مقايسه عملكرد مدل هاي مختلف در خصوص پيش بيني نوسان بازده بورس اوراق بهادار تهران و تحليل تأثير برخي عوامل بر رفتار نوسان بازده

Comparing the Forecasting Ability of Deferent Models of Volatility in Tehran Exchange Dividend Price Index

The present research analyzes the forecasting performance of a variety of conditional and non-conditional models of TEDPIX volatility at the daily frequencies under three performance criteria: The root mean square error (RMSE), the mean absolute error (MAE) and the Theil index. Under RMSE and Theil criteria, results show MA250, exponential smoothing, and COARCH models have better performance among non conditional and conditional models respectively. Comparing forecasting performance of conditional and non conditional models shows that MA250 and ES models had better performance relative to conditional models. Other results of the study also reveal that according to conditional volatility models (except PARCH) there is a significant relationship between volatility and the targeted volatility range This result cannot be approved by ARMA. Furthermore, volatility varies in different return.