This book
aims to provide practically useful models and methods
for quantitative analysis of financial asset prices,
construction of various portfolios, and computer-assisted trading
systems. In particular, this book will be beneficial
for
1) "quants" (quantitatively-inclined analysts) in financial industries,
2) financial
engineers in investment banks, securities companies,
derivative-trading companies, software houses, etc.,
who are developing portfolio trading systems,
3) graduate
students and specialists in the areas of finance,
business, economics, statistics, financial engineering
and
4) investors
who are interested in Japanese financial markets.
Throughout the book a great emphasis
is put on the practical use usefulness of models and
methods for investment decision-making, and some
examples are demonstrated with practical analyses and
models on Japanese financial markets. The statistical
procedures treated in this book are somewhat advanced
because our financial world is of many complexities.
In fact, the time series variations of financial asset
prices are of the complicated features;
(a) they are multi-dimensional
phenomena,
(b) their variational
structure is evolving over time and
(c) profitable structure
of variations will be eventually exploited.
Hence to make
effective portfolios, it will be important to
empirically model the evolving multi-dimensional
variations by a proper multivariate time series model
and to manage portfolios constructed from the model
with proper trading (rebalancing) rules for profitable
investment. Therefore our methods and models for
analysis of prices, construction of portfolios and
management of portfolios are closely associated with
various multivariate time series methods in
statistics. In particular, the MTV ( multivariate time
series variance component ) model the author (1987)
proposed plays an important role as a basic
information-summarizing and result-stabilizing model
in many ways. In this book we respect the MPT (modern
portfolio theory) to the extent that the models
derived from the MPT perform well empirically as
statistical models for investment decision making. In
other words, from our viewpoint of practical
usefulness we consider it more important to extract by
statistically scientific methods significant and
useful information for investments from available date
than to follow the MPT or traditional approach is
fully discussed in Chapter 1. The book consists of the
following three parts:
Part
I Quantitative models for portfolio analysis,
Part
II
Quantitative asset-allocation systems,
Part
III Statistical approaches to option
pricing and bond pricing,
In general,
the procedure of making and managing a portfolio
consists of the following five elements:
(1)
Selection
of investment stances.
(2) Selection
of a population of financial assets for investment.
(3) Modeling
and predicting future prices of financial assets.
(4)
Construction of an optimal portfolio.
(5) Choice of
a trading (rebalance) rule for managing the portfolio.
The most
important element in this procedure will be (3)
because an investment is a commitment to uncertainties
in future and a better prediction leads us to a better
performance. In Part I, we introduce basic and
practical models and methods for analyzing and
predicting financial asset prices. In particular,
univariate nonlinear models and multivariate time
series models are focused upon. In addition, we
observe some empirical features of returns on stocks
and exchange rates, which can be taken into account in
selecting ore screening assets for a portfolio
population in (2) and forming an optimal portfolio in
(4). One of the features is nonlinearity for a time
series process of (one-period) returns and it will
provide profitable information in option trading. In
Part II, multivariate time series models introduced in
Part I are used to develop various procedures of
making portfolios. Above all, we propose some new
procedures such as the procedures of making predictive
MTV market portfolios, MTV index portfolios, MTV
canonical correlation portfolios, predictive
classification portfolios, etc. Also some procedures
associated with the APT (arbitrage pricing theory) and
Rosenbergfs models are introduced with some
practices in the Japanese financial industries. In
Part III, practical option pricing is discussed, where
the variational features observed in Part I are taken
into account. A review and development is also made on
the Black-Scholes type option theory. Some new bond
pricing models are also presented in Part III.
