Deep learning for financial time series forecasting
With the growing knowledge of deep learning, the deep learning knowledge and skills are used more and more in our daily life. The project is aimed to investigate the application of deep learning methods for financial time series forecasting. Financial time series forecasting is extremely challenging...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Theses and Dissertations |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/76037 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | With the growing knowledge of deep learning, the deep learning knowledge and skills are used more and more in our daily life. The project is aimed to investigate the application of deep learning methods for financial time series forecasting. Financial time series forecasting is extremely challenging due to the inherent non-linear and non-stationary characteristic of the trading market and financial time series. Stock market price is one of the most important indicators of a country’s economic growth. That’s why determining the exact movements of stock market price is considerably regarded. However, complex and uncertain behaviors of stock market exact determination impossible and hence strong forecasting models are deeply needed for making decision. Recurrent neural network (RNN) is a kind of artificial neural network which a sequence of information goes from one model to another model. This make the information can be different from each time step and may also maintain some information stay the same. We use many different kinds of RNN models to forecast the stock price. In real situation, we are more care about the price will increase or decrease. So we can verify our forecast results to find out the increase or decrease accuracy. Random forests (RF) is an ensemble learning method. It operate by constructing a multitude of decision trees at training time and outputting the class that is the mode of the classes (classification) or mean prediction of the individual trees. We also used some preprocessing data skill to try to improve the performance. We take the data as a kind of signal. We use the Empirical Mode Decomposition (EMD) and Discrete Wavelet Transform (DWT) to do this job. |
|---|