Evaluating strategic periodisation in team sport

Abstract

The planned peaking for matches or events of perceived greatest priority or difficulty throughout a competitive season is commonplace in high-level team sports. Despite this prevalence in the field, little research exists on the practice. This study aimed to provide a framework for strategic periodisation which team sport organisations can use to evaluate the efficacy of such plans. Data relating to factors potentially influencing the difficulty of matches were obtained for games played in the 2014 Australian Football League season. These included the match location, opposition rank, between-match break and team "form". Binary logistic regression models were developed to determine the level of association between these factors and match outcome (win/loss). Models were constructed using "fixed" factors available to clubs prior to commencement of the season, and then also "dynamic" factors obtained at monthly intervals throughout the in-season period. The influence of playing away from home on match difficulty became stronger as the season progressed, whilst the opposition rank from the preceding season was the strongest indicator of difficulty across all models. The approaches demonstrated in this paper can be used practically to evaluate both the long- and short-term efficacy of strategic periodisation plans in team sports as well as inform and influence coach programming.

Kinematic and Electromyographic Activity Changes during Back Squat with Submaximal and Maximal Loading

“…For each trial, required portions of the video recordings were trimmed, the anthropometric points were digitized, and 2D positional data were obtained. To analyze angular kinematics of the knee and hip joints, the raw data points were calculated and then smoothed using a moving average filter [29]. All EMG data were partitioned into ascending and descending phases.…”

Multifactorial analysis of factors influencing elite australian football match outcomes: a machine learning approach

“…In elite Australian football (AF) many research studies have investigated the impact that various factors may have on the likelihood of winning any given game. These factors include fixture schedules (Gastin, Fahrner, Meyer, Robinson, & Cook, 2013;Lazarus, Hopkins, Stewart, & Aughey, 2018), opposition team strength (Robertson & Joyce, 2018), current team form (Robertson & Joyce, 2018), players' age (Gastin et al, 2013;Lazarus et al, 2018) and experience (Gastin et al, 2013;Mooney et al, 2011;Piggott, McGuigan, & Newton, 2015;Robertson & Joyce, 2018), plus many others (Lazarus et al, 2018). However, few studies have assessed combinations of factors in multivariable analyses; therefore, the impact of, and interaction between various factors remains unclear.…”

“…Recently, attempts to model "fixed" and "dynamic" pre-match factors that influence game performance across multiple team sports (Robertson & Joyce, 2015 have been made. Specifically, in elite AF, "match difficulty indexes" were created to quantify periods of the 2014 season when winning may have been more difficult (Robertson & Joyce, 2018). In these models, factors such as opposition rank (previous year), match location (home/away) and number of days break between games were considered "fixed", as they were known prior to the start of the competitive season.…”

“…Although Lazarus et al (2018) reported on a large data set (5109 AFL games), the number of variables included was only small (6 features) and focused mainly on absolute data with limited consideration for the differential between the teams. While this may conform with Occam's razor principle (Coutts, 2014), where the demand for more variables must be weighed against the ability to collect and report on them, these authors did not include some important variables (such as team quality) that are known to relate to game outcome (Robertson & Joyce, 2018). Without the inclusion of these fundamental variables, it is difficult to have a full understanding of the true impact that the variables studied by Lazarus et al (2018) may have had.…”

“…Prior research attempts to model pre-match factors affecting match outcomes have generally been limited to long periods of time with a small number of features (Lazarus et al, 2018) or a reduced time period and larger number of features (Robertson & Joyce, 2018); however, a consistent theme is that statistical approaches have often been confined to inference-based linear approaches (Robertson, Back, & Bartlett, 2016). However, with greater access to complex data sets, machine learning (ML) and specifically supervised learning analysis is increasing in use, with the potential benefits over traditional modeling approaches being suggested in specific instances (Robertson, Back, & Bartlett, 2016).…”