Dissertations / Theses on the topic 'Price indexes – Data processing'

Thesis (Ph.D.)--University of Mississippi, 2008.<br>Typescript. Vita. "August 2008." Committee chair : Walter Mayer Includes bibliographical references (leaves 82-83). Also available online via ProQuest to authorized users.

Thesis (M.S.)--Worcester Polytechnic Institute.<br>Keywords: comprehensiblity; technical analysis; genetic programming; overfitting; cooperative coevolution. Includes bibliographical references (p. 82-87).

Ce travail porte sur l'analyse d'images numériques biologiques. Il vise à définir et mettre en œuvre des processus de mesures automatiques de données biologiques à partir d'images numériques dans un cadre de traitement de masse, et aborde notamment : l'incidence des choix méthodologiques sur la stabilité des résultats, l'étude de la validation des mesures produites et les limites de la généricité des méthodes et modèles appliquées à la biologie végétale.La réflexion est menée dans le cadre de l'étude de certaines organisations cellulaires, et plus particulièrement de l'identification et l'analyse automatique de files cellulaires dans des mosaïques d'images microscopiques de bois. En effet, l'étude des tendances biologiques le long de ces structures est nécessaire pour comprendre la mise en place des différentes organisations et maturations de cellule. Elle ne peut être conduite qu'à partir d'une grande zone d'observation du plan ligneux. Pour cela,- nous avons mis en place un nouveau protocole de préparation (rondelles de bois poncées) et de numérisation des échantillons permettant d'acquérir entièrement la zone d'observation sans biais- nous avons développé une chaîne de traitement permettant l'extraction automatique des files cellulaires dans des mosaïques images numériques.- nous avons proposé des indices de fiabilité pour chaque mesure effectuée afin de mieux cibler les études statistiques à venir.Les méthodes développées dans la thèse permettent l'acquisition et le traitement rapide d'un volume important de données. Ces données devraient servir de base à de nombreuses investigations : des analyses architecturales des arbres avec le suivi de file cellulaire et/ou la détection de perturbations biologiques, des analyses de variabilité intra et inter arbres permettant de mieux comprendre la croissance endogène des arbres<br>This study focuses on biological numeric picture processes. It aims to define and implement new automated measurements at large scale analysis. Moreover, this thesis addresses: The incidence of the proposed methodology on the results reliability measurements accuracy definition and analysis proposed approaches reproducibility limits when applied to plant biology.This work is part of cells organization study, and aims to automatically identify and analyze the cell lines in microscopic mosaic wood slice pictures. Indeed, the study of biological tendencies among the cells lines is necessary to understand the cell migration and organization. Such a study can only be realized from a huge zone of observation of wood plane. To this end, this work proposes:- a new protocol of preparation (slices of sanded wood) and of digitizing of samples, in order to acquire the entire zone of observation without bias,- a novel processing chain that permit the automated cell lines extraction in numeric mosaic pictures,- reliability indexes for each measurement for further efficient statistical analysis.The methods developed during this thesis enable to acquire and treat rapidly an important volume of information. Those data define the basis of numerous investigations, such as tree architectural analysis cell lines following and/or detection of biological perturbations. And it finally helps the analysis of the variability intra- or inter- trees, in order to better understand the tree endogenous growth

A importância de se considerar os volumes na análise dos movimentos de preços de ações pode ser considerada uma prática bastante aceita na área financeira. No entanto, quando se olha para a produção científica realizada neste campo, ainda não é possível encontrar um modelo unificado que inclua os volumes e as variações de preços para fins de análise de preços de ações. Neste trabalho é apresentado um modelo computacional que pode preencher esta lacuna, propondo um novo índice para analisar o preço das ações com base em seus históricos de preços e volumes negociados. O objetivo do modelo é o de estimar as atuais proporções do volume total de papéis negociados no mercado de uma ação (free float) distribuídos de acordo com os seus respectivos preços passados de compra. Para atingir esse objetivo, foi feito uso da modelagem dinâmica financeira aplicada a dados reais da bolsa de valores de São Paulo (Bovespa) e também a dados simulados por meio de um modelo de livro de ordens (order book). O valor do índice varia de acordo com a diferença entre a atual porcentagem do total de