Using network science to analyze tennis stroke patterns

This research aimed to describe and quantify the stroke patterns profiling professional tennis players by con-structing a complex network specifically related to ball bounce locations. A total sample of 16,863 points from 127 Australian Open matches played by 128 male tennis single players were gathered and scrutinized for creating the bipartite tennis stroke network (TSN) with court zones divided into 40 nodes. Afterwards, network metrics were used to assess the prominence of different court zones: in/out-degree centrality (I/ODC), eigen-vector centrality (EC), betweenness centrality (BC) and shortest-path length. The results showed that zone-5 (zone-36) and zone-8 (zone-33) generally had higher levels of zone utilization considering their I/ODC, EC and BC. Zone-8 (zone-33) to zone-33 (zone-8) turned out to have the least short path length with a value of 0.0008 in all paths which had the most path utilization. Moreover, different rally lengths and tournament rounds can lead to a series of stroke patterns. As the tournament rounds progressed from the 1st round to the final, the consistency in ball bounces of the zones tended to decrease during short to medium or long rallies. Future in-vestigations and sport performance analysts could adopt the TSN method to analyze tactics and specific striking styles of individual players.

Automated summary

This research aimed to study the stroke patterns of professional tennis players by constructing a complex network based on ball bounce locations. Data from 127 Australian Open matches played by 128 male tennis players were analyzed to create the tennis stroke network. The results showed that certain court zones were more frequently used and had shorter path lengths. The study also found that rally lengths and tournament rounds influenced stroke patterns.