Setting the scene: A cohort of professional football players participated in an intensive 6-day off-season player training camp, held in Portugal, June 2024. Players underwent a periodised microcycle delivered by Traainer performance support staff and supported by Sportable technological solutions. Prior to arriving on camp, players had accumulated two weeks of general physical preparation. Following baseline diagnostics conducted on day 1 of the camp, the subsequent 5 days comprised a specific individual preparation period. Players then returned to their professional clubs to commence pre-season preparations, accompanied by their player load reports. Traainer and Sportable then conducted an integrated analysis of the player load across the training week.
Rationale: The challenge facing performance practitioners and coaches in football is that whilst external load data (e.g. derived from global positioning systems) provide a global measure of physical workload, it does not directly link to what matters for player or team performance. This point is made emphatically in the below example (Figure 1), illustrating the distance metric outputs from the 2022 FIFA World Cup. The two best national teams and tournament finalists, France and Argentina, were ranked at the bottom end of the distance charts (1). Tracking data measures how much physical workload players perform, but not how effective that effort was, nor what level of movement the opposition were demanding of the team. This shortcoming has become increasingly recognized by applied sport scientists with an emerging body of scientific literature (2).
Figure1: Team total distance versus high intensity (>20kph) distance in the FIFA2022 World Cup. Extracted from Bradley, P. (2023) ‘Setting the Benchmark’ Part2: Contextualising the Physical Demands of Teams in the FIFA World Cup Qatar2022. Biology of Sport. 41. 271–278. DOI:10.5114/biolsport.2024.131091.
Players perform ‘football actions’ to build-up to attacking positions, which in turn create chances to shoot and to score. Conversely, in the defensive reciprocal, players perform actions to disrupt opposition build-up play and to delay or deny goal-scoring opportunities. The physical movement is simply the means to achieving those actions. Table 1 below lists a selection of ‘football actions’ that players perform for their team. This football, not sport science, specific list highlights that the common methods of measuring physical workload are decontextualised from football’s fundamental requirements. For example, when metrics such as high speed running distance are disconnected from concomitant ‘football actions’, the interpretation and application of this data may be misconstrued, devoid of meaning for the coach and offer a misrepresentation of training or match demands (see Verheijen (3)). Thus, it can be argued that assessing players’ physical outputs in isolation constitutes a partial, one dimensional approach, removed from the same technical and tactical events regulating these outputs. Alternatively, an integrated approach unifies the actions with the outputs and encourages a holistic, interdisciplinary approach to describing player load.
Table1 – list of varied football actions players perform as attacking, transitioning and defending tasks. This list is not intended to be complete.
The aforementioned explains why France and Argentina can have lower end physical outputs, yet still remain the highest performing teams. The total team distance measured is by and large a consequence of the playing style the team is adopting and the extent to which they are able to maintain 'tempo’, as well as the influence of opposition players and tactics. The total distance values are not the cause of performance, but rather a function of the football actions executed by the team. For a given playing style or game model, practitioners can learn what a ‘good’ set of physical numbers looks like and therefore, interpret performance indirectly drawing upon their specialist expertise. However, team and player performance evaluations would be easier and more reliable if an objective and more complete method were available.
Bradley et al. (4) proposed an answer to this problem with an integrated tactical and physical approach, providing a convincing rationale and substantiated by impactful examples of how to apply it. As the authors describe it, moving from‘blind’ quantification to contextualised physical data tells the coach why the player performed their various game runs, as well as what running the player did. The why combined with the what enables teams to evaluate if the players’ physical workload was spent on the correct footballing actions. In other words, were they performing their role as the tactical plan or game model required? For example, coaches may see players ‘hitting their numbers’ but know subjectively the high intensity running is not occurring during phases of play where it is most needed for their role in the team. With contextualized metrics, this can be evaluated correctly. The other benefit is to understand if training is providing the appropriate action specific stimulus. For example, JackAde’s work has shown integrated analysis performed over a series of matches can inform training session design, such as positional specific fitness drills, with appropriate technical actions as well as distance and speeds (5).
