Thuy Duong To

Computational Intelligence for Evolving Trading Rules

This paper describes an adaptive computational intelligence system for learning trading rules. The trading rules are represented using a fuzzy logic rule base, and using an artificial evolutionary process the system learns to form rules that can perform well in dynamic market conditions. A comprehensive analysis of the results of applying the system for portfolio construction using portfolio evaluation tools widely accepted by both the financial industry and academia is provided.

The Return-Volatility Relation in Commodity Futures Markets

By employing a continuous time multi‐factor stochastic volatility model, the dynamic relation between returns and volatility in the commodity futures markets is analyzed. The model is estimated by using an extensive database of gold and crude oil futures and futures options. A positive relation in the gold futures market and a negative relation in the crude oil futures market subsist, especially over periods of high volatility principally driven by market‐wide shocks. The opposite relation holds over quiet periods typically driven by commodity‐specific effects. According to the proposed convenience yield effect, normal (inverted) commodity futures markets entail a negative (positive) relation.

Humps in the Volatility Structure of the Crude Oil Futures Market: New Evidence

This paper analyzes the volatility structure of commodity derivatives markets. The model encompasses stochastic volatility that may be unspanned by futures contracts. A generalized hump-shaped volatility specification is assumed that entails a finite-dimensional affine model for the commodity futures curve and quasi-analytical prices for options on commodity futures. An empirical study of the crude oil futures volatility structure is carried out using an extensive database of futures prices as well as futures option prices spanning 21 years. The study supports a hump-shaped, partially spanned stochastic volatility specification. Factor hedging, which takes into account shocks to both the volatility processes and the futures curve, depicts the presence of unspanned components in the volatility of commodity futures and the outperformance of the hump-shaped volatility in comparison to the more popular exponential decaying volatility. This hump shaped feature is more pronounced when the market is volatile.

Stochastic Correlation and Risk Premia in Term Structure Models

This paper proposes and analyses a term structure model that allows for both stochastic correlation between underlying factors and an extended market price of risk specification. The issues of invariant transformation and different normalization are then considered so that a comparison between different restrictions can be made. We show that significant improvement in bond fitting is obtained by both allowing the market price of risk to have an extended affine form, and allowing the correlation between underlying factors to be stochastic as well as of variable sign. The overall model fit is more negatively impacted by the restriction on the market price of risk than the restriction of correlated factors. However, the stochastic correlation is priced significantly by market participants, though its impact on the risk premia reduces gradually as time to maturity increases. In addition, stochastic correlation is vital in obtaining good hedged portfolio positions. Certainly, the best hedged portfolio is the one that is built based on the model that takes into account both stochastic correlation and extended market price of risk.

A Computational Intelligence Portfolio Construction System for Equity Market Trading

This paper describes an adaptive computational intelligence system for learning trading rules used in equity market trading. The rules are represented using fuzzy logic, an evolutionary process facilitates the learning process. By controlling the evolutionary process and through selection of training data the trading rules are adapted to market conditions. Results of the systems performance are obtained using historical data from the Australian stock exchange (ASX).

The performance of an adaptive portfolio management system

This paper describes the operation and performance of a computational intelligence rule-base system that manages a portfolio of stocks according to investment objectives. We present an overview of several improvements to the system presented in previous papers and provide detailed results from applying the system in representative scenarios toward determining the robustness of the approach.

Abnormal Returns After Large Stock Price Changes: Evidence from Asia-Pacific Markets

This chapter aims to expand the overreaction literature by examining whether the price reversals occur in the short-term period (i.e., 3 days) and long-term period (i.e., up to 20 days), following large 1-day price changes in Asia-Pacific markets over the period 2001–2005. Our results based on firm data in three Asia-Pacific markets, namely, Australia, Japan, and Vietnam, and static and dynamic measures of large price changes indicate the followings. First, stock prices tend to reverse over the short-term period after large price changes. Second, in the case of large price declines defined by arbitrary trigger values, investors may earn profit from exploiting the phenomena of price reversals; however, the profit is not large enough to exploit since it is less than the profit from passive funds. This result is supportive of the weak form of efficient market hypothesis. Third, we find mixed evidence of long run price reversal across markets. Forth, market conditions (i.e., bear or bull) may not explain the magnitude of price reversals. Finally, the dynamic measures of large price changes based on individual firms provide more consistent evidence across markets, which is supportive of short-term price reversals and overreaction hypothesis. This evidence exists in the emerging market of Vietnam as well as developed Australian and Japanese markets.

The Multifactor Nature of the Volatility of the Eurodollar Futures Market

This paper seeks to estimate a multifactor volatility model so as to describe the dynamics of interest rate markets, using data from the highly liquid but short term futures markets. The difficult problem of estimating such multifactor models is resolved by using a genetic algorithm to carry out the optimization procedure. The ability to successfully estimate a multifactor volatility model also eliminates the need to include a jump component, the existence of which would create difficulties in the practical use of interest rate models, such as pricing options or producing forecasts.

Evolving Trading Rules

Summary. This chapter describes a computational intelligence system for portfolio management and provides a comparison of the relative performance of portfolios of managed by the system using stocks selected from the ASX (Australian Stock Exchange). The core of the system is the development of trading rules to guide portfolio management. The rules the system develops are adapted to dynamic market conditions. An integrated process for stock selection and portfolio management enables a search specification that produces highly adaptive and dynamic rule bases. Rule base solutions are represented using fuzzy logic and an evolutionary process facilitates a search for high performing fuzzy rule bases. Performance is defined using a novel evaluation function involving simulated trading on a recent historical data window. The system is readily extensible in terms of the information set used to construct rules, however to produce the results given in this chapter information derived only from price and volume history of stock prices was used. The approach is essentially referred to as technical analysis – as opposed to using information from outside the market such as fundamental accounting and macroeconomic data. A set of possible technical indicators commonly used by traders forms the basis for rule construction. The fuzzy rule base representation enables intuitive natural language interpretation of trading signals and implies a search space of possible rules that corresponds to trading rules a human trader could construct. An example of a typical technical �

Optimal Factor Strategy in FX Markets

We construct mean-variance optimized currency portfolios and analyze the time- series variation of the conditional Sharpe ratio. Returns, volatility and skewness are predictable. Market timing – i.e., trading more (less) aggressively when the conditional risk-return trade-off is more (less) favorable – significantly increases the unconditional Sharpe ratio from 0.72 to 1.21, improves the skewness of the monthly return distribution from -0.79 to +0.89, and reduces the downside risk from 8.68% to 1.57% maximum loss per 1% expected excess return. Thus, restricting risk taking, i.e., prohibiting market timing, is costly. Understanding and quantifying these costs is important when considering constraints in asset allocations.