Data analytics in performance of kick-out distribution and effectiveness in senior championship football in Ireland

2.1Participants

Due to the nature of the data collection in this primarily quantitative research study, participants were not required for the notational analysis aspect of the study. Video footage of the eight matches (Table 1) in the Ulster Men’s Senior Inter-County Championship involving nine teams was analysed. In total, 288 players (age 25.6±3.9 years, height 181.5±5.4 cm, weight 80.9±10.3 kg) actions were tracked during the analysis, which included substitutions. Player profiles (Appendix 2) of every squad member were retrieved from the Ulster GAA Council. A questionnaire (Appendix 3) was designed and completed by 100 participants, all of whom were a selection of Gaelic Football players (75) and coaches (25) from Ulster. The participants surveyed varied from presently involved players/coaches to retired players/coaches, and from those with club level to inter-county level experience.

2.2Data definition

To define the variables used for video tagging, experts in the field of Gaelic football were surveyed via email to clarify a definition of each key element (Appendix 1). The precision of accurately defining the key elements was essential for repeatability of the test. The same operator notated each game, and therefore eliminated the possibility of operator judgement cross-over.

2.3Data collection

All of the video footage was received directly from the Ulster GAA Council. Statistical data was obtained through the video analysis software Dartfish (Connect Plus 5.5) which, along with training in its use, was provided by the Sports Institute of Northern Ireland (SINI). SINI were invested in provision of such support because of the lack of Gaelic Games research at the time. A tagging panel specifically designed for the study was produced in conjunction with the Lead Performance Analyst in SINI. The role of this Lead Analyst was specifically outlined from the outset of the study –the role was agreed as assisting in making the decisions on what to analyse, and he acted as a support if anything was needed during the process; he reviewed the findings, and being satisfied with the results, did not suggest any changes. The tagging panel was piloted on two matches prior to initiation of the study’s notation. The tagging process used in this study is displayed in Appendix 4.

The questionnaire was developed through www.surveymonkey.com and was distributed via email to key stakeholders within the context of the study. The purpose of the questionnaire (which contained both open ended and closed questions) was to gather opinions and attitudes on notational analysis, using quantitative and qualitative analysis, to add further depth to the statistical evidence. Contact via email with the various GAA County Boards across Ulster was made in order to contact to the people in question. The survey was anonymous, with confidentiality guaranteed and had a 100% response rate. The study was approved by the University Ethics Committee prior to data collection.

2.4Data analysis

The computerized notation system (Dartfish) used enabled the development of a large database of raw data, which was manually analysed to highlight any significant patterns, trends and differences in the data and to specifically answer the research questions of the study. Data obtained from the questionnaire was analysed (manually and computerized) to extract consistent themes from a range of the written and verbal communication, and involved a process proposed by Miles and Huberman (1994) (Appendix 5). The website used to create the questionnaire automatically analysed the questionnaire data by calculating both the response count and response percentage of each answer.

3.1Kick-out distribution

Compiled results included analysis of kick-outs (KO’s) from all teams from eight matches, thus amassing 351 kick-outs. The mean number of kick-outs per match was 44 (±9.9). The most common distance KO was 65m+ with 53.8% (±19.9), whilst 57.6% (±13.9) of KO’s were directed to the ‘wings’ (Table 2).

The average time taken to take a kick-out was 32.1 seconds (Table 3). In total, 3.5 hours (38%) of match time was taken up in anticipation of kick-outs. The top two ranking teams were found to have averaged the shortest kick-out times across all teams.

Approximately 44% of all kick-outs resulted in loss of possession with only a quarter of KO’s resulting in possession won cleanly (Fig. 1).

Fig.1

Percentage outcome of KO’s.

3.2Kick-out effectiveness

Classification of KO effectiveness was based on three areas: Possession Regain, Attack Building, and Scores/Scoring Opportunity (Tables 4 and 5). In terms of direction, a greater rate of possession was regained from KO’s to the wings (60.4%), with 54% of kick-outs to the centre won. The 21–45 m KO (87.1%) was the most successful KO distance regarding retaining possession. Kick-outs to 21–45 m yielded the greatest percentage of attack building (54.8% ± 11), whilst in terms of direction, KO’s to the wings (34.5% ± 27.6) were considerably superior in attack building than KO’s to the centre (22.3% ± 11.4). Kick-outs to the wings (18%) contributed greater to creating scoring opportunities than centrally directed KO’s (10%).

Tables 4 and 5 present the net difference various kick-outs have on the scoring opportunities of both teams. It was highlighted that KO’s to both 21–45 m (+6%) and the wings (+5%) result in the greatest positive net difference for the KO team. Kick-outs to the centre were the only kick-out parameter that resulted in a negative net difference (–1%).

Based on completed questionnaires of 100 key stakeholders, Fig. 2 highlights that ‘off the ball’ runs (45%) were the most common, successful kick-out strategy employed over the past thirty years.

Fig.2

Most frequent successful kick-out strategies over the past 30 years.

4.1Distribution

In total, 351 kick-outs (KO’s) were analysed during the study, with an average of 44 KO’s per match; which is more than one KO every two minutes and is in adherence to Martin’s (2003) findings. Table 2 established that KO’s to 65m+ (53.8%) were considerably the most common distribution outcome. This suggests that teams either prefer to take a direct approach to score building, or view the kick-out primarily as a defensive parameter to clear the ball as far away from the goal as possible. The results correlate to the findings of Appleby and Dawson (2002) on Australian Rules who found that on average the favoured kick type for kick-ins’ was ‘long’ to 50m+ (47%). In coherence with Carolan’s (2005) findings, kick-outs to the wings (57.6%) were considerably the most common directional kick-out; although it must be noted that the ‘wings’ takes up two thirds of the field position.

