Prelims

Jose Joy Thoppan (Saintgits Institute of Management, India)
M. Punniyamoorthy (National Institute of Technology, India)
K. Ganesh (McKinsey & Company, India)
Sanjay Mohapatra (Xavier Institute of Management, India)

Developing an Effective Model for Detecting Trade-based Market Manipulation

ISBN: 978-1-80117-397-1, eISBN: 978-1-80117-396-4

Publication date: 5 May 2021

Citation

Thoppan, J.J., Punniyamoorthy, M., Ganesh, K. and Mohapatra, S. (2021), "Prelims", Developing an Effective Model for Detecting Trade-based Market Manipulation, Emerald Publishing Limited, Leeds, pp. i-xiv. https://doi.org/10.1108/978-1-80117-396-420211022

Publisher

:

Emerald Publishing Limited

Copyright © 2021 M. Punniyamoorthy, Jose Joy Thoppan, K. Ganesh, and Sanjay Mohapatra. Published under an exclusive license by Emerald Publishing Limited


Half Title Page

Developing an Effective Model for Detecting Trade-based Market Manipulation

Title Page

Developing an Effective Model for Detecting-trade Based Market Manipulation

Jose Joy Thoppan

Saintgits Institute of Management, India

M. Punniyamoorthy

National Institute of Technology, India

K. Ganesh

McKinsey & Company, India

And

Sanjay Mohapatra

Xavier Institute of Management, India

United Kingdom – North America – Japan – India Malaysia – China

Copyright Page

Emerald Publishing Limited

Howard House, Wagon Lane, Bingley BD16 1WA, UK

First edition 2021

© 2021 M. Punniyamoorthy, Jose Joy Thoppan, K. Ganesh, and Sanjay Mohapatra

Published under an exclusive license by Emerald Publishing Limited

Reprints and permissions service

Contact:

No part of this book may be reproduced, stored in a retrieval system, transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without either the prior written permission of the publisher or a licence permitting restricted copying issued in the UK by The Copyright Licensing Agency and in the USA by The Copyright Clearance Center. Any opinions expressed in the chapters are those of the authors. Whilst Emerald makes every effort to ensure the quality and accuracy of its content, Emerald makes no representation implied or otherwise, as to the chapters' suitability and application and disclaims any warranties, express or implied, to their use.

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

ISBN: 978-1-80117-397-1 (Print)

ISBN: 978-1-80117-396-4 (Online)

ISBN: 978-1-80117-398-8 (Epub)

Abstract

‘Every day criminals may be stealing up to $400 million – 1 quarter of a percent of total trades – by manipulating the stock market’, says Alex Frino of the Sydney based Capital Markets Co-operative Research Centre. Most manipulation is detrimental to the trading venue and its participants. Market manipulation impairs price discovery and misrepresent the fair value of a security. The distorted prices force investors to migrate to more efficient markets for deploying their capital. This reduces order flow and increases the cost of trading at a particular trading venue. It further motivates companies coming up with new issue to list their securities at other markets where there are better regulations and more efficient monitoring. Hence, ways and means of understanding and eliminating manipulative practices attract great interest from researchers, regulators and exchanges.

This research seeks to determine an appropriate model to help identify stocks witnessing activities that are indicative of potential manipulation through three separate but related studies. In a market like India, where there are about 5,000 plus securities listed on its major exchanges, it becomes extremely difficult to monitor all securities for potential market abuse. In this research, classifiers based on three different techniques namely discriminant analysis, a composite classifier based on Artificial Neural Network and Genetic Algorithm and Support Vector Machines are proposed. The proposed models help investigators, with varying degree of accuracy, to arrive at a shortlist of securities which could be subject to further detailed investigation to detect the type and nature of the manipulation, if any.

Chapter 1 provides an introduction to the topic. In this chapter, the market structure and an efficient stock market are discussed. The topics covering Indian stock markets, stock price manipulation and stock market surveillance are also introduced.

Chapter 2 provides a detailed literature survey on the topics covering efficient markets, market integrity, market manipulation and market surveillance. In Chapter 3 the issues, scope and objectives of the research are discussed. In Chapter 4, the data and the three techniques that are used in the research are discussed.

