Analysis of external workload in soccer training and competition: generic versus individually determined speed thresholds

Dear Editor, we read with interest a recent article published in Science and Medicine in Football employing data mining techniques in an attempt to determine new time-motion analysis speed thresholds for elite women football players (Park et al. 2018). This article fits in with the continual need to hone monitoring techniques and aid understanding of external loads in contemporary training and match-play. The information can aid practitioners in manipulating physical output and monitoring responses to the stimulus to help players respond to playing demands whilst attempting to reduce the risk of incurring injury. Historically, external workload has been determined using high-speed and sprinting outputs generally represented by distances covered above generic or arbitrary player-independent speed thresholds (or zones) of 5.5 and 7 m/s, respectively. These thresholds have frequently been used in professional football especially since the introduction of semi-automated camera systems, and were universally adopted by the major contemporary commercial GPS and Optical tracking companies. As a result, they have found their way into the scientific literature and football industry despite a lack of scientific investigations providing an empirical technical, tactical and physiological grounding. As performance indicators, generic thresholds allowpractitioners to compare running outputs at an absolute level across teams, individual players, playing positions and standards using the same criteria. However, when a generic speed threshold of 5.5 m/s was used on a squad average basis to quantify the high-intensity running distance covered by elite players in competition, outputs were substantially underestimated in comparison to data adjusted according to individual speed thresholds derived from physiological testing (Abt and Lovell 2009). Similarly, while a threshold of 7 m/s is widely used to classify sprinting distance in elite professional football, peak speeds ranging between 8.2 and 9.7m/s have been reported across players (Rampinini et al. 2007). As such, sprinting distance can be substantially overestimated in training and match-play in some players. Recently, Colby et al. (2018) suggested that in order to reduce injury risk, athletes should be exposed to nearmaximal velocities on a regular basis. As a result, it would seemmore logical tomonitor running activity above 95%of each individual’s peak speed as opposed to a generic threshold. In our opinion, the article by Park and colleagues has employed a fresh approach to determining time-motion analysis speed thresholds via data mining techniques. These techniques can be used to group athlete velocity data and determine patterns within athlete movements, without the requirement of a human input threshold based on a physiologically defined or arbitrary value (Sweeting et al. 2017). Yet we ask whether high speed running and sprinting data derived using these techniques are sufficient to provide an accurate representation of the true loads elicited upon players especially if we are to account for inter-individual differences in physical characteristics? It is recognised that there can be substantial discrepancies in locomotor outputs if absolute data are not adjusted using individualised speed thresholds (Schimpchen et al. 2016), especially when thresholds are derived from values for peak sprinting speed and/or aerobic fitness (Abt and Lovell 2009; Lovell and Abt 2013; Hunter et al. 2015; Abbott et al. 2018a) and more recently, maximum accelerative capacity (Abbott et al. 2018b). In the absence of adjustments, identical external training loads could elicit considerably contrasting internal loads in players with different individual characteristics. Practitioners unable to administer a player specific approach to performance monitoring and training prescription might find the training stimulus appropriate for one athlete, but inappropriate (too high or too low) for another. Subsequently, players may be underprepared for the physical demands of the game or exposed to ‘spikes’ in external load potentially increasing the risk of them being pushed beyond their physical limits and eventually breaking down. Indeed, there are difficulties in definingwhich acute:chronic workload ratio values are critical when monitoring players with varying or unknown fitness levels (Buchheit 2016). While we acknowledge that a simplemeasure of aerobic fitness does not enable prediction of injury or performance, an easilyadministered field test to determine maximal aerobic speed as a speed threshold (despite its acknowledged limitations) could enable prescription of external loads tailored to each individual or if practically difficult, to small groups including players with similar values. A more tailored approach to training prescription could engender improvements in aerobic fitness thereby increasing athletes’ resilience to higher workloads through protectively moderating the workload effect by ‘dimming’ or reducing the risk of rapid workload increases (Windt et al. 2017). Similarly, if a player performs poorly in a pre-season fitness test or is returning to play following injury, practitioners could theoretically adjust his/her ‘permitted’workload threshold according to current fitness status, whilst providing personalised attention to address the deficiency (Windt et al. 2017). In linewith these points, external workloads are sometimes used as indicators of competitive performance and therefore running outputs of certain players may again be underor over-estimated. Some practitioners also attempt to make inferences from external match load data to post-match stress and readiness to play status and adjust training loads accordingly (Carling et al. 2018). Again, arbitrary speed thresholds might not truly depict players efforts possibly leading to errors in interpreting fatigue and readiness for participation in training and/or selection for competition. Accounting for individual fitness measures could also have pertinence when monitoring youth players moving across age categories and when changes in maturation status occur. While current practice commonly assesses academy players using the same generic speed thresholds as senior squad peers, it SCIENCE AND MEDICINE IN FOOTBALL 2019, VOL. 3, NO. 1, 83–84 https://doi.org/10.1080/24733938.2018.1562279