Data Hackathon

Market Scenario Generator Hackathon: From Stability to Storms!-Award WinningModels

We are excited to announce the top two winners of the Market Scenario Generator Hackathon. 

1. City and Warwick Team: Szymon Kubiak (City, University of London), TillmanWeyde (City, University of London), Oleksandr Galkin (City, University of London), Dan Philps (University of Warwick) and Ram Gopal (University of Warwick).
2. Tuan Anh Nguyen.

Congratulations to the City and Warwick Team and Tuan Anh Nguyen！

About the competition, please refer to Market Scenario Generator Hackathon Link.

Award-Winning Models Introduction:

Based on previous research, the City and Warwick Team used the Diffusion Model for the Hackathon, with U-Net as the base architecture. The U-Net architecture played a crucial role in the denoising phase. The second-place model, Conditional Variational Autoencoder (Conditional VAE), was developed by Tuan Anh Nguyen. This model uses differentiated activation functions for log returns and volatility, and its loss function is designed by incorporating correlation matrices and KL divergence. Their model designs, along with key numerical results, can be found through the following links.

First Place Model：

• Watch Video (YouTube Link)

Second Place Model

• Watch Video (YouTube Link)
• View PPT (PPT Link)
• Codes (GitHub Link)

Thank you to all participants for making this event a success!