Tactical positioning in short-track speed skating: The utility of race-specific athlete-opponent interactions

Figures & data

Table 1. Point estimates (and 95% confidence intervals) of tactical positioning importance calculated without race-specific athlete-opponent interactions.

Lap	τb	95% CI
Start	0.16	0.14–0.17
1	0.21	0.19–0.23
2	0.21	0.20–0.23
3	0.28	0.26–0.29
4	0.36	0.35–0.37
5	0.45	0.43–0.46
6	0.55	0.54–0.56
7	0.70	0.69–0.71
8	0.86	0.85–0.86

Each point estimate represents the similarity between athletes’ intermediate and final rankings measured using Kendall’s Tau-b, τ_b.

Figure 1. Evaluating tactical positioning using the with race-specific athlete-opponent interactions approach. (a) 2D histogram of the similarity between start position/ intermediate rankings (Laps 1–8) and final rankings (Lap 9). The density represents the proportion of races in the dataset, and the white dot denotes each lap’s modal Kendall’s τ_b value. (b) 1D histogram of Lap 2’s similarity between intermediate and final rankings.

Figure 2. Tactical positioning sequences in a 1,000 m race with four athletes. Each node represents the similarity between start position/ intermediate rankings (Laps 1–8) and final rankings (Lap 9). Each edge represents how a node can transition to future and past laps. The highlighted nodes and edges illustrate an example sequence of (−0.67, −0.67, 0, 0, −0.33, −0.33, 0.33, 0.67, 1). This sequence is one of 40,353,607 possible tactical positioning sequences in this race scenario.

Figure 3. Effect of treating laps as discrete or sequential events. Comparison of with race-specific athlete-environment interactions discrete, point estimates, i.e. the modal similarity between start position/ intermediate rankings (Laps 1–8) and final rankings (Lap 9), with two observed tactical positioning sequences of the form $(Star{t}_{{\tau }_b},Lap{1}_{{\tau }_b},\dots ,Lap{8}_{{\tau }_b})$. (a) the most commonly occurring sequence (absolute support = 18 races; relative support = 1.2%), and (b) an observed tactical positioning sequence (absolute support = 4 races; relative support = 0.3%). In both plots, the shaded region represents Kendall’s τ_b values where the start position/ intermediate rankings and final rankings are deemed to have a strong similarity.