Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
braverock
GitHub Repository: braverock/portfolioanalytics
 Path: blob/master/sandbox/RFinance2014/data_analysis.R
1433 views
1
library(PerformanceAnalytics)

2


3
source("data_prep.R")

4


5
figures.dir <- "figures"

6


7
# mix of blue, green, and red hues

8
my_colors <- c("#a6cee3", "#1f78b4", "#b2df8a", "#33a02c", "#fb9a99", "#e31a1c")

9


10
##### Equity Data for Example 1 and Example 2 #####

11
colors <- c(rep(my_colors[1], 15), rep(my_colors[2], 15), rep(my_colors[3], 5))

12
sd.order <- order(StdDev(equity.data))

13


14
# boxplot to compare return distributions

15
# mar(bottom, left, top, right)

16
# default: par(mar=c(5, 4, 4, 2) + 0.1)

17
png(paste(figures.dir, "equity_box.png", sep="/"), height = 500, width = 1000)

18
boxplot(coredata(equity.data[,sd.order]),

19
        cex.axis=0.8, las=3, ylab="Returns", pch=18,

20
        col=colors[sd.order],

21
        main="Return Distribution\n(sorted by StdDev)")

22
legend("topleft", legend=c("Large Cap", "Mid Cap", "Small Cap"), 

23
       fill=c(my_colors[1], my_colors[2], my_colors[3]), bty="n", cex=0.8)

24
dev.off()

25


26
##### edhec Data for Example 3 and Example 4 #####

27
p <- 0.95

28


29
png(paste(figures.dir, "relative_barvar.png", sep="/"))

30
charts.BarVaR(R[,1:3], width=60, methods=c("ModifiedES", "ModifiedVaR"), 

31
              main="Relative Value", colorset=rep(my_colors[2], 3))

32
dev.off()

33


34
png(paste(figures.dir, "directional_barvar.png", sep="/"))

35
charts.BarVaR(R[,4:6], width=60, methods=c("ModifiedES", "ModifiedVaR"), 

36
              main="Directional", colorset=rep(my_colors[4], 3))

37
dev.off()

38


39


40
colors <- c(rep(my_colors[2], 3), rep(my_colors[4], 3))

41
ES.order <- order(ES(R, p=p, invert=FALSE))

42


43
png(paste(figures.dir, "edhec_box.png", sep="/"), height = 500, width = 1000)

44
boxplot(coredata(R[,ES.order]),

45
        cex.axis=0.8, las=3, ylab="Returns", pch=18,

46
        col=colors[ES.order],

47
        main="Return Distribution\n(sorted by Modified ES (95%))")

48
legend("topleft", legend=c("Relative Value", "Directional"), 

49
       fill=c(my_colors[1], my_colors[2]), bty="n", cex=0.8)

50
dev.off()

51


52
# script for data analysis

53


54
# library(PerformanceAnalytics)

55
# library(lattice)

56
# library(corrplot)

57


58
# load("data/edhec.rda")

59
# 

60
# head(edhec)

61
# R <- edhec[,1:4]

62
# p <- 0.95

63
# 

64
# first(R)

65
# last(R)

66


67
# plot the timeseries of returns

68
# plot(as.zoo(edhec))

69
# xyplot(R, scales=list(y="same"))

70
# charts.BarVaR(R, width=36, methods=c("ModifiedES", "ModifiedVaR"))

71
# dev.off()

72


73
# boxplot to compare return distributions

74
# mar(bottom, left, top, right)

75
# default: par(mar=c(5, 4, 4, 2) + 0.1)

76
# par(mar=c(10, 4, 4, 2) + 0.1)

77
# boxplot(coredata(R[,order(ES(R, p=p, invert=FALSE))]),

78
#         cex.axis=0.8, las=3, ylab="Returns", pch=18,

79
#         main="Return Distribution\n(sorted by Modified ES (95%))")

80
# par(mar=c(5, 4, 4, 2) + 0.1)

81
# dev.off()

82


83
# head(R[,order(ES(R, invert=FALSE))])

84
# head(R[,order(StdDev(R))])

85
# chart.Boxplot(R[,order(ES(R, invert=FALSE))])

86
# chart.Boxplot(R[,order(StdDev(R))])

87
# boxplot(coredata(R), col=c(2:5), cex.names=0.8, las=3)

88


89
# chart the distribution of returns

90
# for(i in 1:ncol(R)){

91
#   chart.Histogram(R[,i], methods=c("add.density", "add.normal"), 

92
#                   colorset=c("lightgray", "black", "blue"))

93
#   legend("topleft", legend=c("kernel density estimate", "normal"), 

94
#          lty=c(1,1), col=c("black", "blue"), bty="n")

95
#   Sys.sleep(1)

96
# }

97


98


99
# chart the correlation and covariance

100
# from http://cran.r-project.org/web/packages/corrplot/vignettes/corrplot-intro.html

101
# cor.mtest <- function(mat, conf.level = 0.95) {

102
#   mat <- as.matrix(mat)

103
#   n <- ncol(mat)

104
#   p.mat <- lowCI.mat <- uppCI.mat <- matrix(NA, n, n)

105
#   diag(p.mat) <- 0

106
#   diag(lowCI.mat) <- diag(uppCI.mat) <- 1

107
#   for (i in 1:(n - 1)) {

108
#     for (j in (i + 1):n) {

109
#       tmp <- cor.test(mat[, i], mat[, j], conf.level = conf.level)

110
#       p.mat[i, j] <- p.mat[j, i] <- tmp$p.value

111
#       lowCI.mat[i, j] <- lowCI.mat[j, i] <- tmp$conf.int[1]

112
#       uppCI.mat[i, j] <- uppCI.mat[j, i] <- tmp$conf.int[2]

113
#     }

114
#   }

115
#   return(list(p.mat, lowCI.mat, uppCI.mat))

116
# }

117
# res <- cor.mtest(R)

118
# 

119
# corrplot(cor(R), p.mat=res[[1]], main="Correlation", 

120
#          sig.level=0.05, tl.cex=0.8)

121


122
# corrplot(M, method="number", bg="gray", tl.cex=0.8)

123
# corrplot.mixed(M, bg="gray", tl.cex=0.8)

124


125
# If I compare sample min variance portfolio to a ledoit-shrinkage or robust, 

126
# I should use plotcov to compare covaiance matrices

127


128


129


130


131


132