Rmetrics {Rmetrics} | R Documentation |
A collection and description of functions for
teaching computational finance and financial
engineering.
Some of the Rmetrics packages have become so huge that it is
for many Rmetrics programmers a problem to find out where
to search for a desired function. So w started to split the
biggest packages in smaller ones.
So far Rmetrics (<= R-2.5.1) has packaged functions according to a quite simple subject classification scheme. We have extended it a litte bit, and now this scheme comes with the following topics:
A) DATA SETS AND UTILIY FUNCTIONS
B) CHRONOLOGICAL OBJECTS, FINANCIAL CENTERS AND TIME SERIES
C) ANALYSIS OF FINANCIAL RETURNS AND VOLATILITY
D) TECHNICAL ANALYSIS, DECISION MAKING AND PREDICTION
E) EXTREME VALUE THEORY, COPULAE AND RISK MANAGEMENT
F) VALUATION OF FINANCIAL INSTRUMENTS
G) ASSET MANAGEMENT, PORTFOLIO ANALYSIS AND OPTIMIZATION
We think, this scheme starting with R-2.6.0, will help programmers and developers to search faster around for functions in the Rmetrics packages.
Changes to previous Rmetrics Packages:
Concerning A)
We splitted the package fEcofin
in two packages:
fEcofin
now holds all economic and financial data sets, and
fUtilities
now holds all kinds of utility functions and general tools
which are needed for Rmetrics (most as internal functions).
Concerning B)
We splitted the package fCalendar
now into three
packages:
fCalendar
now holds timeDate
class, zone and DST
information, and everything what has to do with Holidays and
Holiday Calendars,
fSeries
now holds everything about timeSeries
class,
fImport
now holds download functions for several web data
bases, like Yahoo, Economagic, Federal Reserve, and others.
Concerning C)
fBasics
still holds everything about financial returns,
stable and hyperbolic distributions, distributional fits,
stylized facts, and hypothesis tests.
The content of the former fSeries
package has moved
to four new packages
fArma
holds linear and long range time series models,
fractional ARMA and fractional Brownian Nois/Motion,
mainly wrapper functions allowing for timeSeries objects
and providing easy to use,
fGarch
deals with volatility modeling using ARCH, GARCH,
APARCH and related heteroskedastic models, also providing
an interface to OXGarch (Windows only)
fNonlinear
is thought mainly for nonlinearity models,
chaos and nonlinearity tests, e.g. like BDS,
fUnitRoots
deals with time series trends and unit
root testing, implementing MacKinnon's pValues and
interfacing Pfaff's urca package.
Concerning D)
The previous fMultivar package has been splitted now into
four packages:
fMultivar
now contains bivariate and multivariate return
distributions and tools like bivariate binning and gridding,
interpolation,
fTrading
deals with technical analysis, benchmark analysis
and rolling analysis,
fRegression
holds convenient wrappers allowing for
timeSeries objects to be modeled by regression methods
for building trading systems, for decision making and
prediction)
Concering E)
Nothing has changed so far.
fExtremes
deals with GEV and GPD modeling, the extremal
index, and Risk estimation,
fCopulae
has functions for bivariate (only) copulae
including elliptical, archimedean, extreme value, mixed,
and empirical copulae, tail dependency estimators.
Concerning F)
The huge fOptions package has been splitted into three
new packages:
fOptions
holds the option basics, like Black-Scholes,
Heston-Nandi, Binomial Trees, American Option Approximations,
low discrepancy sequences and Monte Carlo valuation of
options,
fExoticOptions
has pricing formulas for dozens of exotic
European options
fAsianOptions
is made for Exponential Brownian Motion for
(exact) pricing formulas of Asian Options including
moment matching methods, Gram-Charlier Series Expansion,
PDE pricing, Laplace Inversion approach, Spectral
Expansion approach, and symmetry relations.
Bonds
has been started up currently with functions for
yield curve modeling.
Concerning G)
The package fPortfolio has been splitted into new
packages:
fAssets
deals with many aspects of asset selection and
asset analysis, including robust statistics, clustering,
correlation analysis, lower partial moments, etc.
fPortfolio
is made for design and optimization of
portfolios using either the standard Markowitz or the
Conditional Value-at-Risk approach, included are
functions for rolling efficient portfolios and frontiers.