papéis existentes no mercado que foram comprados no passado a um preço maior do que o preço atual da ação e a sua respectiva contrapartida, que seria a atual porcentagem de papéis existentes no mercado que foram comprados no passado a um preço menor do que o preço atual da ação. Apesar de o modelo poder ser considerado matematicamente bastante simples, o mesmo foi capaz de melhorar significativamente a performance financeira de agentes operando com dados do mercado real e com dados simulados, o que contribui para demonstrar a sua racionalidade e a sua aplicabilidade. Baseados nos resultados obtidos, e também na lógica bastante intuitiva que está por trás deste modelo, acredita-se que o índice aqui proposto pode ser bastante útil na tarefa de ajudar os investidores a definir intervalos ideais para compra e venda de ações no mercado financeiro.<br>The importance of considering the volumes to analyze stock prices movements can be considered as a well-accepted practice in the financial area. However, when we look at the scientific production in this field, we still cannot find a unified model that includes volume and price variations for stock prices assessment purposes. In this paper we present a computer model that could fulfill this gap, proposing a new index to evaluate stock prices based on their historical prices and volumes traded. The aim of the model is to estimate the current proportions of the total volume of shares available in the market from a stock distributed according with their respective prices traded in the past. In order to do so, we made use of dynamic financial modeling and applied it to real financial data from the Sao Paulo Stock Exchange (Bovespa) and also to simulated data which was generated trough an order book model. The value of our index varies based on the difference between the current proportion of shares traded in the past for a price above the current price of the stock and its respective counterpart, which would be the proportion of shares traded in the past for a price below the current price of the stock. Besides the model can be considered mathematically very simple, it was able to improve significantly the financial performance of agents operating with real market data and with simulated data, which contributes to demonstrate its rationale and its applicability. Based on the results obtained, and also on the very intuitive logic of our model, we believe that the index proposed here can be very useful to help investors on the activity of determining ideal price ranges for buying and selling stocks in the financial market.

This thesis explores whether scanner data can be used to inform Consumer Price Index (CPI) construction, with particular reference to the issues of substitution bias and choice of aggregation dimensions. The potential costs and benefits of using scanner data are reviewed. Existing estimates of substitution bias are found to show considerable variation. An Australian scanner data set is used to estimate substitution bias for six different aggregation methods and for fixed base and superlative indexes. Direct and chained indexes are also calculated. Estimates of substitution bias are found to be highly sensitive to both the method of aggregation used and whether direct or chained indexes were used. The ILO (2004) recommends the use of dissimilarity indexes to determine the issue of when to chain. This thesis provides the first empirical study of dissimilarity indexes in this context. The results indicate that dissimilarity indexes may not be sufficient to resolve the issue. A Constant Elasticity of Substitution (CES) index provides an approximate estimate of substitution-bias-free price change, without the need for current period expenditure weights. However, an elasticity parameter is needed. Two methods, referred to as the algebraic and econometric methods, were used to estimate the elasticity parameter. The econometric approach involved the estimation of a system of equations proposed by Diewert (2002a). This system has not been estimated previously. The results show a relatively high level of substitution at the elementary aggregate level, which supports the use a Jevons index, rather than Carli or Dutot indexes, at this level. Elasticity parameter estimates were found to vary considerably across time, and statistical testing showed that elasticity parameter estimates were significantly different across estimation methods. Aggregation is an extremely important issue in the compilation of the CPI. However, little information exists about 'appropriate' aggregation methods. Aggregation is typically recommended over 'homogenous' units. An hedonic framework is used to test for item homogeneity across four supermarket chains and across all stores within each chain. This is a novel approach. The results show that treating the same good as homogenous across stores which belong to the same chain may be recommended.