In Chapter 5 and 6, the first classifier built based on discriminant analysis, which is one of the most popular classification techniques, is developed and applied. As a first step, the most popular and widely used Linear Discriminant Function is discussed as it has been widely used by researchers. It was also observed that researchers have used this technique without validating the assumption that governs the model. It is shown that the data collected from the Indian exchanges do not comply with the assumptions that govern the use of the Linear Discriminant Function. Based on literature review, it is shown that the Quadratic Discriminant Function (QDF) is the appropriate discriminant analysis based classification technique for instances where the data does not meet the stated assumptions of the Linear Discriminant Function, to categorize stocks as manipulated and non-manipulated. This classification is archived based on certain key market data variables that capture the characteristics of the stock.

In Chapter 7, a hybrid model using advanced data mining techniques like Artificial Neural Network and Genetic Algorithm is developed. An empirical analysis of this model is carried out to evaluate its ability to predict stock price manipulation for the same data that was used earlier. Further, the performance of this hybrid model is compared with a conventional standalone model based on Quadratic Discriminant Function (QDF). Based on the results obtained, it is concluded that the hybrid model offers better prediction accuracy than the conventional model.

In Chapter 8, the essentials of a Support Vector Machines (SVM) based model, first proposed by Vapnik, is presented in a simplified but detailed elucidation. Subsequently, a detailed description for applying SVMs to identify stocks that are witnessing activities indicative of potential manipulation is provided. Finally, the superiority of the model for the data has been established by comparing with the results obtained from the QDF and the ANN-GA composite classifier.

Keywords: Artificial neural network, genetic algorithm, market manipulation, quadratic discriminant function, radial basis function, support vector function, surveillance

List of Tables

Table 4.1. Misclassification Table.
Table 4.2. Summary of Results.
Table 5.1. Linear Discriminant Function Results.
Table 5.2. Confusion Matrix – Linear Discriminant Function.
Table 5.3. Error Count Estimates for Stock.
Table 6.1. D 2 Values for Plotting the Q-Q Plot.
Table 6.2. F Distribution Values.
Table 6.3. Calculated and Tabulated Values for F Distribution.
Table 6.4. Quadratic Discriminant Function Results.
Table 6.5. Confusion Matrix – Quadratic Discriminant Function.
Table 6.6. Error Count Estimates for Stock.
Table 7.1. Confusion Matrix.
Table 7.2. Error Count Estimates for Stock.
Table 8.1. Confusion Matrix.
Table 9.1. Summary of Results.

List of Figures

Figure 4.1. Artificial Neural Network Model.
Figure 4.2. A Sample Chromosome.
Figure 6.1. Q-Q Plot for Non-manipulated and Manipulated Dataset.
Figure 7.1. A Sample Chromosome.
Figure 7.2. Neural Network Based on Weights Extracted from Genetic Algorithm.
Figure 8.1. Hyperplanes Separating Data Into Two Categories.
Figure 8.2. Maximum Margin Hyperplane.

List of Abbreviations

ADS

Advanced Detection System

ANN

Artificial Neural Network

ARIMA

Autoregressive Integrated Moving Average

ASBA

Application Supported by Blocked Accounts

BBC

British Broadcasting Corporation

BSE

Bombay Stock Exchange

CDSL

Central Depository Services (India) Limited

CMCRC

Capital Markets Co-operative Research Centre

DCA

Department of Company Affairs

DEA

Department of Economic Affairs

DMA

Direct Market Access

DSE

Dakha Stock Exchange

EMH

Efficient Market Hypothesis

EPS

Earnings Per Share

ETF

Exchange Traded Funds

FIX

Financial Information eXchange

GA

Genetic Algorithm

IMSS

Integrated Market Surveillance System

IOSCO

International Organization of Securities Commissions

IPO

Initial Public Offering

KKT

Karush Kuhn Tucker

LDF

Linear Discriminant Function

MCX

Multi Commodities Exchange

MDA

Multiple Discriminant Analysis

NASD

National Association of Securities Dealers

NSDL

National Securities Depository Limited

NSE

National Stock Exchange

OTC

Over the Counter

P/E

Price to Equity

QDF

Quadratic Discriminant Function

RBF

Radial Basis Function

RBI

Reserve Bank of India

SEBI

Securities Exchange Board of India

SEC

Securities Exchange Commission

SONAR

Securities Observation, News Analysis and Regulation

SOR

Smart Order Routing

SRO

Self-Regulating Organisation

SVM

Support Vector Machines

TSE

Tunisian Stock Exchange