| Summary: | 碩士 === 南台科技大學 === 財務金融系 === 98 === ETF, traditionally, tracks the specific index as the main method of operation where ,theoretically, the tracking object should match the performance of fund. This research had focused on 5 China CSI 300 ETF, analyzed by Sharpe and Jensen's α performance indicators, and discovered the remarkable differences of each fund in terms of performance.
This paper use HS、EWMA、GARCH、TGARCH were the four estimation models which used to estimate 125 and 250 moving windows for the VaR in daily basis follow by back testing and Likelihood Ratio Test to find out the performance of estimation models. According to the facts of analysis, the choice of estimation model may varied from the research sample. Since the accuracy of Value-at-risk Based on Forecast might differ from estimation period, confidence level to parameter settings. Therefore, the selections of models and parameter settings were the main causes for forecast. The result demonstrated no model outperformed others. The causes might be the effectiveness of strategies which lead to the variations of tracking errors followed by the difference of assets, effecting the goodness of fit on both assets and estimation models.
|
|---|