by Ng Wai-Leung Anthony.<br>Thesis (M.Phil.)--Chinese University of Hong Kong, 1999.<br>Includes bibliographical references (leaves 83-87).<br>Abstracts in English and Chinese.<br>Chapter I. --- INTRODUCTION --- p.1<br>Chapter II. --- LITERATURE REVIEW --- p.5<br>Chapter 2.1 --- The Importance of Stock Index and Index Futures --- p.6<br>Chapter 2.2 --- Importance of Index Forecasting --- p.6<br>Chapter 2.3 --- Reasons for the Lead-Lag Relationship between Stock and Futures Markets --- p.9<br>Chapter 2.4 --- Importance of the lead-lag relationship --- p.10<br>Chapter 2.5 --- Some Empirical Findings of the Lead-Lag Relationship --- p.10<br>Chapter 2.6 --- New Approach to Financial Forecasting - Artificial Neural Network --- p.12<br>Chapter 2.7 --- Artificial Neural Network Architecture --- p.14<br>Chapter 2.8 --- Evidence on the Employment of ANN in Financial Analysis --- p.20<br>Chapter 2.9 --- Hong Kong Securities and Futures Markets --- p.25<br>Chapter III. --- GENERAL GUIDELINE IN DESIGNING AN ARTIFICIAL NEURAL NETWORK FORECASTING MODEL --- p.28<br>Chapter 3.1 --- Procedure for using Artificial Neural Network --- p.29<br>Chapter IV. --- METHODOLOGY --- p.37<br>Chapter 4.1 --- ADF Test for Unit Root --- p.38<br>Chapter 4.2 --- "Error Correction Model, Error Correction Model with Short- term Dynamics, and ANN Models for Comparisons" --- p.38<br>Chapter 4.3 --- Comparison Criteria of Different Models --- p.39<br>Chapter 4.4 --- Data Analysis --- p.39<br>Chapter 4.5 --- Data Manipulations --- p.41<br>Chapter V. --- RESULTS --- p.42<br>Chapter 5.1 --- The Resulting Models --- p.42<br>Chapter 5.2 --- The Prediction Power among the Models --- p.45<br>Chapter 5.3 --- ANN Model of Input Variable Selection Using Contribution Factor --- p.46<br>Chapter VI. --- CAUSALITY ANALYSIS --- p.54<br>Chapter 6.1 --- Granger Casuality Analysis --- p.55<br>Chapter 6.2 --- Results Interpretation --- p.56<br>Chapter VII --- CONSISTENCE VALIDATION --- p.61<br>Chapter VIII --- ARTIFICIAL NEURAL NETWORK TRADING SYSTEM --- p.67<br>Chapter 7.1 --- Trading System Architecture --- p.68<br>Chapter 7.2 --- Simulation Runs using the Trading System --- p.77<br>Chapter XI. --- CONCLUSIONS AND FUTURE WORKS --- p.79

Then, we consider the patterns between news stream and time series indices stream. We first transform the news stream into a set of bursty feature (keywords) time series streams and propose three technique to study their relationship to time series index. First, we explore a Non-homogeneous Hidden Markov Model (NHMM) to predict the stock market process which takes both stock prices and news articles into consideration. Second, we propose a risk analytical model to predict the volatility of price indices by integrating news information. Finally, we devise an algorithm to detect the priming event from text and a time series index. The evaluation on real world dataset suggests the significant correlation exists between news stream and time series stream and our pattern discover algorithm can detect promising patterns from this relationship to support real world applications effectively.<br>We start from investigating the co-movement relationship of multiple time series. We propose techniques to study two aspects of this problem. First, we propose a co-movement model for constructing financial portfolio by analyzing and mining the co-movement patterns among two time series. Second, we presents an efficient streaming algorithm to discover leaders from multiple time series stream. Both of the algorithms are evaluated using real time series indices data and the result proves that co-movement patterns and detected leaders are promising and can support various applications including portfolio management, high frequency trading and risk management.<br>With the increasing amount of data in financial market, there are two types of data streams attracting a lot of research and studies, time series index stream and related news stream. In this thesis, we focus on discovering patterns from these data streams and try to answer the following challenging questions, (I) given two co-evolving time series indices, what is the co-movement dependency between them. (II) given a set of evolving time series, could we detect some leaders from them whose rise or fall impacts the behavior of many other time series? (III) could we integrate the news stream information into stock price prediction? (IV) could we integrate the news stream information into stock risk analysis? and (V) could we detect what are those events that trigger time series index movement. For each of the question, we design algorithms and address three technique issues (I) how to detect promising patterns from the noisy financial data; (II) how to update the old patterns when new data arrives in high frequency; (III) how to use the pattern to support the financial applications.