This is good step forward, but most of the actions categorised by Bradley et al lend themselves best to match play or match scenario drills. However, much of Football training differs to matches, with variety in the scale of movement and density of actions. Training sessions contain technical games and drills with the purpose of overloading skills and decision making, as much as executing effective attacks, transitions or defending. These drills, such as Small sided games(SSG), Possession drills and Rondos differ significantly from match play in density and dimensions and comprise a significant proportion of training time, both within session and across the week. Therefore, it would be beneficial for coaches and practitioners to understand more than the ‘blind’ what has physically happened from training, as well as matches. To analyse training with the addition of context along the lines of the match examples above, then a complete set of football actions in needed. This would need to include individual on-the-ball actions, such as passing, receiving, crossing, shooting, as these are the technical actions often performed during technical training.
Objective: This exploratory paper aims to demonstrate the benefits of integrating technical with physical data, as it expands the quantification of what the players have done in training, and creates insight into how different training drills and games produce technical and physical load outputs. We do this with illustrative data taken from anascent Smart Ball tracking system, alongside more commonly utilised GPS metrics. The Smart Ball system can identify player on-the-ballevents in real time, during training drills or match play. The data was captured from days two to six of a pre- pre-season training camp, where three professional men’s players conducted football specific technical and fitness drills with professional coaches and practitioners. This camp took place in June 2024, 10 days prior to the players start of team pre-season training. Day 1 of the camp was light and less structured, to acclimatize the players after a recovery period, so analysis of days 2 to 6 inclusive are presented here. The following details 1) the physical workload outputs across the camp, then 2) combines technical and physical load to build a more complete quantification and illustrates how to clearly visualise the two dimensions side by side, and finally 3) we create an example application of this integrated analysis to build a more informative picture of a training drills library.
Table 2 below describes the workloads of each day of the training camp, broken down by drill. Workload is expressed using 4 commonly used metrics in team sports. Taken together these numbers describe the volume and intensity performed by players on each day of the training camp. For example, day 4 is clearly the lightest day across all the drills and in total, whereas day 5 built up intensity and volume across the 3 drills as the session progressed.
Table 2: Duration (hh:mm:ss), Distance (m), Distance per minute (m /min), High Speed Running Distance (m) and Acceleration Count (#) of each training Drill across training Days.
The drill names will make sense to Football specialists, who will have an instinct for the football actions performed by a small group of players in these kinds of drill. We suggest the physical outputs shown in each drill are ‘as expected’, given the activity name. During the camp, the data was used to check the players had covered an appropriate level of running, helping to plan the next day’s session. However, the ‘GPS numbers’ do not inform the players’ level of involvements, nor the skill demand placed upon the players. Coaches or analysts would currently have to review the video and record involvements and skills performed by players to obtain this objective data set.
Table 3 adds the description of the football actions alongside the physical workload data. We are calling this the Technical Load™. For this exploratory analysis, we have represented Technical Load as a combination of the total actions the players made in each drill, as a measure of volume, plus the actions per minute as a measure of density, plus the average ball speed during the drill, as an indicator of intensity. Expressed in football specific language, then Technical Load could be seen as an objective representation of the volume and tempo of football actions (3).
Table 3: Duration (hh:mm:ss), Action (#), Action per minute (#/min), Average Ball Speed (mph), Distance (m), Distance per minute (m / min), and Acceleration Count (#) of each training Drill across days.
For example, the ‘Activation with Ball’ drills at the beginning of days 2, 4 and 5 are performed at higher tempo, but lower levels of running than most other drills across the training camp. Interestingly, the ball speed was significantly higher on day 4’s ‘Activation with Ball’ drill, pushing the tempo up in comparison to days 2 and 5. In contrast the ‘Pass and Move’ drill that kicked off day 3, was relatively high tempo with moderate levels of running.
This analysis is more specific to football with the addition of the Technical Load metrics and coaches and practitioners can evaluate more completely the demands of each drill upon players. The physical numbers quantify the (running) movement required to perform the football actions. The technical numbers quantify the volume and tempo of the football actions. For example, the finishing drills on days 4 versus 6 generated similar physical workloads, but day 4 the volume, density and intensity of the drill was higher, making it a higher tempo demand on the players.
