High frequency trading models website

Major High-Frequency Trading Firms in the United States 6 Existing Revenue Models of High-FrequencyRevenue Models of Investment... Revenue Model 8: Venture Capital 20High-Frequency Tradi

iv

High-Frequency Trading Models

G E W E I Y E , P h D

John Wiley & Sons, Inc.

Copyright
C 2011 by Gewei Ye All rights reserved.

Published by John Wiley & Sons, Inc., Hoboken, New Jersey.

Published simultaneously in Canada.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 646-8600, or on the Web at www.copyright.com Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permissions

Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose No warranty may be created

or extended by sales representatives or written sales materials The advice and strategies contained herein may not be suitable for your situation You should consult with a

professional where appropriate Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002.

Wiley also publishes its books in a variety of electronic formats Some content that appears in print may not be available in electronic books For more information about Wiley products, visit our web site at www.wiley.com

Library of Congress Cataloging-in-Publication Data:

Ye, Gewei, 1971–

High-frequency trading models / Gewei Ye.

p cm – (Wiley trading series) Includes bibliographical references and index.

ISBN 978-0-470-63373-1 (cloth)

1 Investment analysis 2 Speculation–Mathematical models 3 Portfolio

management–Mathematical models 4 Financial engineering I Title.

To my parents, Lei, Jessica, and friends

iv

Major High-Frequency Trading Firms in the United States 6 Existing Revenue Models of High-Frequency

Revenue Models of Investment

Revenue Model 8: Venture Capital 20

High-Frequency Trading

Technology Inventions to Drive Financial Inventions 34 The Ultimate Goal for Models and Financial Inventions 34

Computer Algos for High-Frequency

Explaining the Finding with the Black-Scholes Formula 63

CHAPTER 6 Expanding the Size of Options in

Analysis 1: Examination of Correlations and a

and Signal Detection Models for

Transforming Likeability Rating Data into Observed

Assessing Value at Risk with Risk Propensity of SDT for

CHAPTER 9 Behavioral Economics Models

on Fund Switching and Reference

Pricing Engine for Portfolio

SAPE Extensions: TopTickEngine, FundEngine,

Alternative Assessment Tools of Macro Investor

Case Study 1: Execution of SAPE Investment Strategies 206

Case Study 3: Advanced Trading Strategies with SAPE 217

Creating a Successful Fund with SAPE and

New Profitable Financial Instruments by Writing Options 236 The Black-Scholes Model as a Special Case of the

Pricing an Interest Rate Swap with Prospect Theory 241 Behavioral Investing Based on Behavioral Economics 243

Jump-Starting Algo Development with PHP Programming 256 Jump-Starting Algo Development with Java Programming 266 Jump-Starting Algo Development with C ++ Programming 273 Jump-Starting Algo Development with Flex Programming 274

Common UNIX/LINUX Commands for Algo Development 276

Developing a Flex User Interface for Computer Algos 286

Let’s start the book by explaining the title: High-Frequency

Trad-ing Models First, there are three types of models of high-frequencytrading: revenue models, theoretical (including behavioral, quanti-tative, and financial) models, and computer models Revenue models arestrategies, means, and ways to generate revenue and profit for a financialinstitution Theoretical models are foundations for building computer mod-els for high-frequency trading operations Computer models refer to thecomputer algorithms (algos for short) that program the theoretical modelsand trading ideas To summarize, the computer algos automate the tradingideas and the theoretical models with computer programming languagesand technology infrastructure so that the revenue models of financial insti-tutions may be materialized in a systematic way

