Tuesday, May 22, 2012

knitr Performance Report–Attempt 3

please see knitr Performance Report-Attempt 2 and knitr Performance Report-Attempt 1

Since the time of my last reporting post, RStudio, knitr, and Sweave have worked extremely hard to make document creation easier by becoming even more streamlined and cooperative (New Version of RStudio (v0.96)).  Thanks so much for these extremely helpful and generous contributors.

Over the years, I have developed an outline for client reports that I think applies in most situations in the money management world.  This also fits for performance reporting and marketing, except the transaction history section would either be eliminated or replaced by a section highlighting trades or ideas that have been used in the past.

From TimelyPortfolio

I thought I would have a completed product before displaying progress, but I just could not wait.  I started working on the Return section.  Please let me know what you think as your comments will guide the rest of the report.  If you are viewing in a RSS reader, you might need to view through this link, since the embed might not work.

R code from GIST:

%great guides at epslatex.pdf
%check miniplot for potential use
%\documentclass[english,nohyper,noae]{tufte-handout}
\documentclass{article}
\usepackage{graphics}
\usepackage{caption}
\usepackage{sidecap}
\usepackage{textpos}
%\usepackage[section]{placeins}
\title{Performance Report from knitr}
\author{Timely Portfolio}
\begin{document}
\maketitle
\SweaveOpts{concordance=TRUE}
<<eval=TRUE,echo=FALSE,results=hide,warning=FALSE>>=
#do requires and set up environment for reporting
require(ggplot2)
require(directlabels)
require(reshape2)
require(lattice)
require(latticeExtra)
require(xtable)
require(quantmod)
require(PerformanceAnalytics)
#trying some new colors out
mycolors=c(brewer.pal(9,"Blues")[c(7,5)],brewer.pal(9,"Greens")[6])
#mycolors=c(brewer.pal(6,"Blues)[c(3,5)],"slategray4")
data(managers)
#get xts in df form so that we can melt with the reshape package
#will use just manager 1, sp500, and 10y treasury
managers <- managers[,c(1,8,9)]
#add 0 at beginning so cumulative returns start at 1
#also cumulative will match up datewise with returns
managers <- as.xts(rbind(rep(0,NCOL(managers)),coredata(managers)),
order.by=c(as.Date(format(index(managers)[1],"%Y-%m-01"))-1,index(managers)))
managers.df <- as.data.frame(cbind(index(managers),coredata(managers)),stringsAsFactors=FALSE)
#melt data which puts in a form that lattice and ggplot enjoy
managers.melt <- melt(managers.df,id.vars=1)
colnames(managers.melt) <- c("date","account","return")
managers.melt[,1] <- as.Date(managers.melt[,1])
#get cumulative returns starting at 1
managers.cumul <- as.xts(
apply(managers+1,MARGIN=2,FUN=cumprod),
#add end of first month to accommodate the 1 that we add
order.by=index(managers))
managers.cumul.df <- as.data.frame(cbind(index(managers.cumul),
coredata(managers.cumul)),
stringsAsFactors=FALSE)
managers.cumul.melt <- melt(managers.cumul.df,id.vars=1)
colnames(managers.cumul.melt) <- c("date","account","return")
managers.cumul.melt[,1] <- as.Date(managers.cumul.melt[,1])
#this is tricky but necessary
#reorder accounts and indexes to preserve order with manager and then benchmarks
managers.cumul.melt$account <- factor(as.character(managers.cumul.melt$account),colnames(managers)[c(2,3,1)],ordered=TRUE)
#get rolling returns for 1y, 3y, 5y, since inception
trailing <- table.TrailingPeriods(managers[,c(2,3,1)], periods=c(12,36,60,NROW(managers)),FUNCS=c("Return.annualized"),funcs.names=c("return"))
trailing.df <- as.data.frame(cbind(c("1y","3y","5y",paste("Since Inception ",format(index(managers)[1],"%b %Y"),sep="")),
c(rep("return",4)), #will allow for multiple measures if we decide to include later
coredata(trailing)),
stringsAsFactors=TRUE)
trailing.melt <- melt(trailing.df,id.vars=1:2)
colnames(trailing.melt) <- c("period","measure","account","value")
#this is tricky but necessary
#reorder the period so that they will be in correct chronological order
trailing.melt$period <- factor(as.character(trailing.melt$period),rev(c("1y","3y","5y",paste("Since Inception ",format(index(managers),"%b %Y"),sep=""))),ordered=TRUE)
#reorder accounts and indexes to preserve order with manager and then benchmarks
trailing.melt$account <- factor(as.character(trailing.melt$account),colnames(managers)[c(3,2,1)],ordered=TRUE)
@
%\newpage
\section{Overview}
This section should serve as a dashboard or executive summary for quick and easy access to the most informative risk and return measures. Also, most marketing will have a text description of the strategy, process, objective, category, and potential investments of the product.
\newpage
\section{Returns}
Unfortunately, the Return section is generally the focus of the sales pitch and also is often the biggest concern for the prospect. Although it easiest to sell on return in the short-term, long-term success requires much more focus on the graphs presented in the Overview and Risk sections.
\begin{figure}
\begin{wide}
%\begin{minipage}[t]{1.25\linewidth}
