#################################
### R commands used in Lecture 9:
#################################
> da=read.table("d-ibm-0110.txt",header=T)
> head(da)
> ibm=log(da$return+1)
> nibm=-ibm
### RiskMetrics #########
> source("RMfit.R")
> RMfit(nibm)
### One can use default parameter beta = 0.96 wihtout estimation with the following command
> RMfit(nibm,estim=F)
### Econometric modeling
> require(fGarch)
> m1=garchFit(~garch(1,1),data=nibm,trace=F)
> summary(m1)
> pm1=predict(m1,10)
> pm1
> source("RMeasure.R")
> RMeasure(-.0006,.00782)
> names(pm1)
#### 10-day VaR
> v1=sqrt(sum(pm1$standardDeviation^2))
> RMeasure(-0.006,v1)
> m2=garchFit(~garch(1,1),data=nibm,trace=F,cond.dist="std")
> summary(m2)
> pm2=predict(m2,1)
> pm2
> RMeasure(-.000411,.0081,cond.dist="std",df=5.751)

### Empirical quantile and quantile regression
> quantile(nibm,c(0.95,0.99,0.999))
> da1=read.table("d-ibm-rq.txt",header=T)
> fix(da1)
> require(quantreg)
> m3=rq(nibm~vol+vix,data=da1,tau=0.95)
> summary(m3)
> ts.plot(nibm)
> lines(m3$fitted.values,col="red")

### Extreme value theory
> require(evir)
> m4=gev(nibn,block=21)
> m4
> source("evtVaR.R")
> evtVaR(0.2517,0.0103,0.0297)

### Peaks over threshold
> m4a=pot(nibm,thres=0.01)
> plot(m4a)
> riskmeasures(m4a,c(0.95,0.99,0.999))

### generalized Pareto distribution
> m5=gpd(nibm,0.01)
> m5
> plot(m5)
> riskmeasures(m5,c(0.95,0.99,0.999))

#### Simulation to check on limiting distribution of maximum
source("EVTsim.R")  ## The file has two functions: EVTsim and qgumble
m1 <- EVTsim(n=500,iter=5000)
quantile(m1$stmax,prob=c(0.95,0.975,0.99))
qgumble(prob=c(0.95,0.975,0.99))  ## Compute quantile of the standard Gumble distribution