Thus, the high-frequency trading models may expand to the following:(1) existing revenue models; (2) new revenue modes, for example, high-frequency trading in derivatives markets; (3) theoretical (behavioral, finan-cial, and quantitative) models for building unique investment strategies forhigh-frequency trading; and (4) computer algos for high-frequency tradingand portfolio management These four topics make up the central themes

of the book

Second, the high-frequency trading models belong to investment search and practice that are part of investment management Investmentmanagement provides professional asset and portfolio management forfinancial institutions or private investors As part of the functions of

a financial institution, investment management provides the people, sources, and objectives to conduct high-frequency trading operations withthe high-frequency trading models as the tools or goals The advent ofhigh-frequency trading may impact the investment management industryprofoundly if not revolutionarily As a result, investment management, in-cluding portfolio management, will benefit from computer algos e.g., Sen-timent Asset Pricing Engine (SAPE) designed for high-frequency tradingoperations

re-To elaborate, SAPE is a unique set of computer algos that are built ontop of several Nobel models such as modern portfolio theory (MPT) andthe capital asset pricing model (CAPM), the Black-Scholes option pricingmodel, and the autoregressive conditional heteroskedastic (ARCH) model,

by engaging a human behavioral factor, namely, traders’ sentiment Thoughthe Nobel models have considered important elements such as risk andreturn, future-dated option pricing, and volatility clustering, traders’ sen-timent can also affect stock prices As the Nobel models did not considerbehavioral factors in asset pricing, SAPE fills in the gaps by adding traders’real-time sentiment to the equation SAPE estimates future prices of in-dividual assets by aggregating traders’ real-time sentiment Compared tothe Nobel models that provide theories and formulas, SAPE provides anend-to-end solution to portfolio management, including a new theory onbehavioral investing, a new formula on estimating future prices of individ-ual assets, and a new computer system for real-time future asset pricing,asset allocation, and market timing

At a higher level of abstraction, SAPE for portfolio management, with

a collection of computer algos for high-frequency trading, represents nology as the driver of financial innovation and risk management SAPEalgos reflect the principle that technology may have a profound impact

tech-on new financial instruments and applicatitech-ons Similar to the probabilitytheory driving inventions in portfolio management, insurance, and riskmanagement, the advent of high-frequency trading, with an emphasis oninformation technology, may give rise to more liquidity in securities (espe-cially derivatives) markets, and to inventions in investment managementand risk management

High-frequency trading has swept Wall Street with the stunning profitgenerated by top tier investment banks In the meantime, high-frequencytrading has been mentioned repeatedly in headline news such as congres-sional hearings on the practices of Goldman Sachs, and the record-highmarket volatility on May 6, 2010

Many financial institutions, regulators, and financial professionalswould like to know how high-frequency trading works, how it profits, andwhat is needed to build the algorithms with technologies available in thepublic domain There is a lot of demand in the financial services and regu-latory community for an in-depth book on this topic

The audience for this book includes traders, regulators, portfolio agers, financial engineers, IT professionals, graduate or senior undergradu-ate students in finance, investment analysts, financial advisors, investmentbankers, hedge fund managers, and financial institutions

man-This book may be instrumental to the effort of reforming domestic orglobal financial systems and improving financial regulations Imagine if fi-nancial regulators could develop a new high-frequency trading monitoring

system based on the theoretical models and computer algos of this book.The monitoring system might automatically detect the preconditions ofmarket anomalies and prevent the occurrence of undesirable anomalies.

It would be especially useful for financial regulators to use computer algos

to monitor and regulate trading As a result, abnormal market behaviorslike the one on May 6, 2010, could be anticipated

The book comprises four parts: Part I describes the fundamental enue models of high-frequency trading; Part II discusses theoretical models

rev-as a foundation of the computer algos used in high-frequency trading; PartIII creates a unique model of sentiment asset pricing engine for portfoliomanagement and high-frequency trading; Part IV discusses new models andcomputer algos of high-frequency trading The four parts are illustrated inthis outline

Option Pricing Computer

Algos

Unique Model (Section 3)

Behavioral Economics Models

Option Pricing Models

Behavioral Investing

Revenue Models (Section 1)

Theoretical Models (Section 2)

Investment Management

SAPE

Yeswici.com provides ancillary materials for the computer algos tioned in the book It is a quantitative modeling and computing platformfor innovative investment research The platform transfers investment re-search to Internet and mobile apps

I would like to thank Kevin Commins, Meg Freeborn, Michael Lisk,

Pamela van Giessen, and other members of the team at John Wiley &Sons for their hard work to make this book possible