<br>Wu, Di.<br>Adviser: Jeffrey Xu Pu.<br>Source: Dissertation Abstracts International, Volume: 73-01, Section: B, page: .<br>Thesis (Ph.D.)--Chinese University of Hong Kong, 2010.<br>Includes bibliographical references (leaves 124-131).<br>Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web.<br>Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [201-] System requirements: Adobe Acrobat Reader. Available via World Wide Web.<br>Electronic reproduction. Ann Arbor, MI : ProQuest dissertations and theses, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web.<br>Abstract also in Chinese.

by Lam King Wan.<br>Thesis (M.B.A.)--Chinese University of Hong Kong, 1999.<br>Includes bibliographical references (leaves 72-73).<br>ABSTRACT --- p.ii.<br>TABLE OF CONTENTS --- p.iv.<br>Section<br>Chapter I. --- OBJECTIVE --- p.1<br>Chapter II. --- INTRODUCTION --- p.3<br>The principal investment risk --- p.3<br>Effect of risk on investment --- p.4<br>Investors' concern for investment risk --- p.6<br>Chapter III. --- THE INPUT PARAMETERS --- p.9<br>Chapter IV. --- LITERATURE REVIEW --- p.15<br>What is an artificial neural network? --- p.15<br>What is a neuron? --- p.16<br>Biological versus artificial neuron --- p.16<br>Operation of a neural network --- p.17<br>Neural network paradigm --- p.20<br>Feedforward as the most suitable form of neural network --- p.22<br>Capability of neural network --- p.23<br>The learning process --- p.25<br>Testing the network --- p.29<br>Neural network computing --- p.29<br>Neural network versus conventional computer --- p.30<br>Neural network versus a knowledge based system --- p.32<br>Strength of neural network --- p.34<br>Weaknesses of neural network --- p.35<br>Chapter V. --- NEURAL NETWORK AS A TOOL FOR INVESTMENT ANALYSIS --- p.38<br>Neural network in financial applications --- p.38<br>Trading in the stock market --- p.41<br>Why neural network could outperform in the stock market? --- p.43<br>Applications of neural network --- p.45<br>Chapter VI. --- BUILDING THE NEURAL NETWORK MODEL --- p.47<br>Implementation process --- p.48<br>Step 1´ؤ Problem specification --- p.49<br>Step 2 ´ؤ Data collection --- p.51<br>Step 3 ´ؤ Data analysis and transformation --- p.55<br>Step 4 ´ؤ Training data set extraction --- p.58<br>Step 5 ´ؤ Selection of network architecture --- p.60<br>Step 6 ´ؤ Selection of training algorithm --- p.62<br>Step 7 ´ؤ Training the network --- p.64<br>Step 8 ´ؤ Model deployment --- p.65<br>Chapter 7 --- RESULT AND CONCLUSION --- p.67<br>Result --- p.67<br>Conclusion --- p.69<br>BIBLIOGRAPHY --- p.72

A Dissertation submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Master of Science, 2012.<br>The purpose of this study is to investigate if the International Comparison Program (ICP) methodology could be used to examine the different buying power (worth) of the currency on the same products or goods amongst South African provinces. The method will be tested on the Consumer Price Index (CPI) food data collected from January 2006 to December 2006 from the main cities in the provinces. The food basket is obtained via the Income and Expenditure Survey (IES), which is generally updated every 5 years. South Africa (SA) has disparities and differentials in economic indicators such as the CPI, Gross Domestic Product and employment, amongst the provinces which are caused by among other things geographic set-up, urbanisation, inflation rates, and expenditure patterns. We use the monthly data to do an inter-provincial comparison of food prices by deriving annual purchasing power parities (PPPs) for each of the provinces, using the Country Product Dummy (CPD) method recommended as best practice by the World Bank. The CPI data is validated using the SEMPER software developed by the African Development Bank (AfDB). The validated data is examined for variability over the months and between the provinces