Simulating financial time series using attention

Fu, Weilong; Hirsa, Ali; Osterrieder, Jörg Robert (2022). Simulating financial time series using attention Cornell University 10.48550/arXiv.2207.00493

[img] Text
2207.00493.pdf
Restricted to registered users only

Download (1MB) | Request a copy

Financial time series simulation is a central topic since it extends the limited real data for training and evaluation of trading strategies. It is also challenging because of the complex statistical properties of the real financial data. We introduce two generative adversarial networks (GANs), which utilize the convolutional networks with attention and the transformers, for financial time series simulation. The GANs learn the statistical properties in a data-driven manner and the attention mechanism helps to replicate the long-range dependencies. The proposed GANs are tested on the S&P 500 index and option data, examined by scores based on the stylized facts and are compared with the pure convolutional GAN, i.e. QuantGAN. The attention-based GANs not only reproduce the stylized facts, but also smooth the autocorrelation of returns.

Item Type:

Working Paper

Division/Institute:

Business School > Institute for Applied Data Science & Finance > Finance, Accounting and Tax
Business School

Name:

Fu, Weilong;
Hirsa, Ali and
Osterrieder, Jörg Robert0000-0003-0189-8636

Subjects:

H Social Sciences > HG Finance

Publisher:

Cornell University

Language:

English

Submitter:

Jörg Robert Osterrieder

Date Deposited:

22 Aug 2022 12:57

Last Modified:

22 Aug 2022 12:57

Publisher DOI:

10.48550/arXiv.2207.00493

Uncontrolled Keywords:

deep learning, generative adversarial networks, attention, time series, stylized facts

ARBOR DOI:

10.24451/arbor.17396

URI:

https://arbor.bfh.ch/id/eprint/17396

Actions (login required)

View Item View Item
Provide Feedback