Comparing Conventional and Artificial Neural Network Models for the Pricing of Options

The pricing of options on futures is compared using conventional models and artificial neural networks. This work demonstrates superior pricing accuracy using the artificial neural networks in an important subset of the input parameter set.

Using Neural Networks to Discover Patterns in International Equity Markets

The use of neural networks represents a new approach to how this type of problem can be investigated. The economics and finance literature is full of studies that require the researcher to prespecify the exact nature of the relationship and select specific variables to test. In this study, we use a multistage approach that requires no prespecification of the model and allows us to look for associations and relationships that may not have been considered. Previous studies have been limited by the nature of statistical tools, which require the researcher to determine the variables, time frame, and markets to test. An intelligent guess may lead to the desired outcome, but neural networks are used to produce a more thorough analysis of the data, thus improving the researcher’s ability to uncover unanticipated relationships and associations.

Credit Scoring Using Supervised and Unsupervised Neural Networks

Some of the concerns that plague developers of neural network decision support systems include: (a) How do I understand the underlying structure of the problem domain; (b) How can I discover unknown imperfections in the data which might detract from the generalization accuracy of the neural network model; and (c) What variables should I include to obtain the best generalization properties in the neural network model? In this paper we explore the combined use of unsupervised and supervised neural networks to address these concerns. We develop and test a credit-scoring application using a self-organizing map and a multilayered feedforward neural network. The final product is a neural network decision support system that facilitates subprime lending and is flexible and adaptive to the needs of e-commerce applications.

Segmentation of the Portuguese Clients of Pousadas de Portugal

The aim of our work is to perform a market segmentation of the clients of Pousadas de Portugal, a network for over 40 high-end small hotels, ENATUR. The data for this work was provided by a sample of more than 2500 clients that filled in a given questionnaire. The segmentation is based on how often the clients used the hotels, and on the type of stay they were seeking. A few different techniques were used: mixed approaches using a-priori constitution of clusters and/or neural nets (SOM – Self-Organizing Maps) and/or k-means. Profiling the obtained segments adds some new insights about the clients and helps ENATUR managers to better support new marketing decisions.

A Neural Network Application to Identify High-Value Customers for a Large Retail Store in Japan

The data mining activities studied in this chapter concern the early identification of potential high-value customers. Member stores can use this information to establish a close relationship with this select group of customers, thus reducing the chances of losing them. Traditionally, Japanese drugstore chains, unlike their American counterparts, have enjoyed close ties with their customers. For example, cash register clerks at Pharma stores may have substantial interactions with customers using online market research questionnaire forms. By closely monitoring the purchasing behavior of relatively new visitors to the store and applying data mining tools to pertinent data, the company can provide decision support to clerks and to the marketing department to aid in relationship building. For instance, sales campaign information, customized coupons, and free samples can be directly mailed to members of the targeted group or given to them at the checkout counter.

Using Neural Networks to Model Premium Price Sensitivity of Automobile Insurance Customer

This paper describes a neural network modelling approach to premium price sensitivity of insurance policy holders. Clustering is used to classify policy holders into homogeneous risk groups. Within each cluster a neural network is then used to predict retention rates given demographic and policy information, including the premium change from one year to the next. It is shown that the prediction results are significantly improved by further dividing each cluster according to premium change. This work is part of a larger data mining framework proposed to determine optimal premium prices in a data-driven manner.

Neural Networks for Target Selection in Direct Marketing

The outline of the chapter is as follows. The section on direct marketing explains briefly what it is and discusses the target selection problem in direct marketing. Target selection for a charity organization is also explained. The next section discusses how neural networks can be used for building target selection models that a charity organization can use. The section on data preparation considers how the actual data for training the neural networks is obtained from the organization’s database. The actual model building steps are explained in the following section. The results of the neural network models are discussed afterwards, followed by a comparison of the results with some other target selection methods. Finally, the chapter concludes with a short discussion.

Neural Networks for Technical Forecasting of Foreign Exchange Rates

This chapter describes the application of neural networks in foreign exchange rate forecasting between American dollar and five other major currencies: Japanese yen, Deutsch mark, British pound, Swiss franc and Australian dollar. Technical indicators and time series data are fed to neural networks to mine, or discover, the underlying “rules” of the movement in currency exchange rates. The results presented in this chapter show that without the use of extensive market data or knowledge, useful prediction can be made and significant paper profit can be achieved for out-of-sample data with simple technical indicators. The neural-network-based forecasting is also shown to compare favorably with the traditional statistical approach.

Corporate Strategy and Wealth Creation

Corporate managers, business consultants, stock analysts, and academic researchers have long maintained that the strategic decisions of managers have a direct influence on firm performance. Although societal and economic trends, industry characteristics, and chance all influence performance, the strategic decisions made by managers are believed to play a decisive role in shaping financial performance. Even so, researchers investigating this relationship have reported largely ambiguous results (Rumelt, 1974; Ramanujam and Varadarajan, 1989; Hoskisson and Hitt, 1990; Robbins and Pearce, 1992; Markides and Williamson, 1994; Barker, 1994). Furthermore, attempts by analysts to forecast future financial performance by scrutinizing current strategy decisions have been plagued with problems. Can firm financial performance be predicted with accuracy from the corporate strategy decisions of the executive management team?

Combining Supervised and Unsupervised Neural Networks for Improved Cash Flow Forecasting

This chapter examines the use of neural networks as both a technique for pre-processing data and forecasting cash flow in the daily operations of a financial services company. The problem is to forecast the date when issued cheques will be presented by customers, so that the daily cash flow requirements can be forecast. These forecasts can then be used to ensure that appropriate levels of funds are kept in the company’s bank account to avoid overdraft charges or unnecessary use of investment funds. The company currently employs an ad-hoc manual method for determining cash flow forecasts and is keen to improve the accuracy of the forecasts. Unsupervised neural networks are used to cluster the cheques into more homogeneous groups prior to supervised neural networks being applied to arrive at a forecast for the date each cheque will be presented. Accuracy results are compared to the existing method of the company, together with regression and a heuristic method.