using Analysis of Variance. Significant price differences are found for various products over the months and between provinces. The validated data was used to compute PPPs at the group and basic heading level. PPPs were investigated for differences in the provinces on grouped level of food products using Analysis of Variance. The reliability of PPPs between provinces is investigated both at grouped and basic heading level of products using the Cronbach-alpha statistic. The results show that there are no significant variations in PPPs across provinces. This could be due to the similar business opportunities or developments in the provinces or due to the aggregation of prices from the individual product (basic heading) to the main product group level. This implies that the cost of the food basket is the same across provinces.

This thesis consists of three parts: a genetic fuzzy trading model for stock trading, incremental intraday information for financial time series forecasting, and intraday effects in conditional variance estimation. Part A investigates a genetic fuzzy trading model for stock trading. This part contributes to use a fuzzy trading model to eliminate undesirable discontinuities, incorporate vague trading rules into the trading model and use genetic algorithm to select an optimal trading ruleset. Technical indicators are used to monitor the stock price movement and assist practitioners to set up trading rules to make buy-sell decision. Although some trading rules have a clear buy-sell signal, the signals are always detected with 'hard' logical. These trigger the undesirable discontinuities due to the jumps of the Boolean variables that may occur for small changes of the technical indicator. Some trading rules are vague and conflicting. They are difficult to incorporate into the trading system while they possess significant market information. Various performance comparisons such as total return, maximum drawdown and profit-loss ratios among different trading strategies were examined. Genetic fuzzy trading model always gave moderate performance. Part B studies and contributes to the literature that focuses on the forecasting of daily financial time series using intraday information. Conventional daily forecast always focuses on the use of lagged daily information up to the last market close while neglecting intraday information from the last market close to current time. Such intraday information are referred to incremental intraday information. They can improve prediction accuracy not only at a particular instant but also with the intraday time when an appropriate predictor is derived from such information. These are demonstrated in two forecasting examples, predictions of daily high and range-based volatility, using linear regression and Neural Network forecasters. Neural Network forecaster possesses a stronger causal effect of incremental intraday information on the predictand. Predictability can be estimated by a correlation without conducting any forecast. Part C explores intraday effects in conditional variance estimation. This contributes to the literature that focuses on conditional variance estimation with the intraday effects. Conventional GARCH volatility is formulated with an additive-error mean equation for daily return and an autoregressive moving-average specification for its conditional variance. However, the intra-daily information doesn't include in the conditional variance while it should has implication on the daily variance. Using Engle's multiplicative-error model formulation, range-based volatility is proposed as an intraday proxy for several GARCH frameworks. The impact of significant changes in intraday data is reflected in the MEM-GARCH variance. For some frameworks, it is possible to use lagged values of range-based volatility to delay the intraday effects in the conditional variance equation.<br>Ng, Hoi Shing Raymond.<br>Adviser: Kai-Pui Lam.<br>Source: Dissertation Abstracts International, Volume: 72-01, Section: B, page: .<br>Thesis (Ph.D.)--Chinese University of Hong Kong, 2010.<br>Includes bibliographical references (leaves 107-114).<br>Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web.<br>Electronic reproduction. Ann Arbor, MI : ProQuest Information and Learning Company, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web.<br>Abstract also in Chinese.

