R/sup_functions.R

In learnfin/learnfin: Statistical Learning Methods for Financial Decision Making

#Below functions are supplementary functions used in learnfin package

#' Calculate Pricing Error
#' @param mar_price Market price of the contract.
#' @param mod_price Model price of the contract.
#' @param error_type The pricing error type. It will be automatically calculated. Choose from (Absolute) Relative Pricing Error (denoted with RPE or ARPE), squared error (denoted with SE) or absolute pricing error (APE).
#' @export
calculate_pricing_error<-function(mar_price,mod_price,error_type){
  if(error_type=="ARPE"){
    #Absolute Relative Pricing Error
    return(abs((mod_price-mar_price)/mar_price))
  }
  if(error_type=="RPE"){
    #Relative Pricing Error
    return((mod_price-mar_price)/mar_price)
  }
  if(error_type=="APE"){
    #Absolute Pricing Error
    return(abs(mod_price-mar_price))
  }
  if(error_type=="SE"){
    #Squared Error
    return((mod_price-mar_price)^2)
  }
  #If error type is incorrect show the warning
  stop("Incorrect error type")
}

#' @export
rescale_error<-function(errors,mederror=NA){
	if(is.na(mederror)){
    #If there is no predetermined medium error value, use median errors to scale the errors.
		ifelse(errors <= median(errors),(errors-min(errors))/(median(errors)-min(errors))*0.5,(errors-median(errors))/(max(errors)-median(errors))*0.5+0.5)
	}else{
		ifelse(errors <= mederror,(errors-min(errors))/(mederror-min(errors))*0.5,(errors-mederror)/(max(errors)-mederror)*0.5+0.5)
	}
}

#' @export
get_the_cluster<-function(Mon_sc,Mat_sc,center_matrix){

	which.min(colSums((t(center_matrix)-c(Mon_sc,Mat_sc))^2))

}


#' @export
get_dm_results<-function(PSet1,PSet2,error_data){
    dm_result<-tryCatch(forecast::dm.test(unlist(error_data[,PSet1]),unlist(error_data[,PSet2]),power=1),error=function(e) return(FALSE))
    if(dm_result[1] == FALSE){
        dm_df<-data.frame(DM= NA,p_val=NA)

    }else{
        dm_df<-data.frame(DM=dm_result$statistic,p_val=round(dm_result$p.value,4))
    }
    return(dm_df)
}