Our hypotheses were that there would be significant differences between the phases of play and that these differences would be dependent upon playing position. The results demonstrated that the physical demands were greater in defence for the backs, and in offence for the forwards, whereas midfielders performed greater distance and high-speed running in offence, without significant differences in accelerations and decelerations between the phases. In-line with our second hypothesis, measures of distance and high-speed running were greater in the larger field locations, however, no specific pattern was noted for accelerations and decelerations. Our final hypothesis was that successful plays would be performed at a greater intensity than unsuccessful plays. This was the case during successful offence for both midfielders (distance and high-speed distance) and forwards (all metrics). However, measures of distance and high-speed distance were greater during unsuccessful defence for all positional groups.
Comparisons between phases of play highlighted the prevalence of significant differences that varied depending upon playing position. Backs displayed higher outputs in defensive phases, whilst forwards displayed higher outputs in offensive phases for all measured metrics, which is indicative of their positional role. However, despite midfielders performing significantly more relative distance and high-speed running in offence, they recorded no significant differences in acceleration and deceleration efforts. Additionally, with the exception of distance amongst the backs, midfielders recorded smaller effect size comparisons than the other two playing positions, which may indicate their physical output is more evenly distributed between offence and defence, in line with previous findings and accordance with their positional role . Furthermore, the effect sizes reported between offence and defence in this study (moderate to large) are greater than those reported in the study by Rennie et al.  (trivial to small), potentially owing to the delineation of physical output into positional groups within this study, or the differences in coding that exist within the literature regarding contested plays [8, 10, 12, 24].
Despite the differences in coding, contested plays were performed at a lower metres per minute than offence and defence, in line with findings from previous research . This finding is unsurprising and owed to contested plays being located at stoppages, where players are required to jostle or wrestle for possession of the ball, thus reducing the requirement to perform locomotion. This is highlighted in previous research using spatiotemporal data, where the density of players was greatest during contested phases of play . An interesting finding within this study relates to acceleration and deceleration efforts relative to playing time amongst the midfielders during contested phases of play, where there were no significant differences observed in comparison to offence and defence. This is likely attributed to the role of midfielders at stoppages who are centred around the ball, thus increasing their requirement to accelerate and decelerate, which is not the case for backs and forwards . This is supported by Rennie et al.  who reported that acceleration and deceleration load is higher during contested play, with the findings of our study indicating that this demand is greatest upon the midfield playing group. It is important for coaches to be cognizant of these differences that exist between the three phases of play, as training is often prescribed with the intention of practicing specific elements of a game (e.g., stoppages), and also within position groups (e.g., line training) . Therefore, having a greater understanding of the physical demands of each phase of play, per playing position, can ensure the appropriate training intensity is matched to specific training design.
When studying successful versus unsuccessful offensive phases, midfielders (distance and high-speed running) and forwards (all metrics) were the only groups to recorded significantly greater outputs during successful offensive plays, whilst backs generally recorded greater values during unsuccessful play. The findings for midfielders and forwards are in line with those reported within rugby union, where relative high-speed running was greater during successful attacking 22 entries . This finding may potentially indicate that successful play relies upon fast ball movement, where players are also required to move at speed to either spread (to create an opportunity for an effective disposal) or to carry the ball . This is somewhat supported by Lane et al.  who suggest that slow ball movement leads to greater congestion and lower scoring. This may have important implications for representative training, where drills aimed to improve a team’s offensive play, such as ball movement drills, should replicate the intensities derived from successful match performance in order to promote positive transfer to competition [27, 28, 30]. Additionally, increasing a player’s physical capacity in order to match these demands may also prove beneficial, this is particularly pertinent amongst the midfield and forward positional groups.