Ng Anny.<br>Thesis (M.Phil.)--Chinese University of Hong Kong, 2002.<br>Includes bibliographical references (leaves 139-142).<br>Abstracts in English and Chinese.<br>Chapter 1 --- Introduction --- p.1<br>Chapter 1.1 --- Problem Definition --- p.2<br>Chapter 1.2 --- Thesis Organization --- p.3<br>Chapter 2 --- Chinese Text Summarization Using Genetic Algorithm --- p.4<br>Chapter 2.1 --- Introduction --- p.4<br>Chapter 2.2 --- Related Work --- p.6<br>Chapter 2.3 --- Genetic Algorithm Approach --- p.10<br>Chapter 2.3.1 --- Fitness Function --- p.11<br>Chapter 2.3.2 --- Genetic operators --- p.14<br>Chapter 2.4 --- Implementation Details --- p.15<br>Chapter 2.5 --- Experimental results --- p.19<br>Chapter 2.6 --- Limitations and Future Work --- p.24<br>Chapter 2.7 --- Conclusion --- p.26<br>Chapter 3 --- Event Extraction from Chinese Financial News --- p.27<br>Chapter 3.1 --- Introduction --- p.28<br>Chapter 3.2 --- Method --- p.29<br>Chapter 3.2.1 --- Data Set Preparation --- p.29<br>Chapter 3.2.2 --- Positive Word --- p.30<br>Chapter 3.2.3 --- Negative Word --- p.31<br>Chapter 3.2.4 --- Window --- p.31<br>Chapter 3.2.5 --- Event Extraction --- p.32<br>Chapter 3.3 --- System Overview --- p.33<br>Chapter 3.4 --- Implementation --- p.33<br>Chapter 3.4.1 --- Event Type and Positive Word --- p.34<br>Chapter 3.4.2 --- Company Name --- p.34<br>Chapter 3.4.3 --- Negative Word --- p.36<br>Chapter 3.4.4 --- Event Extraction --- p.37<br>Chapter 3.5 --- Stock Database --- p.38<br>Chapter 3.5.1 --- Stock Movements --- p.39<br>Chapter 3.5.2 --- Implementation --- p.39<br>Chapter 3.5.3 --- Stock Database Transformation --- p.39<br>Chapter 3.6 --- Performance Evaluation --- p.40<br>Chapter 3.6.1 --- Performance measures --- p.40<br>Chapter 3.6.2 --- Evaluation --- p.41<br>Chapter 3.7 --- Conclusion --- p.45<br>Chapter 4 --- Mining Frequent Episodes --- p.46<br>Chapter 4.1 --- Introduction --- p.46<br>Chapter 4.1.1 --- Definitions --- p.48<br>Chapter 4.2 --- Related Work --- p.50<br>Chapter 4.3 --- Double-Part Event Tree for the database --- p.56<br>Chapter 4.3.1 --- Complexity of tree construction --- p.62<br>Chapter 4.4 --- Mining Frequent Episodes with the DE-tree --- p.63<br>Chapter 4.4.1 --- Conditional Event Trees --- p.66<br>Chapter 4.4.2 --- Single Path Conditional Event Tree --- p.67<br>Chapter 4.4.3 --- Complexity of Mining Frequent Episodes with DE-Tree --- p.67<br>Chapter 4.4.4 --- An Example --- p.68<br>Chapter 4.4.5 --- Completeness of finding frequent episodes --- p.71<br>Chapter 4.5 --- Implementation of DE-Tree --- p.71<br>Chapter 4.6 --- Method 2: Node-List Event Tree --- p.76<br>Chapter 4.6.1 --- Tree construction --- p.79<br>Chapter 4.6.2 --- Order of Position Bits --- p.83<br>Chapter 4.7 --- Implementation of NE-tree construction --- p.84<br>Chapter 4.7.1 --- Complexity of NE-Tree Construction --- p.86<br>Chapter 4.8 --- Mining Frequent Episodes with NE-tree --- p.87<br>Chapter 4.8.1 --- Conditional NE-Tree --- p.87<br>Chapter 4.8.2 --- Single Path Conditional NE-Tree --- p.88<br>Chapter 4.8.3 --- Complexity of Mining Frequent Episodes with NE-Tree --- p.89<br>Chapter 4.8.4 --- An Example --- p.89<br>Chapter 4.9 --- Performance evaluation --- p.91<br>Chapter 4.9.1 --- Synthetic