Application of Distributed Computing in Developing Architecture of Intelligent Information System for Automated Stock Exchange Trading

Kristina Reizenbuk, Tatyana Sarapulova, Semen Shchedrin, Irina Shchedrina

The paper analyses the principles of building distributed systems and the methods of load balancing among the nodes of such networks. An algorithm of distributed training of neural networks is described and architecture of an intelligent information system is developed for automated stock exchange trading with the potential to be used further for designing interactive trading terminals for stock, currency or cryptocurrency markets.To speed up the training of neural networks, a distributed algorithm is implemented, operating with a core application as the central node of a distributed system connected to auxiliary nodes as may be required.The paper includes a description of the methods of fundamental and technical analysis of stock markets, the principles of designing and training neural networks and an overview of the Russian market of software for analysing and forecasting security prices. The research also concerns the main principles and mechanisms of operation and design of distributed networks, as well as the technologies and techniques used in these processes. The architecture of the Financial Neural Network software complex is described.

Volume 11 | 08-Special Issue

Pages: 2549-2555