r语言arch模型分析报告 附数据代码

R代码复制到相应后面（能附上运行得到的图不）数据读取和处理（r=log(/)。##读取数据golddata=read.csv("数据.csv")head(golddata)##日期收盘价##12008/1/25385.103##22008/1/35422.034##32008/1/45483.650##42008/1/75556.593##52008/1/85528.054##62008/1/95613.758golddata=golddata[,2]head(golddata)##[1]5385.1035422.0345483.6505556.5935528.0545613.758Valuedata<-golddata##ValuedataValuedata=ts(Valuedata,start=c(2008,2),frequency=365)n<-length(Valuedata)###为减少误差，在估计时，根据每个交易日的收盘价对日收益率进行自然对数处理，即将收益率根据以下公式进行计算：#绘制收益率波动图Valuedata1<-log(lag(Valuedata))-log(Valuedata)R软件，画出日对数收益率线形图（1）plot.ts(Valuedata1)#收益率的基本统计表出下表summary(Valuedata1)## Min. 1stQu. Median Mean 3rdQu. ##-0.0915400-0.00825900.0004899-0.00020730.00901000.0893100library(asbio)library(asbio)#Functionsforskewnessandkurtosis.##Loadingrequiredpackage:tcltk#datadescriptionfunction#datadescriptionfunctiondatadesc=function(X){result=list(0);#resultlisttoreturnmean=mean(X);#meanvar=var(X)#variance,pearsonskew=3*(mean(X)-median(X))/sd(X)#Pearsoncoefficientofskewnesskurt=kurt(X) #kurtosis,quantile1=quantile(X,probs=0.25)# firstquartile,med=median(X)# median,quantile3=quantile(X,probs=0.25)# thirdquartile,max=max(X)# minimumandmin=min(X)# maximum.result=list(mean=mean,variance=skewness=pearsonskew,kurtosis=kurt,"firstquartile"=quantile1,median=med,"thirdquartile"=quantile3,"maximum"=max,minimum=minimum=min)return(result)}datadesc(Valuedata1)##$mean##[1]-0.0002073343####$variance##[1]0.0003538641####$skewness##[1]-0.1111916####$kurtosis##[1]3.309377####$`firstquartile`## 25%##-0.008258792####$median##[1]0.0004898845####$`thirdquartile`## 25%##-0.008258792####$maximum##[1]0.08931021####$minimum##[1]-0.09154204##直方图hist(Valuedata1)通过R3.3093773.309377，远高于正态分布的峰度值3率序的检验均失效#收益率序列的平稳性检验F检验）library(tseries)平稳性检验最常用的方法为单位根方法，运R验结果如下print(adf.test(diff(Valuedata1),alternative="stationary",k=0))##Warninginadf.test(diff(Valuedata1),alternative="stationary",k=0):##p-valuesmallerthanprintedp-value####AugmentedDickey-FullerTest####data:diff(Valuedata1)##Dickey-Fuller=-76.851,Lagorder=0,p-value=0.01##alternativehypothesis:stationaryp-value0.05，从而拒绝原假设，表明收益率不存在单位根，是平稳序列，即服I(0)过程通过通过Racf(Valuedata1)pacfpacf(Valuedata1)##从自相关图和偏自相关图的结果来看，对数收益率的自相关函数值和偏自相关函数值很快落入置信区间，因此对数收益率稳定。#ARCH效应检验#1.滞后阶数的选折及均值方程的确定library(FinTS)##Loadingrequiredpackage:zoo####Attachingpackage:'zoo'##Thefollowingobjectsaremaskedfrom'package:base':#### as.Date,as.Date.numeric#getSymbols("XPT/USD",src="oanda")#Valuedata1ones<-rep(1,length(Valuedata1))ols<-lm(Valuedata1~ones);ols####Call:##lm(formula=Valuedata1~ones)##lm(formula=Valuedata1~ones)####Coefficients:##(Intercept)##-0.0002073onesNAresiduals<-ols$residualsArchTest(residuals,lags=1)ArchTest(residuals,lags=5)ArchTest(residuals,lags=12)####ARCHLM-test;Nullhypothesis:noARCHeffects####data:residuals##Chi-squared=242.63,df=12,p-value<2.2e-16根据Chi-squared11，则公式可以写成2.残差序列自相关检验（日收益率的残差和残差平方自相关图）6：日收益率差平方自相关图acf(residuals)########ARCHLM-test;Nullhypothesis:noARCHdata:residualseffects##Chi-squared=66.824,df=1,p-value=3.331e-16####ARCHLM-test;Nullhypothesis:noARCHeffects####data:residuals##Chi-squared=191.09,df