Many thanks go to Dr Fred van Raaij, Dr Naresh Malhotra, Dr TraceyKing, and Dr Curtis Haugtvedt for their help with the theoretical models(in Part II)

Many thanks also go to the colleagues and graduate students at JohnsHopkins University for their constructive feedback to the unique modeland algorithms (in Part III)

xiv

P A R T I

Revenue Models of High-Frequency

Trading

In Part I, we cover the introduction to high-frequency trading and theexisting revenue models of high-frequency trading In addition, we dis-cuss the roots, history, and future of the industry

2

C H A P T E R 1

High-Frequency Trading and Existing Revenue Models

In this chapter, we discuss the basic concepts of high-frequency trading:

what it is; why it is important; who the major players are in the UnitedStates; how it earns a profit; and how to categorize high-frequency trad-ing operations

WHAT IS HIGH-FREQUENCY TRADING?

High-frequency trading has swept Wall Street and made quite a few newsheadlines since 2008, with the stunning profit generated by top tier invest-ment banks Only a small percentage (about 2 percent) of all the tradingfirms operate the high-frequency trading business So many financial insti-tutions and professionals would like to know what high-frequency trading

is Let’s answer the question here

High-frequency trading extends program trading that normally usescomputer algorithms to execute a collection of trading orders at fastspeeds, boosting market liquidity (see Hendershott, Jones, and Menkveld,

in press) The word high-frequency implies the boosting effect on

mar-ket liquidity compared to manual trading practices For example, as aliquidity booster, high-frequency trading operations may use sophisticatedcomputer algorithms to analyze multiple markets and execute arbitragestrategies with many orders at the same time

Based on the definition, there are four elements of a high-frequencytrading system: computer algorithms (algos); market liquidity booster;

collection of orders; and faster speed than manual executions Among thefour elements, computer algos and liquidity booster are the necessary con-ditions to forming a high-frequency trading operation.

As there is not yet a widely acceptable definition of high-frequencytrading among academia, practitioners, and regulators, we hope that thedefinition in this book will be adopted by many, as it addresses thebenefit of high-frequency trading to market liquidity at a higher level

of abstraction This benefit is of critical importance to the existence ofhigh-frequency trading practices, as regulators, who are the major force

in monitoring and regulating high-frequency trading, have acknowledgedthe benefit of liquidity boosting of high-frequency trading to the securitiesmarkets More on the concept of high-frequency trading may be found in

Irene Aldridge’s book High-Frequency Trading (2009).

Computer algos are programs written by financial engineers or ware engineers that automate the trading activities or quantitative modelsthat are traditionally conducted by human traders or researchers Some ofthe algos are based on mathematical models and some are not For exam-ple, a computer algo to compute the value at risk with Monte Carlo simu-lation uses statistical models Another computer algo to get the real-timequote of a security over the Internet may not require statistical or mathe-matical backing

soft-Collection of orders refers to the grouping of buying or selling ders that are normally used for arbitrage operations that, for example,attempt to profit from price differences between securities or exchangeswithin a short period of time Another example may be hedging: A tradermay use computer algos to combine a long position of an underlying as-set (stock or bond) and in the meantime hedge the risk of losing money

or-by buying a put option on this asset As a result, a collection of orders(buy the asset and the option) are formed as part of the automated tradingstrategy

The advantage of speed has been the major factor of success for earlyhigh-frequency trading operations This is enabled by super computingtechnology and sophisticated computer algos For example, it has beenreported that Goldman Sachs and IBM collaborated on this type of high-frequency trading to catch the price difference between buying and sellingorders within milliseconds, which has delivered significant profits In gen-eral, it is well received by the practitioners that the speed of order execu-tion would position the trades with a better chance of profit-making Theprinciple of co-location, requiring that the trading servers be located asclose as possible to an exchange, emphasizes this point for high-frequencytrading However, the speed of execution is not a required condition for ahigh-frequency trading system that may use advanced computer algos tooutperform peers

