Mark T. Leung

Application of neural networks to an emerging financial market: forecasting and trading the Taiwan stock index

In this study, we attempt to model and predict the direction of return on market index of the Taiwan Stock Exchange, one of the fastest growing financial exchanges in developing Asian countries. Our motivation is based on the notion that trading strategies guided by forecasts of the direction of price movement may be more effective and lead to higher profits. The probabilistic neural network (PNN) is used to forecast the direction of index return after it is trained by historical data. Statistical performance of the PNN forecasts are measured and compared with that of the generalized methods of moments (GMM) with Kalman filter. Moreover, the forecasts are applied to various index trading strategies, of which the performances are compared with those generated by the buy-and-hold strategy as well as the investment strategies guided by forecasts estimated by the random walk model and the parametric GMM models. Empirical results show that the PNN-based investment strategies obtain higher returns than other investment strategies examined in this study. Influences of length of investment horizon and commission rate are also considered.

Forecasting Stock Indices: A Comparison of Classification and Level Estimation Models

Despite abundant research which focuses on estimating the level of return on stock market index, there is a lack of studies examining the predictability of the direction/sign of stock index movement. Given the notion that a prediction with little forecast error does not necessarily translate into capital gain, we evaluate the efficacy of several multivariate classification techniques relative to a group of level estimation approaches. Specifically, we conduct time series comparisons between the two types of models on the basis of forecast performance and investment return. The tested classification models, which predict direction based on probability, include linear discriminant analysis, logit, probit, and probabilistic neural network. On the other hand, the level estimation counterparts, which forecast the level, are exponential smoothing, multivariate transfer function, vector autoregression with Kalman filter, and multilayered feedforward neural network. Our comparative study also measures the relative strength of these models with respect to the trading profit generated by their forecasts. To facilitate more effective trading, we develop a set of threshold trading rules driven by the probabilities estimated by the classification models. Empirical experimentation suggests that the classification models outperform the level estimation models in terms of predicting the direction of the stock market movement and maximizing returns from investment trading. Further, investment returns are enhanced by the adoption of the threshold trading rules.

Forecasting exchange rates using general regression neural networks

Predicting currency movements has always been a problematic task as most conventional econometric models are not able to forecast exchange rates with significantly higher accuracy than a naive random walk model. For large multinational firms which conduct substantial currency transfers in the course of business, being able to accurately forecast the movements of exchange rates can result in considerable improvement in the overall profitability of the firm. In this study, we apply the General Regression Neural Network (GRNN) to predict the monthly exchange rates of three currencies, British pound, Canadian dollar, and Japanese yen. Our empirical experiment shows that the performance of GRNN is better than other neural network and econometric techniques included in this study. The results demonstrate the predictive strength of GRNN and its potential for solving financial forecasting problems.

Regression neural network for error correction in foreign exchange forecasting and trading

Predicting exchange rates has long been a concern in international finance as most standard econometric methods are unable to produce significantly better forecasts than the random walk model. Recent studies provide some evidence for the ability of using multivariate time series models to generate better forecasts. At the same time, artificial neural networks have been emerging as alternatives to predict exchange rates. In this paper, we propose an adaptive forecasting approach which combines the strengths of neural networks and multivariate econometric models. This hybrid approach contains two forecasting stages. In the first stage, a time series model generates estimates of the exchange rates. In the second stage, General Regression Neural Network is used to correct the errors of the estimates. A number of tests and statistical measures are then applied to compare the performances of the two-stage models (with error-correction by neural network) with those of the single-stage models (without error-correction by neural network). Both empirical and trading simulation experiments suggest that the proposed hybrid approach not only produces better exchange rate forecasts but also results in higher investment returns than the single-stage models. The effect of risk aversion in currency trading is also considered.

Stock auction bidding behavior and information asymmetries: An empirical analysis using the discriminatory auction model framework

Abstract This paper empirically analyzes bidding behavior and information asymmetries of stock auctions using a discriminatory auction model framework. Analyzing stock auctions using auction theory is important because it provides a logical framework for explaining observable behaviors and a solid foundation for empirical testing. Because of limited availability of stock auction data, existing empirical research based on auction theory focus mainly on treasury auctions. Away from the treasury markets, however, there is a significant paucity of literature on this subject. In this study, we make use of a detailed stock auction data set from the Taiwan Stock Exchange to empirically examine the behaviors of the Taiwan stock auction market within the framework of the auction theory. Our results show that the level of competition, the dispersion of opinion among bidders, and the bidder’s risk aversion are significant in determining auction prices. Results also show that if the offering prices are set equal to the average offering prices, the first-day post-initial public offering abnormal return will be equal to zero. Additionally, we find institutional bidders possess superior bidding skills compared to small bidders and that underwriter’s characteristics influence the bidding results.

