What is it about?
This paper is about generating completely synthetic and new financial time series with the help of generative deep learning methods. Its purpose is to augment already existing financial datasets (stock markets) in order to develop more accurate financial tools such as stock market predictors, portfolio managers etc.
Featured Image
Photo by olieman.eth on Unsplash
Why is it important?
It is among the first works of its kind, demonstrating the use of generative models and introducing several qualitative and quantitative metrics for assessing the goodness of synthetically generated samples.
Perspectives
This paper has the potential to start a new direction in the field of generative methods focused on financial data. It also offers a starting point for following researchers as to what generative methods work better for this type of data. Moreover, it opens an infamous discussion regarding the way in which synthetic data should be evaluated - this problem is also present in other fields such as image generation.
Mihai Dogariu
University Politehnica of Bucharest
Read the Original
This page is a summary of: Generation of Realistic Synthetic Financial Time-series, ACM Transactions on Multimedia Computing Communications and Applications, November 2022, ACM (Association for Computing Machinery),
DOI: 10.1145/3501305.
You can read the full text:
Contributors
The following have contributed to this page