WHY HIGH-FREQUENCY TRADING

IS IMPORTANT

Why is high-frequency trading so important today? First, from afunctional perspective, high-frequency trading is a type of electronictrading that uses information technology to increase transparency andliquidity of securities markets The primary benefit of engaging high-frequency trading is boosting liquidity for the securities markets As such,according to Bloomberg news on May 13, 2009, Tim Geithner,1the U.S Sec-retary of the Treasury, has urged electronic trading for over-the-counter(OTC) derivatives that lack liquidity and transparency (Leising and Seeley2009) Note that the 2008 U.S GDP is $14.4 trillion (data from the WorldBank) The OTC derivatives market is a major part of the $600 trillionglobal derivatives markets that have been frozen due to the lack of liq-uidity, which was triggered by the financial crisis from late 2008 to the end

of 2009

The second reason for high-frequency trading’s importance is its ing volume According to the data circulating in the trading community, bythe end of 2009, the high-frequency trading firms, approximately 2 percent

trad-of the 20,000 trading firms trad-of the U.S markets account for over 70 percent

of all U.S equity trading volume These 2 percent of companies includeproprietary trading desks for a few major investment banks, less than ahundred of the hedge funds, and hundreds of small trading shops Theyall operate with one mission: maximizing profit by being smarter or fasterthan their peers

An illustrative work by advancedtrading.com (citing the research byTABB Group’s Iati) visualizes the three types of trading firms that createthe over 70 percent trading volume of U.S equity with high-frequency trad-ing First are the traditional broker-dealers who undertake high-frequencystrategies, separate from their client business Second are high-frequencyhedge funds Third are proprietary trading firms that are mainly usingprivate money

The third reason for high-frequency trading’s importance is that ithas produced a high-paying labor market that is hiring large numbers oftraders, developers, strategists, and analysts, even in an economy in finan-cial crisis with an unemployment rate of more than 10 percent in Decem-ber 2009, when the recovery of the economy and employment situationwas remote

Here is a job posted in December 2009 by a Wall Street cruiter seeking a quantitative algo trader: “Looking for someone with

re-a qure-antitre-ative bre-ackground who comes with/cre-an crere-ate their own re-rithms Must have experience with putting algo’s into a live production

algo-environment on equities, FX/futures or fixed Income PhD in FinancialEngineering, Physics, Chemistry, Economics, Finance, or similar pre-ferred Roles in NYC and Fairfield County, CT Total compensation

$350k–$750k+.”

Apparently, this job posting is very attractive to students or enced professionals, especially as it claims to pay three times the averagesalary of an equivalent employee in other fields such as technologyconsulting with similar technical skills, or scientific research with similarquant skills

experi-I teach graduate courses on Financial Engineering (with a focus onhigh-frequency trading algos, financial institutions, and derivatives) atJohns Hopkins University Carey Business School One of the courses had

an original limit of 30 enrollments It was quickly wait-listed for the firsttime The school then raised the enrollment limit to 40 Within weeks, itwas wait-listed again with 40 graduate students enrolled This happenedthree to four months before the start of the class

These examples demonstrate the popularity and importance of frequency trading and financial engineering in the investment managementindustry It has been perceived to be the future of major investing activitiesfor private investors such as hedge funds, and for public investors such asinvestment banks and mutual funds

high-MAJOR HIGH-FREQUENCY TRADING

FIRMS IN THE UNITED STATES

In the United States, the major locations of the headquarters for some

of the major high-frequency trading firms are New York City in theNortheast and Chicago in the Midwest Part of the reason is the needfor co-location; an important requirement for high-frequency trading is

to place the algo trading servers as close as possible to the exchange.The New York Stock Exchange (NYSE) is located in New York City.CME Group is in Chicago, which operates two self-regulatory futures ex-changes, the Chicago Mercantile Exchange (CME) and the Chicago Board

of Trade (CBOT) CME Group is the result of a merger of CBOT andCME in 2007 that formed one of the largest derivatives exchanges inthe world

