This section reviews research that has focussed on specific technical or physical aspects of play whilst including measures for the performance of the opponent. The aims and objectives of the studies reviewed in this section vary, however all are specific to football and all account for opponent interactions.
Gerisch and Reichelt (1993) conducted a computer and video based analysis of one- on-one situations (duels) between two opposing players participating in the first leg of a European cup semi final between FC Bayern Munich and Red Star Belgrade. Player’s and their opponent in the duel were identified using player numbers and it was revealed that in a total of 250 duels in the first game between the two sides, each team won 125 duels each. Although the number of duels won was even amongst both teams it was not divided equally amongst individual players. Some players playing in key positions for Munich (identified to be a central defender, two wide players and both strikers) were revealed to have low success in their duels, which were ultimately cited as the reason for why Red Star Belgrade created more chances and eventually won the tie.
Yamanaka et al (2002) performed a computerised notational analysis of three World cup 1998 games played between Japan, Argentina, Croatia and Jamaica with specific focus on the three fixtures in which Japan played. Playing patterns were analysed and technical actions such as dribbling, passing and shooting were compared amongst the four competing teams. In total 32 playing actions were categorised in to ‘time’, ‘place’, ‘player’ and ‘action’ and included in the analysis and performance data was entered after repeatedly viewing videotaped recordings of matches. The football pitch was divided into six areas horizontally and three areas vertically and the frequency of each play action per area was recorded. Yamanaka et al (2002) concluded that Japan did not utilise pitch spaces as well as Argentina, Croatia and Jamaica (contributing to their loss in each fixture against them) and that Japan should look to play more backward passes from the midfield area that enable them to retain possession and build attacks.
More recently, Lago and Martin (2007) conducted an investigation on the importance of possession (of the ball) using data on 170 matches played in the 2002-03 season in
the Spanish soccer league. In particular, four variables were examined, evolving match status, venue, the identity of the subject team and the identity of their opponent in each match. Nineteen dummy variables were created in order to identify the 20 teams in the league with Real Madrid the reference team (as they finished the season as champions). The results obtained from a linear regression analysis revealed that all four variables were statistically significant and together explained most of the variance in possession. Furthermore, the identity of the opponent was revealed to be of importance as the worse the opponent the greater the amount of possession.
Taylor et al (2008) examined the effects of match location, quality of the opponent and match status on the technical aspects of performance for a single professional British football team using data on forty matches played in the 2002-03 season. A computerised notational analysis system was used to notate the 40 match sample with 13 on the ball behaviours (play actions) and their corresponding outcomes (successful or unsuccessful). The quality of the opposition was dichotomised into two categories, “strong” or “weak” depending on whether the opponent finished in the top half (position 1-12) or bottom half (position 13-24) of the division upon conclusion of the season.
The independent and interactive effects of the three situation variables on the play action incidence were examined through log linear modelling. The results indicated that incidences of all technical play actions, with the exception of “set pieces” were influenced by at least one of the three situation variables with both independent and interactive effects discovered. The independent and interactive effects of the situation variables on play action outcomes were investigated through logit modelling, the results of which revealed that the situation variables had no influence on the outcome of play actions.
Lago (2009) examined the effect of match location, opponent quality and match status on possession strategies for a professional Spanish football team (RCD Espanyol). A computerised match analysis system was used to notate (post-event) 27 matches played throughout the 2005-06 season. Quality of the opponents was measured by the distance between end of season league rankings of competing teams. A linear regression analysis revealed that time spent in possession of the ball was greater when
ranking between competing teams increased or decreased the team’s possession by 0.2%.
Redwood-Brown et al (2012) analyse the impact of different standards of opponents on an observed team’s performance at a team level, positional unit level and individual player level. 29 Premier League matches were analysed during the 2010-11 season for 18 performance indicators. The standard of the opposing team was categorised as “top”, “middle” or “bottom” depending upon their final league position. The participating (observed) team was categorised in the “middle” category and 18 players from this team’s squad were selected for the purposes of the study. A one-way ANOVA20 analysis was conducted that assessed the observed team’s performance behaviour along with five positional units (centre-back, full-back, central midfield, wide midfield and centre forward) and individual player performance behaviour.
At team level, successful passes were revealed to be significantly higher against middle teams than top or bottom teams and interceptions were also revealed to be significantly higher against middle standard teams than top teams. In general the observed team performed better when playing against teams of similar calibre (other middle ranked teams) rather than top or bottom teams. Performance indicators at the positional unit and individual player level were also revealed to be sensitive to the standard of opponent. Furthermore Redwood-Brown et al (2012) state differences in individual player performances are not always evident at a positional unit or team level.
As can be seen above there are some studies that account for the quality of the opponent in football, however as stated by Tenga et al (2010) these studies tend to be limited by univariate data analyses or by small sample sizes, and are hence not representative of a population (Mackenzie and Cushion 2013). In a league consisting of twenty teams, a traditional season would consist of 380 fixtures. However, none of the studies reviewed above surpass 170 matches (study conducted by Lago and Martin 2007) and even this sample consisted of games played between teams across different Spanish leagues. Furthermore, although these studies have accounted for
20 Analysis of variance (ANOVA) tests are used to compare several samples (three or more) in terms of some numerical
opponent interactions, absolute measures of their ability (league position) have been utilised meaning the impact of the manager (in terms of specific match tactics and team selections) and relative team form have been ignored.
In various team sports it has been generally accepted that manager or head coach of a team has some influence upon overall performance (Audas et al 1999; Scully 1994; Hadley et al 2000; Dobson and Goddard 2011). At a team level, tactics and strategy could potentially decide between the winners and losers of a contest. At an individual player level coaching, training and motivating could all play a part in improving player ability. Section 3.6 presented below shifts the focus towards another factor of production in the form of a manager’s contribution.