Using the models of neural networks for making forecast of the securities pricing policy dynamics

The paper studies the problem of planning and forecasting, improving forecasts accuracy while constructing the strategy of investment portfolio management under the influence of various factors. The information flow control mechanism for investment portfolio based on the forecasts using neural network with the solution through control matrix is proposed. The results of neural network predicting for time series data, statistics, of different securities types return dynamics are presented. Analysis of possibilities to improve the accuracy of prediction is carried out and algorithms capable of making such forecasts are presented.