# R programs to use for Math/Econ 565 class
# Fall 2017 at American University   John Nolan   jpnolan@american.edu
# Revised 6/19/2024 to read data from Yahoo finance 
# as described in https://www.datasciencecentral.com/getting-historical-data-from-yahoo-finance-in-r/
# Dates are in (year,month,day) format, e.g. c(2023,12,31)  is December31, 2023 with most recent data first
##########################################################################################
get.stock.price <- function( symbol, start.date, stop.date, full.table=FALSE, print.info=TRUE ) {
# gets stock price data from Google Finance for specifed symbol
# default is to return ADJUSTED closing price 

  if( print.info ) cat(symbol," ")

  t1 <- ISOdate(start.date[1],start.date[2],start.date[3],hour=0)
  t2 <- ISOdate(stop.date[1],stop.date[2],stop.date[3],hour=24)
  url <- paste("https://query1.finance.yahoo.com/v7/finance/download/",
               symbol,
               "?period1=", as.integer(t1),
               "&period2=", as.integer(t2),
               "&interval=1d&events=history", sep="")

# changed from Yahoo Finance (who apparently stopped supporting this) to Google Finance  6 Oct 2017
#url <- paste("http://finance.google.com/finance/historical?q=",symbol,
#             "&startdate=",paste(start.date,collapse="-"),"&enddate=",paste(stop.date,collapse="-"),
#             "&output=csv",sep="")
#cat("url=",url,"\n\n")
  x <- read.csv(url) # in S-Plus, use read.table(url,sep=","), but note that it treats 
                   # the dates as row labels, not a separate column 

# data has most recent days first, going back to start date
n <- length(x$Date)
date <- as.character(x$Date[c(1,n)])
if (print.info) cat("has", n,"values from",date[1],"to",date[2],"\n") 
# data comes back in reverse order, starting with most recent.
# reverse the order
if (full.table) { # return full table
    return( x )
} else 
    return(x$Adj.Close) # return just the closing prices
}
##########################################################################################
get.portfolio.returns = function( symbols, start.date, stop.date  ){
# get a table (data.frame) of values for a list of stocks in the stated time period
n = length(symbols)
for (i in 1:n) {
  t1 = get.stock.price( symbols[i], start.date=start.date, stop.date=stop.date,full.table=T)
  #  need to merge columns, possibly with mismatching dates
  a = data.frame(t1$Date,t1$Close)
  names(a) = c("Date",symbols[i])
  if (i == 1) {b=a}
  else {b = merge(b,a,sort=FALSE)}
  }
#  leave off the date column
nn = dim(b)[1]
cat("    ",nn,"dates with values for all stocks,",nn-1,"returns calculated\n")
b = b[,2:ncol(b)]
bb = data.frame(apply(b,2,"log.ratio"))
names(bb) = symbols
return(bb) }
##########################################################################################
log.ratio <- function(x) { return(diff(log(x))) }
##########################################################################################
trimmed.hist <- function(x,low=-Inf,high=+Inf,nclass=20,...) {
# histogram with data in x "trimmed" to exclude values below "low" or above "high"
# useful to see a histogram of data that is visually distorted by extreme values
y <- x[ (x > low) & (x < high) ]
hist(y,nclass=nclass,...)
cat("original data had",length(x),"values, trimmed data has",length(y),"values\n")
invisible(y)}
##########################################################################################
# four plots on one window, use some color and show histogram of returns with normal fit
plot.basics <- function( symbol, start.date, stop.date ){

all <- get.stock.price(symbol,start.date=start.date,stop.date=stop.date,full.table=TRUE) 
xx <- all$Close
yy <- log.ratio(xx)
par(mfrow=c(2,2)) # start a new graph window with 4 plot areas
plot(xx,type='l',main=paste(symbol," price\n", 
           start.date[1],"/",start.date[2],"/",start.date[3], 
           " to ",stop.date[1],"/",stop.date[2],"/",stop.date[3],sep=""), 
           col='blue',  xlab="day",ylab="price")
plot(yy,type='l',main="return",col='blue',xlab="day",ylab="return")
plot(all$Volume,type='l',col='blue',xlab="day",ylab="",main="Volume")

bnds <- max(abs(yy))*c(-1,1)
hist(yy,prob=TRUE,main="histogram of returns",col='blue',xlim=bnds,xlab='return',nclass=50)
mu <- mean(yy); s <- sd(yy)
cat("normal fit: mean=",mu,"  std. dev=",s,"\n")
z <- seq(bnds[1],bnds[2],length=101)
w <- dnorm(z,mean=mean(yy),sd=sd(yy))
lines(z,w,type='l', xlab="return",ylab="",col='red',lwd=2 )
text(0.6*bnds[2],max(w)/2,"normal fit",col='red')
return() }
##########################################################################################
cat("Finance tools for R - Math 565/Econ 565   Fall 2017, American University\n")

# simple test - retrieve price of Google stock for January 2024
get.stock.price( "GOOG", start.date=c(2024,1,1), stop.date=c(2024,1,31) )

# plot of price, return, volume and histogram of returns
plot.basics( "GOOG", start.date=c(2023,1,1), stop.date=c(2023,11,31) )

# get returns for a portfolio with 3 stocks: Google, Ford, IBM for January 2024
get.portfolio.returns( symbols=c("GOOG", "Ford", "IBM"), 
                       start.date=c(2024,1,1), stop.date=c(2024,1,31))