Cross European markets examination of using neural network for stock picking
Artificial Neural Network (ANN) is a computer programme that mimics the cognitive processes of the human brain. Empirical researches show that ANN is a viable alternative to traditional statistical methods. ANN is a technique that used to be heavily researched and used widely in engineering and scie...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2009
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/17831 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Artificial Neural Network (ANN) is a computer programme that mimics the cognitive processes of the human brain. Empirical researches show that ANN is a viable alternative to traditional statistical methods. ANN is a technique that used to be heavily researched and used widely in engineering and scientific fields for various purposes ranging from control systems to artificial intelligence. But now, due to its astonishing generalization power, financial researchers and practitioners are taking an interest in the feasibility of applying ANN in financial.
This research attempts to explore the usefulness of neural network in stock index in stock index forecasting in the European context 30 days in the future. Technical indicators are used to train the ANN to forecast three European stock market indices. They are namely the Financial Times Stock Exchange 100 stock index (FTSE100), Compagnie Nationale des Agents de Change (CAC40) and last but not least, Deutscher Aktien-Index (DAX30).
Optimized inputs are also obtained by trials and errors. Input variables found to be useful in forecasting stock indices include 2-years historical opening, high, low, close prices, simple moving average of closing price and volume, relative strength index of price and major world stock indices. Experiments were also carried out to find the most appropriate network parameters that generate the most ideal results.
Comparisons were also done between the predicted outputs with the previous 30 days stock prices. Thus, a list of promising stocks can be selected and be recommended to the investors. The returns from simulated trading also calculated in this study. |
|---|