STOCK MARKET PREDICTION USING GENERATIVE ADVERSARIAL NETWORK (GAN) – STUDY CASE GERMANY STOCK MARKET
MICHAL MEC, MIKULAS ZEMAN, KLARA CERMAKOVA
Abstract:
Using neural networks in economics time series data is a new unexplored field. Lot of companies and economics research use mostly logistic regression or statistical approach when they try to predict the movement of stock market. Neural networks became a frequent tool for prediction in recent years and this approach has been confirmed to provide more reliable and better solutions when it comes to prediction and accuracy power. Within a wider context of current debate on neural networks employment in stock market predictions, we suggest an innovative methodology based on the combination of neural networks. In our analysis we use Wasserstein Generative Adversarial Network (WGAN) on Germany stock market as an example. We present how the trading strategy could be established on the prediction of the model and how it can be compared with other models in terms of returns. Overall, the WGAN monthly prediction outperformed Random Forest by 36%, benchmark by 32% and LSTM by 26% in the testing period. Our results also suggest that the WGAN model has on average higher returns than pure investment into index. Furthermore, WGAN is less volatile, which is always the preferred option for investors. Using neural networks for stock index prediction and confirming that WGAN investment strategy brings higher returns compared to generally used models is our main contribution to the current debate.
Keywords:
Generative adversial network; Germany stock market; neural network; stock prediction
DOI: 10.52950/ES.2024.13.2.005
PDF:
Download
APA citation:
MICHAL MEC, MIKULAS ZEMAN, KLARA CERMAKOVA (2024). Stock market prediction using Generative Adversarial Network (GAN) – Study case Germany stock market. International Journal of Economic Sciences, Vol. XIII(2), pp. 87-103. , DOI: 10.52950/ES.2024.13.2.005
Data:
Received: 23 Oct 2024
Revised: 30 Oct 2024
Accepted: 1 Nov 2024
Published: 15 Nov 2024
Copyright © 2024, Michal Mec et al, michalmec@gmail.com