The Comparison of Models for Forecasting the Prices of Commodity Futures－by Gold, Oil and Natural gas
| Autor: | Jia-yu Lu, 呂佳育 |
|---|---|
| Rok vydání: | 2010 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 98 The traditional forecast models are used to predict cross-section data, the model places the factor into their models to show the impact of price and quantity; the present prediction models do not place the factor into the model, because it has been historical data on the response in the past forecasts. We can know the cumulative change in the past from the history of the data. If you want to place the factors into the model, the factor will have a lot of influences. We will use a single variable to explain the nature of current status and future with the past history information and data. More types of finance information are based on the literature. They use a time series model to analyze the research model and the common use of the three futures markets. To verify the best ability to explain the model used to explain the past and the present, and to connect with the future. Therefore, this study establish a variety of three futures price forecasting model of raw materials, compare their predictive ability, and verify the predictive ability of simple models in three different goods is better than complex models. To avoid the mistakes of the cycle biased data, this article used diary data and weekly data to analysis and compare with it at the same time, and verify that will the cycle information affect the model predictability. Whether day or week of data and information data, in the three indicators in this study, RESE, MAE, MAPE is that exponential smoothing model are ES0.9 shows the best predictive ability for the target, followed by ARIMA model better, the next is the ARCH model, the final prediction error of the overall MA20 bottom. Therefore, we can see that the simple model prediction ability than the complex models to predict the good. Keywords: raw materials futures, Time Series, Forecasting |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|