data --- p.91<br>Chapter 4.9.2 --- Real data --- p.99<br>Chapter 4.10 --- Conclusion --- p.103<br>Chapter 5 --- Mining N-most Interesting Episodes --- p.104<br>Chapter 5.1 --- Introduction --- p.105<br>Chapter 5.2 --- Method --- p.106<br>Chapter 5.2.1 --- Threshold Improvement --- p.108<br>Chapter 5.2.2 --- Pseudocode --- p.112<br>Chapter 5.3 --- Experimental Results --- p.112<br>Chapter 5.3.1 --- Synthetic Data --- p.113<br>Chapter 5.3.2 --- Real Data --- p.119<br>Chapter 5.4 --- Conclusion --- p.121<br>Chapter 6 --- Mining Frequent Episodes with Event Constraints --- p.122<br>Chapter 6.1 --- Introduction --- p.122<br>Chapter 6.2 --- Method --- p.123<br>Chapter 6.3 --- Experimental Results --- p.125<br>Chapter 6.3.1 --- Synthetic Data --- p.126<br>Chapter 6.3.2 --- Real Data --- p.129<br>Chapter 6.4 --- Conclusion --- p.131<br>Chapter 7 --- Conclusion --- p.133<br>Chapter A --- Test Cases --- p.135<br>Chapter A.1 --- Text 1 --- p.135<br>Chapter A.2 --- Text 2 --- p.137<br>Bibliography --- p.139

M.Comm.<br>The availability of large amounts of information and increases in computing power have facilitated the use of more sophisticated and effective technologies to analyse financial markets. The use of neural networks for financial time series forecasting has recently received increased attention. Neural networks are good at pattern recognition, generalisation and trend prediction. They can learn to predict next week's Dow Jones or flaws in concrete. Traditional methods used to analyse financial markets include technical and fundamental analysis. These methods have inherent shortcomings, which include bad timing of trading signals generated, and non-continuous data on which analysis is based. The purpose of the study was to create a tool with which to forecast financial time series on the Johannesburg Stock Exchange (JSE). The forecasted time series information was used to generate trading signals. A study of the building blocks of neural networks was done before the neural network was designed. The design of the neural network included data choice, data collection, calculations, data pre-processing and the determination of neural network parameters. The neural network was trained and tested with information from the financial sector of the JSE. The neural network was trained to predict share prices 4 days in advance with a Multiple Layer Feedforward Network (MLFN). The mean square error on the test set was 0.000930, with all test data values scaled between 0.1 - 0.9 and a sample size of 160. The prediction results were tested with a trading system, which generated a trade yielding 20 % return in 22 days. The neural network generated excellent results by predicting prices in advance. This enables better timing of trades and efficient use of capital. However, it was found that the price movement on the test set within the 4-day prediction period seldom exceeded the cost of trades, resulting in only one trade over a 5-month period for one security. This should not be a problem if all securities on the JSE are analysed for profitable trades. An additional neural network could also be designed to predict price movements further ahead, say 8 days, to assist the 4-day prediction