Similar findings were evident for technical actions during offensive phases, where backs performed kicks, handballs, and marks at a greater rate during unsuccessful plays, indicating that their impact on successful offensive performance is somewhat limited. Conversely, midfielders (kicks and marks) and forwards (kicks, handballs, and marks) performed technical actions at a greater rate during successful plays, highlighting the need to combine skill execution and enhanced running performance to achieve superior offensive outcomes. This combined approach to training appears to be of high importance, particularly when previous research has shown that kicking accuracy is reduced when kicking to a marked target as well as when the kicker has reduced time in possession, and is under increased opposition pressure . However, it has also been reported that AF players are underexposed to these constraints during representative practice . Therefore, there is a need to create training environments where skills (e.g., kicks) are performed under match conditions, which is likely to improve perception action coupling as well as player decision making, and lead to greater transfer to competitive matches [30, 32]. Interestingly, midfielders performed handballs at a higher rate during unsuccessful plays, potentially indicating greater importance of kicking to successful outcomes. The importance of kicking performance to successful match outcomes has been previously demonstrated, where it has been reported that team kick (and goal conversion) values were the two biggest contributors to successful match outcome . This may indicate that handballing offers a greater chance for the opposition to regain possession of the ball or force a stoppage, as a player can be tackled upon receiving a handball. This is not the case following a kick that is secured via a mark, where the receiving player is afforded a short period of time to perform a secondary kick, either to transfer the ball to a teammate or to take a shot at goal, which is unimpeded by opposition players. Previous research lends some support to this theory, where the frequency of handballs performed under physical pressure (3.1 ± 1.7) was greater than that of kicks (1.19 ± 0.83) . This information could be used to benchmark player performance, where a desired kick: handball ratio (number of kicks relative to number of handballs) may be targeted by coaching staff. The value of the kick-to-handball ratio has been highlighted by Robertson et al. , where winning teams demonstrated a higher kick-to-handball ratio compared to losing teams. However, it should be noted that this finding may be specific to the style of play of the study team, and may be relevant to the effectiveness of disposals (i.e. accurately reaching the intended target) and therefore not generalisable to the wider AF population where teams may have differing styles of play, as demonstrated by previous research .
Comparisons of successful and unsuccessful defensive phases highlighted that measures of distance and high-speed running were greater for all positions in unsuccessful phases. It is possible that during unsuccessful defensive play, opposition ball movement may be quicker, increasing the need for the defensive team to chase the ball and opposition . This may be particularly evident during turn-over, where the team now defending is likely to be caught out of position. These occurrences are potentially heightened during unsuccessful play, as score from turnover has been previously identified as a contributing factor to match outcome . Furthermore, Vella et al.  noted that relative high-speed running distances were greatest when defensive phases began with an intercept, adding further evidence to this theory. Conversely, acceleration and deceleration efforts were greater during successful defensive plays for midfielders and forwards. Although this may indicate the importance of accelerating and decelerating to perform successfully in defence for these positional groups, it should be noted that it is difficult to ascertain if these measures are an indication of an athlete changing direction or performing a tackle and collision . This is particularly relevant when tackles were performed at a greater rate during successful defensive plays within these positional groups. Additionally, marks and tackles were performed at a greater rate during successful plays for all positional groups, highlighting that the completion of these actions likely contributes more to successful play than physical output. Furthermore, the completion of tackles appears to be especially important for midfielders, where the effect size was calculated to be large. Therefore, it appears prudent that coaches afford dedicated training time to tackling and marking in defensive scenarios. As previously mentioned, these should be performed under match conditions (e.g., intensity), in order to facilitate positive transfer to performance.
Comparisons between field location demonstrated that relative measures of distance and high-speed running were greatest for both offence and defence in the largest field locations (full ground and attacking and defensive midfield), and lowest in the smallest field locations (defensive and forward-50), with exception of relative distance during defence where measures were greater in the forward-50 than midfield. This larger area potentially affords athletes with less congestion (i.e., number of players in proximity), as well as a larger distance to accelerate to higher velocity running, allowing them to produce superior relative running performance and greater velocities [16, 36]. Therefore, coaches should be cognizant to these outputs when devising and monitoring training drills in order to adequately meet these physical demands, where drills played on a full ground with reduced numbers are likely to elicit higher relative distances and high-speed running to those played in smaller areas. This is supported by previous research which demonstrated that the implementation of small sided games on larger pitch areas leads to greater distance and high-speed running performed by AF players .
Although it is expected that acceleration and deceleration efforts would increase in smaller field locations, this was not always the case, where there appeared to be no specific pattern demonstrated. There was evidence of both larger (e.g., attacking, and defensive midfield) and smaller (e.g., defensive and forward-50) pitch locations showing both comparably higher and lower measures for these metrics. However, previous research in soccer has demonstrated that small-sided games played on medium and larger pitches showed greater acceleration demand than those played on small pitches, although it should be noted that the medium sized pitch demonstrated the highest demand . Another study in soccer populations also support this, where the number of high accelerations and decelerations was similar (p > 0.05) between drill sizes . Combined, this evidence may suggest that if coaches wish to expose athletes in training to similar acceleration and deceleration efforts to those experienced during a game, area size may not be of primary concern.
Finally, there were few differences in physical output during contested phases of play, which is unsurprising considering these phases represent a time where the ball is somewhat locked into a contest during a stoppage. These were greatest in the attacking-midfield location; however, this finding is somewhat difficult to explain and could be potentially owed to the effort of the attacking team attempting to force the ball into the forward-50 location, and therefore closer to goal. Additionally, it should be noted that the effect size comparisons were only trivial to small.