A Bayesian vector error correction model for forecasting exchange rates

This paper develops a new method called Bayesian Vector Error Correction Model (BVECM), which is applied to forecast 1 month ahead changes of currency exchange rates for three major Asia Pacific economies. The study also compares out-of-sample forecasting performance with those of the random walk model and the Bayesian Vector Autoregression (BVAR), which has been shown in recent studies to outperform a variety competing of econometric techniques in exchange rate forecasting. Our experimental results indicate that both BVECM and BVAR are able to forecast the changes in exchange rates better than the random walk model. In terms of conventional forecast evaluation statistics, BVECM outperforms BVAR for all three currencies examined. In addition, the bias tests find that BVECM produces systematically less biased and more efficient out-of-sample forecasts than BVAR. Although the results of market timing tests indicate that both BVAR and BVECM have an economically significant value in predicting the directional change in two of the three exchange rates, BVECM is shown to produce equally or more economically significant directional change forecasts than BVAR.

Corrective maintenance through dynamic work allocation and pre-emption: case study and application

This case study develops an innovative management and scheduling system for corrective maintenance of machines in a manufacturing facility. The study also involves a comparative evaluation of the proposed and the existing systems under a spectrum of operating conditions. A comprehensive simulation is used to evaluate system performances under a variety of settings which include reliability, service level, and cost consequences. The analysis is based on a full factorial experimental design. In summary, the developed self-regulating management system which involves dynamic work allocation and pre-emption is shown to yield higher machine availability and higher mechanic utilisation even with fewer mechanics. The study also finds that the new system is more streamlined, agile, and robust although it is subject to more-constrained machine reliability and mechanic service time environments. Further, a major reduction of current manpower can still achieve at least 95% machine availability, illustrating the cost effectiveness and efficacy of the developed system. This rule-based corrective maintenance system can be operated in uncertain environments on a real-time basis without additional reformatting costs and provides a competitive measure to deal with managerial issues such as low retention rate for skilled mechanics, highly uneven training levels and pay scales. The financial consequences and gains in strategic advantage with respect to the facilitys operational structure are promising after implementation. Moreover, the system developed in this case study represents a meaningful starting point for a more vigorous theoretical research on the bucket brigade system to different functions in industrial and operations management.

Finite mixture partial least squares for segmentation and behavioral characterization of auction bidders

The purpose of this study is to demonstrate how to empirically segment, without a priori knowledge, online auction bidders using experimental data and finite mixture models. The proposed method utilizes a finite mixture partial least squares (FIMIX-PLS) approach to examine bidder behaviors and personality characteristics, evaluate bidder differences, and then segment the bidders. The empirical experiment is conducted for two different auction mechanisms - English and Vickrey. Results from both auction mechanisms indicate that FIMIX-PLS is capable of profiling and segmenting the bidders based on their individual characteristics. The post hoc analysis confirms the segmentation scheme and the capability of FIMIX-PLS in segmenting bidders into statistically identifiable homogeneous groups without a priori information of group characteristics. Such advantage is practical for online businesses dealing with increasing amount of data about their customers on a real time basis.

Adaptive exponential smoothing versus conventional approaches for lumpy demand forecasting: case of production planning for a manufacturing line

Production planning in a lumpy demand environment can be tenuous, with potentially costly forecasting errors. This paper addresses the issue of selecting the smoothing factor used in lumpy demand forecasting models. We propose a simple adaptive smoothing approach to replace the conventional industrial practice of choosing a smoothing factor largely based on the analyst or engineers experience and subjective judgment. The Kalman filter approach developed in this study processes measurements to estimate the state of a linear system and utilises knowledge from states of measurements and system dynamics. Performances of an array of forecasting models that have been shown to work well in lumpy demand environments are compared with respect to the proposed adaptive smoothing factor and the conventional smoothing constant across a spectrum of lumpy demand scenarios. All models using the adaptive smoothing factor based on Kalman filter weighting function generate smaller errors than their conventional counterparts, especially under high lumpiness demand environments. Our proposed approach is particularly useful when production management is concerned about simplicity and transferability of knowledge due to constant personnel turnaround and low retention rate of expertise.

Application of polynomial projection ensembles to hedging crude oil commodity risk

Although the rapid expansion in derivative market in previous decades has drawn research in both theory and practice of hedging against commodity risk, recent volatile fluctuations in crude oil prices in world market have renewed profound interest in examination of existing and development of new hedging models and strategies. In this paper, we propose and develop a methodological framework for applying individual and ensembles of polynomial projection models to hedge against oil commodity price risk. The study also comparatively evaluates the hedging performances of these projection models and benchmarks them against naive hedging, VEC-GARCH model, and the case of no hedging. In addition, the empirical analysis considers a traders level of risk aversion in commodity hedging as well as the adoption of transaction cost. Our findings indicate promising out-of-sample hedging capability by polynomial projection models. Also, different forms of integrated ensembles of projections outperform individual polynomial projections, suggesting the usefulness of ensemble structure in enhancement of hedging in an uncertain environment.