Predictability of stock prices in Tehran stock exchange index (TEPIX) is investigated. Five econometric models including Random Walk, Autoregressive, GARCH(1,1), TGARCH and EGARCH performed to eliminate linear structures form series and five statistical tests employed to examine the nonlinearity in residuals of that econometrical models.
Results from econometric models rejected the random walk hypothesis, this shows that we can accept the possibility of prediction with historical data. From results of nonlinearity tests we can accept existence of nonlinear structures in the residuals of models. This is another evidence to accept the predictability in TEPIX