Prediction of NBA talent and quality from non-professional tracking data

Patent metadata

Field	Value
Publication number	US-11682209-B2
Application number	US-202117449694-A
Country	US
Kind code	B2
Filing date	Oct 1, 2021
Priority date	Oct 1, 2020
Publication date	Jun 20, 2023
Grant date	Jun 20, 2023

Abstract

A computing system identifies broadcast video for a plurality of games in a first league. The broadcast video includes a plurality of video frames. The computing system generates tracking data for each game from the broadcast video of a corresponding game. The computing system enriches the tracking data. The enriching includes merging play-by-play data for the game with the tracking data of the corresponding game. The computing system generates padded tracking data based on the tracking data. The computing system projects player performance in a second league for each player based on the tracking data and the padded tracking data.

First claim

The invention claimed is:

1. A method, comprising:
identifying, by a computing system, broadcast video for a plurality of games in a first league, wherein the broadcast video comprises a plurality of video frames;
generating, by the computing system, tracking data for each game from the broadcast video of a corresponding game;
enriching, by the computing system, the tracking data, the enriching comprising merging play-by-play data for the game with the tracking data of the corresponding game;
generating, by the computing system, padded tracking data based on the tracking data; and
projecting, by the computing system, player performance in a second league for each player based on the tracking data and the padded tracking data.

2. The method of claim 1 , wherein enriching, by the computing system, the tracking data further comprises:
refining player and ball precisions in each frame of a respective broadcast video.

3. The method of claim 1 , wherein enriching, by the computing system, the tracking data further comprises:
automatically detecting events, via a neural network, in each frame of a respective broadcast video.

4. The method of claim 3 , further comprising:
enhancing the detected events with contextual information, the contextual information comprising defensive matchup information.

5. The method of claim 4 , further comprising:
generating an influence score for each defensive matchup, wherein the influence score captures an influence of a defender on a respective defensive matchup.

6. The method of claim 1 , wherein generating, by the computing system, the padded tracking data based on the tracking data comprises:
creating new player representations using mean-regression.

7. The method of claim 1 , further comprising:
providing, by the computing system, the tracking data to a raw data model;
providing, by the computing system, the padded tracking data to a padded data model; and
ensembling, by the computing system, a first output from the raw data model and a second output from the padded data model.

8. The method of claim 7 , wherein projecting, by the computing system, the player performance for each player based on the tracking data and the padded tracking data comprises:
providing the tracking data to a second raw data model;
providing the padded tracking data to a second padded data model; and
ensembling a third output from the second raw data model, a fourth output from the second padded data model, the first output, and the second output to classify the respective player into a bin.

9. A non-transitory computer readable medium comprising one or more sequence of instructions, which, when executed by a processor, causes a computing system to perform operations comprising:
identifying, by a computing system, broadcast video for a plurality of games, wherein the broadcast video comprises a plurality of video frames;
generating, by the computing system, tracking data for each game from the broadcast video of a corresponding game;
enriching, by the computing system, the tracking data, the enriching comprising merging play-by-play data for the game with the tracking data of the corresponding game;
generating, by the computing system, padded tracking data based on the tracking data;
identifying, by the computing system, a subset of players that have at least a threshold percentage chance of being drafted based on the tracking data and the padded tracking data; and
projecting, by the computing system, a range of draft positions for each player of the subset of players based on the tracking data and the padded tracking data.

10. The non-transitory computer readable medium of claim 9 , wherein enriching, by the computing system, the tracking data further comprises:
refining player and ball precisions in each frame of a respective broadcast video.

11. The non-transitory computer readable medium of claim 9 , wherein enriching, by the computing system, the tracking data further comprises:
automatically detecting events, via a neural network, in each frame of a respective broadcast video.

12. The non-transitory computer readable medium of claim 11 , further comprising:
enhancing the detected events with contextual information, the contextual information comprising defensive matchup information.

13. The non-transitory computer readable medium of claim 12 , further comprising:
generating an influence score for each defensive matchup, wherein the influence score captures an influence of a defender on a respective defensive matchup.

14. The non-transitory computer readable medium of claim 9 , wherein generating, by the computing system, the padded tracking data based on the tracking data comprises:
creating new player representations using mean-regression.

15. The non-transitory computer readable medium of claim 9 , wherein identifying, by the computing system, the subset of players that have at least the threshold percentage chance of being drafted based on the tracking data and the padded tracking data comprises:
providing the tracking data to a raw data model;
providing the padded tracking data to a padded data model; and
ensembling a first output from the raw data model and a second output from the padded data model to generate a percent likelihood of a respective player being drafted.

16. The non-transitory computer readable medium of claim 15 , wherein projecting, by the computing system, the range of draft positions for each player of the subset of players based on the tracking data and the padded tracking data comprises:
providing the tracking data to a second raw data model;
providing the padded tracking data to a second padded data model; and
ensembling a third output from the second raw data model, a fourth output from the second padded data model, the percent likelihood of a respective player being drafted, the first output, and the second output to classify the respective player into a bin.

17. 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:
identifying broadcast video for a plurality of games, wherein the broadcast video comprises a plurality of video frames;
generating tracking data for each game from the broadcast video of a corresponding game;
enriching the tracking data, the enriching comprising merging play-by-play data for the game with the tracking data of the corresponding game;
generating padded tracking data based on the tracking data;
identifying a subset of players that have at least a threshold percentage chance of being drafted based on the tracking data and the padded tracking data; and
projecting a range of draft positions for each player of the subset of players based on the tracking data and the padded tracking data.

18. The system of claim 17 , wherein identifying the subset of players that have at least the threshold percentage chance of being drafted based on the tracking data and the padded tracking data comprises:
providing the tracking data to a raw data model;
providing the padded tracking data to a padded data model; and
ensembling a first output from the raw data model and a second output from the padded data model to generate a percent likelihood of a respective player being drafted.

19. The system of claim 18 , wherein projecting the range of draft positions for each player of the subset of players based on the tracking data and the padded tracking data comprises:
providing the tracking

Assignees

• Stats Llc

Inventors

• Patton Andrew
• Walker Nathan
• Scott Matthew
• Ottenwess Alex

Classifications

• G06N5/042

  Backward inferencing · CPC title

• G06T7/292

  Multi-camera tracking · CPC title

• G06V20/44

  Event detection · CPC title

• G06V20/42 Primary

  of sport video content · CPC title

• G06N3/09

  Supervised learning · CPC title