In the camp, practitioners assessed if the players were running at planned levels AND whether they also received an appropriate level of skill stimulus from each drill. This informed coaching cues and organisation of the next day’s session activity. The left hand box of the chart in figure 2 shows the breakdown of the number of football actions performed by one player during each drill on days 3 and 4 of the camp. The right hand box shows the corresponding physical workload required to perform the football actions of drill. We have expressed this required workload as mechanical work, as a valid method to quantify workload proposed by Gray et al (6) and furthered by Vassallo et al (7). It places work (in Joules) as a unifying variable representing the physiological cost to the player from the mechanical running performed. The Severe Domain Work portion quantifies running performed above critical power, which could be from accelerating, decelerating or high speed running. We have used this to visualise a single descriptor of the physical cost of the running ‘work done’.
Figure 2. Days 3 and 4 Drill Actions (#) by on the ball event, with Average Ball Speed (mph) of each action, with physical workload, expressed as mechanical work (kJ). Data shown for one player only.
This chart is intended as an exploratory attempt to show the integrated technical and physical demands of training session drills for coaches and practitioners. Hopefully it provides a clear picture of the stimulus of each drill and how they compared to each other. The football actions describe the drill technically, and the work value quantifies the resulting physiological demand. In this example, certain drills deliver higher levels of football actions, but very little workload and with no intensity. The first 4 drills on day 4 would be examples of this, as passing action dominant drills, with lower levels of work. Other drills can deliver both a high tempo of football actions and physical load with intensity, such Day 3’s ‘Conditioning-with Ball’, and others comprise higher levels of technical and physical load, but limited high intensity, such as day 4’s ‘Finishing Drill’.
Figure 2 above is an attempt to show that a combined analysis of technical and physical load provides a more complete and informative description of football training drills. We have also suggested a method to visualise the football actions and the workload in a single plot to give practitioners aclear picture of how players have performed their running workloads from the football actions. Another potential application of this analysis is to conduct comparisons of the training drills using the two dimensions. Figure 3 below is example of how drill type scan be objectively described using the technical data in conjunction with the usual ‘GPS numbers’, and it plots the Physical against Technical Load, as a normalized STEN value, intuitively showing the relative demands.
Figure 3. Physical Load Index v Technical LoadIndex of drill types performed in the Training Camp.
Briefly, we converted the three Technical Load™ metrics with the three Physical Load metrics in table 3 into STEN scores, and then aggregated these scores into separate and Technical Load and Physical Load Index scores for drills of the same type. E.g. the STEN scores of all the passing type drills were combined and averaged to produce the ‘Passing Drill’ Index scores depicted. Performance practitioners could use this type of analysis to inform session planning, helping coaches choose the optimal mix of drills to achieve both the technical and physical objectives of the session. This helps promote an interdisciplinary approach to session design. Typically, drills are only described by abstract physical workload numbers, which could lead to non-football specific information determining session content. This method helps both coaches and sport scientists achieve the optimal outcomes from sessions and should help teams build more a meaningful ‘Drills Library’.
Conclusion
Objective training data is typically limited to physical workload, giving a ‘blind’ description of what running happened. The addition of Technical Load™ data from Smart Ball technology brings the quantification of football actions alongside the ‘GPS numbers’. This means coaches can evaluate training using their own football language more clearly, answering questions such as what was the tempe quality of the crossing and finishing drill today, versus last week? It also means practitioners can build a complete and objective understanding of commonly used drills, to optimise both the technical and physical stimulus.
REFERENCES:
1) FIFA 2022 World Cup Distance Teams Covered Data
2) Torres-Randaet. al. (2022)
3) Verheijen (2014, Football Periodisation
4) Bradley et. al. (2021), Football Analytics, Chapter 9
5) Ade (2019) PhD Thesis
6) Gray et. al. (2018)
7) Vassallo et. al. (2021)