\begin{textblock*}{165mm}(5mm,50mm)
%\hspace*{-0.5in}
<<echo=FALSE,eval=TRUE,fig=TRUE,width=8,height=8>>=
#while latticeExtra theEconomist.theme is beautiful
#I wanted to stretch my knowledge, so I will start from scratch
#example given to left justify strip
#http://maths.anu.edu.au/~johnm/r-book/xtras/boxcontrol.pdf
stripfun <- function(which.given, which.panel,factor.levels, ...){
grid.rect(name = trellis.grobname("bg", type = "strip"),
gp = gpar(fill = "seashell3", col = "seashell3"))
panel.text(x=0.10, y=0.6,
lab = factor.levels[which.panel[which.given]],
adj=0, font=3, cex=1.3)
}
#heavily stripped and modified theEconomist.axis() from latticeExtra
timely.axis <- function (side = c("top", "bottom", "left", "right"), scales,
components, ..., labels = c("default", "yes", "no"), ticks = c("default",
"yes", "no"), line.col, noleft=TRUE)
{
side <- match.arg(side)
if (side == "top") return()
labels <- match.arg(labels)
ticks <- match.arg(ticks)
if (side %in% c("left", "right")) {
if (side == "right") {
scales$draw=TRUE
labels <- "no"
ticks <- "no"
}
if (side == "left") {
labels <- "yes"
ticks <- "yes"
}
}
axis.default(side, scales = scales, components = components,
..., labels = labels, ticks = ticks, line.col = "black")
if (side == "right" & panel.number()==1) {
comp.list <- components[["right"]]
if (!is.list(comp.list))
comp.list <- components[["left"]]
panel.refline(h = comp.list$ticks$at)
lims <- current.panel.limits()
panel.abline(h = lims$y[1], col = "black")
}
}
#set up ylimits to use for the two scales
ylimits<-c(pretty(c(min(managers.cumul.melt$return),
max(managers.cumul.melt$return)),4),as.numeric(round(last(managers.cumul)[,order(last(managers.cumul))],2)))
ylabels<-c(ylimits[1:(length(ylimits)-3)],colnames(managers)[order(last(managers.cumul))])
returns <- list(
bar = barchart(account~value|period,col=mycolors,data=trailing.melt,
layout=c(1,4),
box.ratio=100,
origin=0,
reference=TRUE,
par.settings=
list(
par.main.text = list(font = 1, cex=1.5, just = "left",x = grid::unit(5, "mm")),
axis.line = list(col = NA)),
scales=list(x=list(
limits=c(0,max(trailing.melt$value)+0.025), #snug labels right up to bars by setting to 0
at=pretty(trailing.melt$value),
labels=paste(round(100*as.numeric(pretty(trailing.melt$value)), 2), "%", sep="")
)),
xlab=NULL,
axis = timely.axis,
strip=stripfun,
strip.left=FALSE,
panel=function(...) {
panel.barchart(...)
tmp <- list(...)
tmp <- data.frame(x=tmp$x, y=tmp$y)
# add text labels
panel.text(x=tmp$x, y=tmp$y,
label=sprintf("%1.2f%%", tmp$x * 100 ),
cex=1, col="black", pos=4)
},
main="Annualized Returns"),
cumulgrowth =
xyplot(return~date,groups=account,data=managers.cumul.melt,
# col=mycolors,
type="l",lwd=3,
xlab=NULL,
ylab=NULL,
par.settings=
list(
par.main.text = list(font = 1, cex=1.5, just = "left",x = grid::unit(5, "mm")),
axis.line = list(col = "transparent"),
superpose.line=list(col=mycolors)), #do this for direct.label
scales=list(x=list(alternating=1,at=index(managers)[endpoints(managers,"years")],
labels=format(index(managers)[endpoints(managers,"years")],"%Y")),
y=list(alternating=3,at=ylimits,labels=ylabels)),
axis=function (side = c("top", "bottom", "left", "right"), scales,
components, ..., labels = c("default", "yes", "no"), ticks = c("default",
"yes", "no"), line.col){
side <- match.arg(side)
labels <- match.arg(labels)
ticks <- match.arg(ticks)
axis.text <- trellis.par.get("axis.text")
if(side == "top") return()
if(side == "right") {
components[["right"]]<-components[["left"]]
components[["right"]]$ticks$at <- components[["right"]]$ticks$at[5:7]
components[["right"]]$labels$at <- components[["right"]]$labels$at[5:7]
components[["right"]]$labels$labels <- components[["right"]]$labels$labels[5:7]
}
if(side %in% c("bottom","right")){
axis.default(side, scales = scales, components = components,
..., labels = labels, ticks = ticks, line.col = axis.text$col)
if (side == "right") {
comp.list <- components[["left"]]
panel.refline(h = comp.list$ticks$at[1:4])
lims <- current.panel.limits()
panel.abline(h = lims$y[1], col = axis.text$col)
comp.list.left<-components[["left"]]
comp.list.left$ticks$at <- components[["left"]]$ticks$at[1:4]
comp.list.left$labels$at <- components[["left"]]$labels$at[1:4]
comp.list.left$labels$labels <- components[["left"]]$labels$labels[1:4]
panel.axis(side="left",at=comp.list.left$ticks$at,outside=TRUE)
}
}
},
main=paste("Cumulative Growth Since Inception ",format(index(managers)[1],"%B %Y"),sep=""))
)
print(returns$cumulgrowth,position=c(0,0.6,1,1),more=TRUE)
print(returns$bar,position=c(0,0,1,0.6))
@
%\end{minipage}
\begin{center}
<<echo=FALSE,eval=TRUE,results=tex>>=
percent <- function(x, digits = 2, format = "f", ...)
{
paste(formatC(100 * x, format = format, digits = digits, ...), "%", sep = "")
}
trailingtable <- apply(trailing,MARGIN=2,FUN=percent)
rownames(trailingtable) <- c("1y","3y","5y",paste("Since Inception ",format(index(managers)[1],"%b %Y")))
print(xtable(trailingtable), floating=FALSE)
@
\end{center}
\end{textblock*}
\end{wide}
\end{figure}
\newpage
\section{Risk}
\end{document}

