Study On Forecasting Methods Of Earnings Rate In Security Investment Managerial

Abstract:Investment activity in securities is a kind of effective and the most common investment behavior in the modern economic market, the rate of return on investment is one of the most important decision basis for the investor. Because the future rate of return is nondeterministic, the forecast before investment is important in the management of securities investment. In this dissertation, we present three forecasting algorithms, which are the regression estimation algorithm, the exponential smoothing algorithm and the GM (1,1) forecast algorithm, respectively, to predict the rate of the future return, and some comparisons among these algorithms are also made. From the numerical results, it is obtained that1. If the source of information is more, since the variables and the dependent variable are strong correlation, and only do short-term forecasting, then the linear regression method of forecasting is good effect.2. If the change is not a stationary time series, and the raw data is with fewer resources, then the prediction method for smoothing is good effect.3. If the original information is less, also with no laws, and not subject to any smooth distribution of the original sequence, then the GM (1,1) forecasting is good effect.In this paper one by one Details exponential smoothing prediction algorithm and multiple linear regression prediction algorithm and the GM (1,1) prediction algorithm, The basic principles of the three algorithms, modeling process, as well as the error of the test are described in detail, Instance of the use of some of the economic and market data on the three algorithms, Chapter5, the predictive effect of the three algorithms, use conditions and characteristics were analyzed to prepare for the practical application of Reference.