Analysing the physical output of international field hockey players through the lens of phase of play: A Commentary

The development of automated player-tracking technology has been the catalyst for an explosion in the amount of research dedicated to understanding the physical demands of team sports. ¹ With the arrival of a whole new raft of ‘metrics’ to describe ‘physical performance’, one could be forgiven for thinking that the distance covered by a player in some way represented how well they played, and that teams completing more ‘work’ were more likely to win matches.

However, after that first wave of papers investigating the ‘demands’ of team sports rolled through, a more intelligent second wave of analysis followed whereby researchers looked at the relationship between physical output and more performance-based factors such as scoring, possession and style of play. After all, the objective in team sports is not to cover more ground than your opponent, but instead to score more goals.

As usual – and unsurprisingly – the first movers in this deeper level of performance analysis for team sports came from soccer.^2,3 It is welcome to see this type of analysis extend to hockey, and the paper by Cunniffe and colleagues is certainly one of the very first studies to investigate the relationship between locomotor activity and phase of play for this particular sport. There are some interesting findings, in particular how locomotor characteristics vary between positions depending on the phase of play. However, there are also some fundamental flaws in the way this study was designed and conducted, rendering the impact somewhat less than it otherwise might have – and could have – been.

The authors identified five discrete phases for their analysis, based on their existing in-house match analysis practices. No doubt this information is considered very useful for the coaching and support staff (and players) of the specific team evaluated here, but whether it has any particular relevance for the wider scientific and practitioner community is debatable. At the very least, it would have been logical to have reciprocal pairs for each phase, with one representing when the analysed team was in possession along with its corresponding partner for when the opposition were in possession. This is indeed the case for four of the five phases, such that Established Attack and Established Defence are paired, as are Attacking Counterattack and Opposition Counterattack. Meanwhile, Offensive Passing sits on its own. Although its counterpart – which perhaps might have been labelled Opposition Passing – was not considered of interest for the authors and their wider coaching staff, from a scientific perspective, it would have been informative to have included this phase. Indeed, having the equivalent phases for both the main team and the opponent would have allowed for an interesting comparison of how possession status impacts on a given phase of play, but the arbitrary and incomplete selection of phases did not allow for this.

Directly related to this point, the five phases adopted in this study collectively accounted for only 34% of playing time. In other words, two-thirds of match time was not accounted for in this analysis. Even though there may have been ‘no clear pattern’ in these ‘unstructured phases of play’, the fact that they accounted for somewhere between 30% and 78% of total match activity, depending on the locomotor category, suggests that this ‘phase’ cannot simply be ignored. At the very least, this 66% of playing time could have been allocated to a category labelled ‘unstructured’, ‘remainder’ or ‘miscellaneous’, and could have provided some very important information and context. For example, whilst differences in various locomotor categories were found between the chosen phases of play, it may well be the case that these values were all well below (or well above – we don’t know) the levels reported for unstructured play. The reader (whether coach, practitioner or researcher – or all three) may well have been able to take a guess, but no figures were provided, let alone analysed, for this substantial proportion of playing time. And in any case, ‘taking a guess’ does not belong in a scientific journal. There is no particular reason why this data – readily available based on the way data collection was described – could not have been at least reported, if not analysed, to allow the findings of this study to extend beyond the specific objectives of the team's coaching and support staff.

Any situation where unclassified activity represents almost double the amount of classified activity suggests that more thought should have been given to what happens in a hockey match. Although this may not have been of interest or utility to the coaching and support staff (although one wonders whether it should), this incomplete classification represents poor science. ⁴ Ideally, the unclassified segment should have represented a negligible component of the overall match activity. Better still, no activity would be left unclassified due to a comprehensive list of phases that account for all the various activities occurring in a hockey match.

In trying to add context to the locomotor activity recorded in a hockey match, the authors have not provided sufficient context as to how the chosen phases of play sit within the overall match demands. This information would have been easy to include in the analysis. Its absence reduces the relevance and impact of an otherwise notable study.