5 comments:

  1. it is recommended to quote character options now, e.g. results='hide'

    ReplyDelete
    Replies
    1. Thanks Yihui Xie for the helpful comment. Unfortunately, my RStudio/Sweave setup dislikes the quoted character options, so I omitted, but that goes into my next comment in response to alstated...

      Delete
  2. I don't understand fully what is this knitr package. I've installed it, but don't know how to start with its features. I know what is Sweave, which allows me to run R codes on TeX. But what is the difference of Sweave and Knitr? What does knitr can do that sweave can't?

    Any link to address my questions?

    I'm reading now on your site Yihui Xie, I hope I can start to use your package.

    ReplyDelete
    Replies
    1. Thanks alstated. Actually this post did not use knitr, and I was able to spot a lot of the power of knitr through that frustrating process. Yihui can correct me, but I believe knitr's power comes primarily through its http://yihui.name/knitr/options chunk options that give you much more control over the output.

      I also have forked this post to http://timelyportfolio.blogspot.com/2012/05/knitr-performance-report-3-really-with.html

      Delete
    2. This is hard to answer, because there are too many new features in knitr. See https://github.com/yihui/knitr#motivation for a small portion of them.

      If you are a beginner, I recommend you to start with R Markdown first (and RStudio is recommended: http://rstudio.org/docs/authoring/using_markdown); the series of Performance Reports here may be too advanced to beginners :)

      Delete