R/parametricES.R

In f-kailin/varmonitor: Risk Management Tools: Value-at-Risk Monitor

parametricES <- function (confidence, position, sigma, horizon = 1) {
  # Reorganize the input
  confidence <- sort(confidence)
  horizon    <- sort(horizon)
  
  # Initializing the output setup
  ES         <- c()
  data       <- matrix(0, nrow = 3, ncol = length(confidence)*length(horizon))
  data[1, ]  <- sort(rep(confidence, length(horizon)))
  data[2, ]  <- rep(horizon, length(confidence))
  
  # Main
  if (is.matrix(sigma) || is.array(position)) {
    # Portfolio with N securities
    for (i in 1:length(confidence)) {
      for (j in 1:length(horizon)) {
        ES[length(ES) + 1]      <- abs(as.numeric(sqrt(position%*%sigma%*%position) * exp(-qnorm(1-confidence[i])^2/2)/(sqrt(2*pi)*(1-confidence[i])) * sqrt(horizon[j])))
      }
    }
  } else {
    # Portfolio with one security
    for (i in 1:length(confidence)) {
      for (j in 1:length(horizon)) {
        ES[length(ES) + 1]      <- as.numeric(abs(position)*sigma * exp(-qnorm(1-confidence[i])^2/2)/(sqrt(2*pi)*(1-confidence[i])) * sqrt(horizon[j]))
      }
    }
  }
  data[3, ]  <- round(ES, 2)
  
  # Output
  return(data.frame(cbind(c("Confidence level", "Time horizon", "Expected Shortfall"), data)))
}