Rețele neuronale utilizate în evaluarea performanței pe piața de capital
Neural networks in capital market performance evaluation

Summary/Abstract: The paper presents how neural networks work and their application to classify companies listed on the stock exchange, which after one year will record higher prices of shares, thus identifying possibilities for capital gains. The data used for the classification are 66 company indicators from the initial and the business sector names. Capital gains are calculated after one year, and based on this result, the classes to be assigned to the firms are determined. The type of neural network used is "feed-forward," and the learning process is "back-propagation". The study is divided into four parts. The first part is the introduction and briefly shows how this field has evolved, the current level and examples of its uses. The second part describes how a neural network works in general, the association with human brain operating rules, the "delta" learning rule and other particularities. The third part presents the application of neural network and the obtained results. In the last part, conclusions and potential developments or uses of the application will be presented.

• Details
• Contents