In rexmacey/fundAnalysis: Provides Analytics on Mututal Funds and ETFs

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

This package helps analyze mutual funds and ETFs both of which are referred to as funds. There is not much new to this package. Rather it brings together functions from several packages.

At this stage in the package development, the emphasis has been on collecting data and performing calculations. Functions to present the information need to be written.

library(MASS, quietly = TRUE)
library(quantmod, quietly = TRUE)
library(readr, quietly = TRUE)
library(rvest, quietly = TRUE)
library(xts, quietly = TRUE)
library(ggplot2, quietly = TRUE)
library(FactorAnalytics, quietly = TRUE)
library(fundAnalysis, quietly = TRUE)
library(PerformanceAnalytics, quietly = TRUE)

For illustrative purposes data has been loaded for funds with the following symbols.

# Data is provided for the following symbols
print(symbols)

Data installed with package

Some data has been provided with the package to save time. The commands to create the data objects are shown as comments.

Getting the Fama-French 5 factor model data

This data is used to calculate exposures to the Fama-French factors. Future versions might have an option for the 3 factor model.

# ff_datam <- download_FF_5_factor("M")
# ff_datad <- download_FF_5_factor("D")

# Monthly Fama-French data
head(ff_datam)
tail(ff_datam)
# Daily Fama-French data
head(ff_datad)
tail(ff_datam)

Retrieve price information for a set of symbols

The function to retrieve prices uses the quantmod package and returns an xts object with just the adjusted close values. Price information is an intermediate step to calculating returns and for plotting price movements.

#prices_d <- getPrices(symbols,startDate="2012-12-31", freq="D")
#prices_m <- getPrices(symbols,startDate="2012-12-31", freq="M")
# Daily prices for a few of the funds
head(prices_d)[,1:6]
tail(prices_d)[,1:6]
# Monthly prices for a few of the funds
head(prices_m)[,1:6]
tail(prices_m)[,1:6]

Convert prices into returns

The convertPricesToReturns function calls the periodReturn function from quantmod.

# Daily returns of the funds - a list with one item per fund
# First fund
#rets_d <- convertPricesToReturns(prices_d, freq="D")
#rets_m <- convertPricesToReturns(prices_d, freq="M")
names(rets_d[1])
head(rets_d[[1]])
tail(rets_d[[1]])

Summary information scraped from Yahoo Finance - a list with one item for each fund

The scrapeQuoteSummary function returns a list with an item for each symbol containing a character vector of the information found on the Quaote Summary tab of the Yahoo Finance page.

#summary_data <- scrapeQuoteSummary(symbols)
# Information for an ETF
summary_data["FNDB"]
# Information for a mutual fund
summary_data["DFALX"]

LM model for fund exposures to the Fama-French 5 factor model

The ffModelLM function produces a linear regression model for each for each symbol and stores the result in a list. Show are the details for two securities and a table of coefficents for all the symbols.

lm_listd <- ffModelLM(rets_d, ff_datad)

print.mdl_list <- function(lm_list){
    print("Summaries")
    for (i in 1:2){
        mdl <- lm_list[[i]]
        print(names(lm_list[i]))
        print(summary(mdl))
    }
}

print.mdl_list(lm_listd)
print(round(100*coefficients_lm(lm_listd),2))

Stepwise model for fund exposures to the Fama-French 5 factor model

The ffModelLM may return coefficient values not significantly different from zero. Such a model may be overfit. To reduce overfitting, we can run a stepwise regression. The

step_listd <- ffModelStepLM(rets_d,ff_datad)
print.mdl_list(lm_listd)
print(round(100*coefficients_step(step_listd),2))

As an indication of how the stepwise regression simplifies the models which is an indication of less overfitting, there are r sum(coefficients_lm(lm_listd)==0) non-zero coefficients in the models without stepwise regression and r sum(coefficients_step(step_listd)==0) non-zero coefficients with stepwise regression.

More fun with models

R offers built-in functions to see the results of a linear model. Above we saw the summary function. A few more are presented here.

mdl <- step_listd[[1]]
print("Model coefficients")
print(coefficients(mdl)) # just view the coefficients 
print("Confidence interval of the coefficients at a specified level")
print(confint(mdl,level=0.9)) # view the confidence interval of the coefficients
print("The fitted (predicted) values - only first few displayed")
print(head(fitted(mdl))) # show the fitted (predicted) values
print("The residuals (errors) - only first few displayed")
print(head(residuals(mdl)))  # show the residuals (errors)

Style Analysis

Another way to look at the exposures of a fund is via returns-based style analysis (RBSA). RBSA regresses the returns of a fund against the returns of several other funds and constrains the coefficients to sum to 1.0. We mainly use the FactorAnalysis package for this. The rbsa function finds one set of coefficients using all the data.

r.fund <- rets_d[[1]]
colnames(r.fund)<-names(rets_d)[1]
r.style <- merge.xts(rets_d[[10]],rets_d[[11]],rets_d[[13]],rets_d[[14]],rets_d[[23]],rets_d[[15]],rets_d[[16]])
colnames(r.style)<-names(rets_d)[c(10,11,13,14,23,15,16)]
RBSA_1 <- rbsa(r.fund,r.style)
print(RBSA_1)

RBSA - rolling window

RBSA_window <- rbsa_rolling(r.fund,r.style,method="constrained", width = 63)
head(RBSA_window$weights)
summary(RBSA_window$weights)
autoplot(RBSA_window$weights, main = "RBSA for FNDB")
boxplot(as.data.frame(RBSA_window$weights), col="blue", main="Style Weights for FNDB", ylim=c(0,1))

Standard performance measurement

This is mostly from the PerformanceAnalytics package.

table.AnnualizedReturns(cbind(r.fund,r.style))
table.TrailingPeriods(r.fund)
table.Stats(cbind(r.fund,r.style))
table.Drawdowns(r.fund)
table.Variability(cbind(r.fund,r.style))
table.Correlation(r.fund,r.style)
TrackingError(r.fund,r.style)*sqrt(252)

---

Below was automatically inserted when this vignette was created and kept for instructional purposes.

Figures

The figure sizes have been customised so that you can easily put two images side-by-side.

plot(1:10)
plot(10:1)

You can enable figure captions by fig_caption: yes in YAML:

output:
  rmarkdown::html_vignette:
    fig_caption: yes

Then you can use the chunk option fig.cap = "Your figure caption." in knitr.

More Examples

You can write math expressions, e.g. $Y = X\beta + \epsilon$, footnotes^[A footnote here.], and tables, e.g. using knitr::kable().

knitr::kable(head(mtcars, 10))

Also a quote using >:

"He who gives up [code] safety for [code] speed deserves neither." (via)

rexmacey/fundAnalysis documentation built on Dec. 2, 2019, 5:50 p.m.