The Superiority of the EGARCH-Odd Exponentiated Skew-t Model in Predicting Financial Returns Volatility

The key point for volatility forecast is to utilize appropriate innovation conditional density. the specification of flexible innovation density is very essential given that it directly affects the accuracy of volatility prediction. in this work, a new odd exponentiated skew-t (oest) innovation density is introduced for exponentiated generalized autoregressive conditional heteroscedasticity (egarch) models for modeling the daily volatility of financial return series. the simulation via monte carlo experiment indicates that the estimators compared are asymptotically unbiased and consistent given that their biases converge to zero as the sample size increases. the maximum likelihood and maximum product of spacing procedures dominate the other procedures. the real dataset application based on the first bank nigeria shock price index is given to show the performance of the egarch model specified under oest innovation density relative to normal, student-t, generalized error, skew-normal, skew student-t, skew generalized error, generalized hyperbolic and johnson reparametrized densities in terms of volatility accuracy. overall, the empirical results show that the egarch model with oest innovation density generates better in-and out-of-sample performance than all the other models.