System and method for content and style predictions in sports

Patent metadata

Field	Value
Publication number	US-12175754-B2
Application number	US-202318154145-A
Country	US
Kind code	B2
Filing date	Jan 13, 2023
Priority date	May 8, 2019
Publication date	Dec 24, 2024
Grant date	Dec 24, 2024

Abstract

A system and method for generating a play prediction for a team is disclosed herein. A computing system retrieves trajectory data for a plurality of plays from a data store. The computing system generates a predictive model using a variational autoencoder and a neural network by generating one or more input data sets, learning, by the variational autoencoder, to generate a plurality of variants for each play of the plurality of plays, and learning, by the neural network, a team style corresponding to each play of the plurality of plays. The computing system receives trajectory data corresponding to a target play. The predictive model generates a likelihood of a target team executing the target play by determining a number of target variants that correspond to a target team identity of the target team.

First claim

What is claimed:

1. A method of generating a play prediction for a team, comprising:
receiving, by a processor of a computing system, match data for a match between a first team and a second team from a memory of the computing system;
extracting, by the processor of the computing system, from the match data, one or more portions of data corresponding to a target play of the first team of the match, wherein the one or more portions of data include tracking data, and wherein the tracking data includes trajectory data;
inputting, by the processor of the computing system, the tracking data into a variational autoencoder, wherein the variational autoencoder includes a generative model configured to receive the trajectory data and generate a plurality of variants of target play based on the trajectory data;
based on the inputting, receiving, by the processor of the computing system, the plurality of variants of the target play based on the tracking data from the variational autoencoder;
inputting, by the processor of the computing system, the target play and the plurality of variants into a neural network;
based on the inputting, receiving, by the processor of the computing system, a team identity and a playing style corresponding to each of the plurality of variants from the neural network;
generating, by the processor of the computing system, a graphical representation of the team identity and the playing style corresponding to each of the plurality of variants; and
outputting, by the processor of the computing system, the graphical representation via a display of the computing system.

2. The method of claim 1 , further comprising:
associating, by the processor of the computing system, each variant of the plurality of variants with at least one team of the first team or the second team.

3. The method of claim 1 , the method further comprising:
identifying, by the processor of the computing system, a content of the target play.

4. The method of claim 3 , wherein inputting, by the processor of the computing system, the tracking data into the variational autoencoder further comprises:
inputting, by the processor of the computing system, the content of the target play or the playing style into the variational autoencoder.

5. The method of claim 3 , wherein each variant of the plurality of variants maintains the content of the target play while changing the playing style exhibited by the target play.

6. The method of claim 1 , wherein generating, by the processor of the computing system, the graphical representation of the team identity and the playing style corresponding to each of the plurality of variants comprises:
generating, by the processor of the computing system, one or more graphical representation trajectories for each of the plurality of variants.

7. The method of claim 1 , further comprising:
predicting, by the processor of the computing system, a likelihood of a third team executing the target play by analyzing the plurality of variants with the neural network to determine a second number of target variants that exhibit a playing style similar to the third team.

8. A method of generating a play prediction for a plurality of teams comprising:
receiving, by a processor of a computing system, match data for a match between a first team and a second team from a memory of the computing system;
extracting, by the processor of the computing system, from the match data, one or more portions of data corresponding to a target play of the first team of the match, wherein the one or more portions of data include tracking data, and wherein the tracking data includes trajectory data;
inputting, by the processor of the computing system, the tracking data into a variational autoencoder, wherein the variational autoencoder includes a generative model configured to receive the trajectory data and generate a plurality of variants of target play based on the trajectory data;
based on the inputting, receiving, by the processor of the computing system, the plurality of variants of the target play based on the tracking data from the variational autoencoder;
inputting, by the processor of the computing system, the target play and the plurality of variants into a neural network;
based on the inputting, receiving, by the processor of the computing system, a team identity for a plurality of teams and a playing style corresponding to each of the plurality of variants from the neural network;
predicting, by the processor of the computing system, a likelihood of the second team executing the target play by:
mapping, by the neural network, each variant of the plurality of variants to a team of the plurality of teams based on the playing style exhibited in each variant,
determining a number of target variants corresponding to the playing style of each team of the plurality of teams, and
identifying the number of target variants corresponding to the playing style of the second team;

generating, by the processor of the computing system, a graphical representation of the number of target variants that exhibit the playing style similar to the second team; and
outputting, by the processor of the computing system, the graphical representation via a display of the computing system.

9. The method of claim 8 , further comprising:
associating, by the processor of the computing system, each variant of the plurality of variants with at least one team of the first team or the second team.

10. The method of claim 8 , the method further comprising:
identifying, by the processor of the computing system, a content of the target play.

11. The method of claim 10 , wherein inputting, by the processor of the computing system, the tracking data into the variational autoencoder further comprises:
inputting, by the processor of the computing system, the content of the target play or the playing style into the variational autoencoder.

12. The method of claim 10 , wherein each variant of the plurality of variants maintains the content of the target play while changing the playing style exhibited by the target play.

13. The method of claim 8 , wherein generating, by the processor of the computing system, the graphical representation of the team identity and the playing style corresponding to each of the plurality of variants comprises:
generating, by the processor of the computing system, one or more graphical representation trajectories for each of the plurality of variants.

14. The method of claim 8 , further comprising:
predicting, by the processor of the computing system, a likelihood of a third team executing the target play by analyzing the plurality of variants with the neural network to determine a second number of target variants that exhibit a playing style similar to the third team.

15. A system comprising:
a processor; and
a memory having programming instructions stored thereon, which, when executed by the processor, causes the system to perform operations comprising:
receiving, by the processor, match data for a match between a first team and a second team from the memory;
extracting, by the processor, from the match data, one or more portions of data corresponding to a target play of the first team of the match, wherein the one or more portions of data include tracking data, and wherein the tracking data includes trajectory data;
inputting, by the processor, the tracking data into a variational autoencoder, wherein the variational autoencoder includes a generative model configured to receive the trajectory data and generate a plurality of variants of target play based on the trajectory data;
based on the i

Assignees

• Stats Llc

Inventors

• Ganguly Sujoy
• Sha Long
• Hobbs Jennifer
• Wei Xinyu
• Lucey Patrick Joseph

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/0475

  Generative networks · CPC title

• G06V10/82

  using neural networks · CPC title