1. 一般这种比较out-of-sample forecasting的,都是 static forecast (one-step-ahead), NOT dynamic forecast. Actual value in time N would be used to forecast N+1, and actual N+1 for N+2, ......and so on.
2. There're many R packages that help forecast and provide statistical accuracy, say Mean Squared Forecast Error, or Mean Absolute Error..... You could download their manuals, and find which one best fits you.
3. I'm not quite familiar with ETV-GARCH..... sorry for this.


雷达卡



以Yen/USD为例,我取Hourly spot rate的Log Return,这里可以把他的绝对值看为其volatility (Conditional Mean=0)

京公网安备 11010802022788号







