This study aimed to compare four standard deviational ellipse models to assess directional behavior and player activity area in four small-sided games (SSG) of soccer played on pitches with the same width (40 m) and different lengths (30, 40, 50, and 60 m). Fourteen participants played four 7-a-side SSGs on each of the four pitch sizes. Based on GPS data, four ellipse models were calculated for each outfield player and pitch size: major ranges (MR) measuring standard deviation in fixed length and width directions, linear regression assuming length (LR_LvsW) or width (LR_WvsL) as the independent variable, and principal component analysis (PCA) assuming both length and width as independent variables. Slope, area, semi-major and semi-minor axes, and eccentricity were calculated for each ellipse model. The PCA and LR_LvsW models showed similar and valid results for each variable, especially for larger pitch sizes. LR_WvsL showed unreliable results. The length axis should be considered as the independent axis when assessing the main direction of players’ movements and playing area through a standard deviational ellipse in soccer. This methodology could also be applied to evaluate a team’s labor distribution and spatial distribution of its players.

本研究旨在比较四种标准差椭圆模型，以评估在相同宽度 (40 m) 和不同长度 (30, 40, 50,和 60 m）。14 名参与者在四种球场尺寸中的每一种上进行了四场 7 人制的 SSG。基于 GPS 数据，为每个外场球员和球场大小计算了四个椭圆模型：主要范围 (MR) 测量固定长度和宽度方向的标准偏差，线性回归假设长度 (LR _LvsW ) 或宽度 (LR _WvsL)) 作为自变量，主成分分析 (PCA) 假设长度和宽度都作为自变量。计算每个椭圆模型的斜率、面积、半长轴和半短轴以及偏心率。PCA 和 LR _LvsW模型对每个变量都显示出相似且有效的结果，尤其是对于较大的节距尺寸。LR _WvsL显示出不可靠的结果。在足球比赛中，通过标准差椭圆评估球员运动的主要方向和比赛区域时，应将长度轴视为独立轴。这种方法也可以用于评估球队的劳动力分布和球员的空间分布。