996). The relationship between basketball shooting kinematics, distance and playing position. Journal of Sports Sciences, 14(3), 243–253. https://doi.org/10.1080/02640419608727708 Muse (2017) http://www.choosemuse.com/what-does-muse-measure/ Newcomb, T. (2016) Tech Talk: New Catapult device tracks player movement for NFL, NBA https://www.si.com/edge/2016/03/01/tech-talk-catapult-data-analytics-nfl-nba-nextgeneration Noah Basketball (2017) Noah Basketball Wins Startup Competition At MIT Sloan Sports Analytics Conference https://www.bizjournals.com/prnewswire/press_releases/2017/03/08/CL31042 Okazaki, V. H. A., & Rodacki, A. L. F. (2012). Increased distance of shooting on basketball jump shot. Journal of Sports Science and Medicine, 11(2), 231–237. Okazaki, V. H. A., Rodacki, A. L. F., & Satern, M. N. (2015). A review on the basketball jump shot. Sports Biomechanics, 14(2), 190–205. https://doi.org/10.1080/14763141.2015.1052541 O'Kane, S. (2015). This smart basketball can calculate your shooting percentage for you https://www.theverge.com/2015/9/17/9347039/wilson-x-connected-smart-basketball Polk, T., Yang, J., Hu, Y., & Zhao, Y. (2014). TenniVis: Visualization for tennis match analysis. IEEE 60 Transactions on Visualization and Computer Graphics, 20(12), 2339–2348. https://doi.org/10.1109/TVCG.2014.2346445 Ray, P. P. (2014). Internet of Things based Physical Activity Monitoring (PAMIoT): An Architectural Framework to Monitor Human Physical Activity. Ieee Calcon, (November), 32–34. Ray, P. P. (2015). Internet of Things for Sports (IoTSport): An architectural framework for sports and recreational activity. International Conference on Electrical, Electronics, Signals, Communication and Optimization, EESCO 2015. https://doi.org/10.1109/EESCO.2015.7253963 Research, L. (2013). “ An Introduction to the Internet of Things ( IoT )” What Is The Internet of Things ( IoT ), (November), 1–6. Rojas, F. J., Cepero, A. M., Oña, L., & Gutierrez, M. (2000). Kinematic adjusments in the basketabll jump shot against an opponent. Ergonomics, 43(Hudson 1985), 1651–1660. Rowlands, D. D., McCarthy, M., & James, D. A. (2012). Using inertial sensors to index into video. Procedia Engineering, 34, 598–603. https://doi.org/10.1016/j.proeng.2012.04.102 ShotTracker (2017) http://shottracker.com/individual Singh, P. (2016). Internet of Things ( IoT ): A Literature Review, 3(12), 164–173. Retrieved from https://www.irjet.net/archives/V3/i12/IRJET-V3I12221.pdf Somayya Madakam, R. Ramaswamy, S. T. (2016). Internet of Things ( IoT ): A Literature Review. International Research Journal of Engineering and Technology, 3(12), 164–173. Retrieved from https://www.irjet.net/archives/V3/i12/IRJET-V3I12221.pdf Swan, M. (2012). Sensor Mania! The Internet of Things, Wearable Computing, Objective Metrics, and the Quantified Self 2.0. Journal of Sensor and Actuator Networks, 1(3), 217–253. https://doi.org/10.3390/jsan1030217 Tomislav Rupčić, Ljubomir Antekolović, Damir Knjaz, Bojan Matković, V., & Cigrovski. (2016). Reliability analysis of the 94 fifty smart sensor basketball. Weartesters (2017). http://weartesters.com/shooters-revolution-evo-one-product-review/ Wundersitz, D. W. T., Josman, C., Gupta, R., Netto, K. J., Gastin, P. B., & Robertson, S. (2015). Classification of team sport activities using a single wearable tracking device. Journal of Biomechanics, 48(15), 3975–3981. https://doi.org/10.1016/j.jbiomech.2015.09.015 Yoon, T.-L. (2017). Validity and Reliability of an Inertial Measurement Unit-Based 3D Angular Measurement of Shoulder Joint Motion. The Journal of Korean Physical Therapy, 29(3), 145– 151. https://doi.org/10.18857/jkpt.2017.29.3.145 Zanella, A., Bui, N., Castellani, A., Vangelista, L., & Zorzi, M. (2014). Internet of things for smart cities. IEEE Internet of Things Journal, 1(1), 22–32. https://doi.org/10.1109/JIOT.2014.2306328 61 ANNEXES 62