DOIONLINE

DOIONLINE NO - IJMAS-IRAJ-DOIONLNE-3872

Publish In
International Journal of Management and Applied Science (IJMAS)-IJMAS
Journal Home
Volume Issue
Issue
Volume-2,Issue-1, Special Issue-1  ( Jan, 2016 )
Paper Title
Spectral Analysis On Forecasting Sri Lankan Share Market Returns
Author Name
W.G. S. Konarasinghe, N. R. Abeynayake, L.H.P.Gunaratne
Affilition
Postgraduate Institute of Agriculture, University of Peradeniya, Sri Lanka. Faculty of Agriculture and Plantation Management, Wayamba University of Sri Lanka, Makandura, Gonawila (NWP), SriLanka. Department of Agricultural Economics and Business Management Faculty of Agriculture, University of Peradeniya Peradeniya, Sri Lanka.
Pages
25-27
Abstract
spectral analysis reveals that a wave can be generated by a packet of waves with different amplitudes and angular speeds. This concept is applied in the present study for modeling sri lankan stock returns, as they show wave like patterns. Daily share prices of random sample of six business sectors of colombo stock exchange (cse) were collected for the period year 1994- 2014.monthly returns were used for the data analysis, taking one third of each data set for model fitting and the rest for model verification. Time series plots were used to check whether data follows wave like patterns and auto correlation functions (acf) were used to test the stationary of series. Fourier transformation was used to transform a time series of returns (rt) into a series of trigonometric functions and multiple regression analysis was used to estimate the amplitudes of waves. Anova technique was used to test the significance of overall model and t- test was used to test the significance of regression coefficients. Model assumptions were tested by residual plots. Model assessment was based on mean square error (mse) and mean absolute deviation (mad). Based on the results it was concluded that fourier transformation along with multiple regression is suitable for forecasting sector returns of cse. However the tested method is successful only if the data series is stationary type. It is recommended to extend the method for non stationary series. Key words- Spectral analysis, Stationary series
  View Paper