The full stack - recognition, analysis, and visualization - explained, start to finish. See how court lines, players and the ball are automatically recognized; how shot-by-shot data like speed, spin and angle is extracted; how that data is queried, filtered and analyzed; and how insights are displayed in the 3D engine.

Do players adjust their serve placement when more is on the line? This analysis explores how serve patterns shift depending on leverage — a measure of the importance of the point being played. The player-specific behaviours revealed in this video highlight the crucial role of data-assisted match preparation. After all, trying to recognize these patterns during the match is costly and difficult, since high-leverage points are important, but occur too infrequently to allow the player to adapt.

By modelling the expected probability to win the point, this case study assesses the effectiveness of the chipped return versus the topspin return. While the chip return may appear inferior at first glance, deeper statistical analysis reveals (somewhat surprising) patterns of play that can make the chip a viable alternative to the topspin return, in certain scenarios.