Due to goalkeepers being in unpressurised control of the ball during kick-outs, it was surprising to report that in total, more KO’s were lost (59.3%) than retained (Fig. 1). An explanation of this may be due to the fact that the opposing team can ‘break’ the ball due to it not being in their half of the pitch. The KO team on the other hand can ill afford not to attempt the ‘clean catch’ as it will otherwise put the opposing team in a dangerous attacking pitch position.

Across the eight matches, barring instances of injuries or unforeseen stoppages, the average time taken to complete a KO was 32.1 s, whilst the ball was out of play 38% of match time in anticipation of KO’s. Table 3 displays a ranking system of the teams based on the speed of their average KO time. Interestingly the table ranks Tyrone, Monaghan and Donegal first, second and last respectively. Given the fact that Tyrone defeated Monaghan in the final and Donegal lost in a preliminary round, it would be easy to assume that a quicker KO time contributes to the overall success of a team. Although this would be a ill advised assumption as there are numerous other factors to be accounted for that attributes to this statistic. It is however likely that quicker kick-outs enable fast counter attacking football, and is advantageous in the sense that it reduces the time for opposition teams to regroup. A key limitation of Table 3 however is that both Tyrone and Monaghan played three matches whilst Donegal only played one, which may have attributed to their longer KO time.

4.2Effectiveness

To distinctively determine ‘effectiveness’ is extremely difficult in relation to this study as there are many individual factors to incorporate such as regaining possession, attack building and scoring opportunities. Indeed simply regaining possession may not always be advantageous to a team. For example, if a player regains possession deep in their own half but is immediately put under pressure from 2-3 opposition players, then it would be difficult to classify the KO as being effective. The lack of research reported in the literature pertaining to performance analysis on kick-outs, leaves it challenging to make any generalisations about its effectiveness. Therefore for the purpose of this study, the effectiveness of the KO’s was broken into three subsections; regaining possession, attack building and scoring opportunities.

The percentage of kick-outs retained decreased the further the distance kicked, which adhered to Carolan’s (2005) findings. A noteworthy difference of 34.4% exists between the percentage of possession regained from KO’s to 21–45 m and KO’s to 65m+, in favour of the shorter distance. Considering only 8.8% of KO’s were distributed to 21–45 m and 53.8% to 65m+ (Table 2), it appears illogical that Gaelic teams persist with the long ball strategy. A long kick will result in the ball travelling in the air for a greater time, allowing more defensive players to arrive at the site of contest (Appleby and Dawson, 2002), and thus therein lies a possible explanation of the lower success rate of longer KO’s.

In the modern game, a goalkeeper is required to be dynamic and involved in attack building. For this reason, the study analysed the percentage of KO’s directly leading to the start of an attack. Substantially, 21–45 m KO’s (55%) lead to an attack at almost twice the rate of the longer ranged KO’s, which is likely a result of the increased rate of retaining possession. The findings are consistent with Appleby and Dawson’s (2002) who found that kick-ins to 25–50 m had the greatest correlation with the ball entering the opponents half of the pitch.

In 1983, Franks et al. concluded that it would be advantageous for soccer players to keep sequential passing to≤3. Bates (1988) interestingly found that 94% of goals scored in varying levels of international soccer resulted from moves involving≤4 passes, enabling a conclusion that direct play is more beneficial than possession play. Albeit in a different code of football, the current study findings failed to detect any difference in benefit of direct play (long KO’s) and shorter KO’s on scoring outcome. However, kick-outs to the wings contributed an 8% greater success rate of creating scoring opportunities as opposed to central kick-outs. With eight player positions consuming central positions, and only four player positions per wing, it appears that the increased space contributes to the creation of scoring opportunities by enabling more room to deliver a pass to an attackingplayer.

Considering the majority of KO’s lost possession (59.3%), it was important to analyse the net difference between kick-out placement on scores/scoring opportunities for and against teams (Tables 4 and 5). Kick-outs to 21–45 m (+6%) had a greater net difference of scores for and against than any other distance and thus was the most effective strategy as regards overall scores. Of greater interest, KO’s to the centre resulted in more scores against than for (net difference –1%) whereas KO’s to the wings (+5%) proved to have the greatest net difference. This suggests that coaches would benefit the team by developing and incorporating kick-out strategies into training that will allow them to create more scoringopportunities.

4.3Team strategies

It is important to acknowledge that the norm does not indicate the most successful strategy. A breakdown for each team’s KO’s (Appendix 6, 7 and 8) was produced prior to the calculation of summative results in order to present evidence of this point. Tyrone, who eventually won the tournament for the second successive year, kicked 22.8% of their KO’s to the centre, which is approximately half of the average, and opted for the strategic placement to the wings instead to maximise retention of possession. It is essential to note that a ‘one size fits all’ approach does not exist, as successful teams develop strategies to suit their strengths (Keane et al., 1993). From studying anthropometric profiles, Tyrone’s central midfield players were on average the smallest and second lightest of all the teams in Ulster; this makes their short game approach understandable.

4.4Player and coach opinions on kick-outs

From the fifty stakeholders interviewed, the most common kick-out strategy appeared to be the use of off the ball runs to create space. Player movement patterns were not analysed in the current study, however off the ball runs can be used to influence the actual directional/placement of kick-outs. What is of interest to the present study, and which adheres to the findings, is that 39% (Fig. 2) of stakeholders stated that kick-outs to the wings was their most frequently successful approach, whilst a mere 14% used kick-outs to the centre as their most common strategy.