Evolutionarily Developed Neural Networks for Investment Strategies Construction
Janusz Morajda · 2003
The study presents the idea of feedforward evolutionary neural network that utilizes (instead of classical training) a specific evolutionary procedure for network development. This procedure is responsible for network evolving, connections selection and weight values determination. Basic features of such networks and the algorithm of network development are submitted. Such a network has been applied to a problem of construction of investment strategy for Polish stock index WIG 20. The evolutionary network development process has been characterized and discussed. Relatively good results of application of generated by the network investment strategies have been obtained for the test data. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.