Table 1.1 shows a small portion of around 200 high-frequency tradingfirms in the United States as of 2009 The table displays the names of majorhigh-frequency trading players in the United States, the location of theirheadquarters, and their main characteristics of trading

EXISTING REVENUE MODELS

OF HIGH-FREQUENCY

TRADING OPERATIONS

In a recent Wall Street Journal article entitled “What’s Behind

High-Frequency Trading,” Scott Patterson and Geoffrey Rogow (2009) gated the goals and revenue models of high-frequency trading firms For amulti-line firm that operates on financial products from stocks to curren-cies to commodities, the revenue model is to profit from fleeting moves

investi-in these products The high-frequency tradinvesti-ing operation looks for nals,” such as the movement of option prices, that indicates which way themarket may move in short periods Some other high-frequency trading op-erations attempt to profit from finding ways to exploit the defects in thenetwork or computer infrastructure of trading

“sig-Market making is another revenue model for some of the frequency trading operations As a result of the frequent buying and selling,securities on both buy and sell side become easy to liquidate

high-Given the assumption of an efficient market for financial marketsand exchanges, some high-frequency trading operations exploit temporary

“market inefficiencies” and trade in ways that can make money before thebrief inefficiency disappears This extends the traditional arbitrage prac-tice that profits from opportunities such as temporary price difference

of a financial product (e.g., gold) in two exchanges (e.g., New York sus Japan) If risk is not involved, such as buying and selling the product

ver-at the same time, it is called pure arbitrage; if risk is involved, such asholding the product with one’s own capital, the operation is called riskarbitrage

Detecting and taking advantage of the bid-ask spread is another enue model for high-frequency trading operations A bid-ask spread indi-cates the difference between investors buying and selling a security Ahigh-frequency operation, for example, a computer algo, may detect thedifference in milliseconds while the trade between the buyer and seller is

rev-to be matched, and make the trade happen for the buyer and seller As aresult, the algo takes the tiny profit for the matching With the algo workingautomatically, the tiny profit may accumulate to a large sum

Many exchanges, such as the New York Stock Exchange, offer liquidityrebates of about one-third of a penny a share to high-frequency tradingoperations that are willing to make trades between buyers and sellerseasier to complete The exchange becomes more liquid than before due

to the high trade volumes The frequent trading volumes would produceprofit for the trading firms as the exchange awards the trades with theliquidity rebates

TABLE 1.2 Existing Revenue Models

Fleeting moves Stocks; currencies;

commodities

Institutions Signal detection Interest rates Institutions; individuals Infrastructure Defects of computing

environment

Institutions Inefficient market Tiny gains; financial models Institutions; individuals

Liquidity rebates Exchange offers 0.33 penny

per share for improving market liquidity

To further categorize high-frequency trading operations at a higher level

of abstraction, let’s discuss the two criteria that organize high-frequencytrading operations, followed by the four types of high-frequency tradingoperations The two criteria are (1) the types of financial markets (efficientversus inefficient) where a high-frequency trading operation produces rev-enue, and (2) the participants or actors of the high-frequency trading oper-ations (institutions or private investors)

The revenue models of high-frequency trading firms may be rized in two parts: Does the model produce profit in an efficient mar-ket, or an inefficient market? The notion of efficient versus inefficientmarkets comes from academic literature An efficient market refers tothe financial market that all players, institutions or individuals, can max-imize the utility of their resources with super computing power and ad-vanced intelligence In reality, the efficient market assumption may nothold, especially for trades that are conducted by human traders Therefore,some financial markets are inefficient in that the trading represents char-acteristics such as bias, overconfidence, sentiment-driven, rumor-led, and

catego-so forth

The participants of a high-frequency trading operation may be looselydefined as two groups: financial institutions or private (“individual”)

TABLE 1.3 Categorizations of High-Frequency Trading

Institutions Examples: Liquidity rebates;

infrastructure defects

Example: Tiny gains on pure arbitrage

Private investors Example: Trades with signals Example: Algo-trade with

financial anomalies such as behavioral economics models

investors Financial institutions refer to the entities that handle a largevolume of financial resources that are pooled from the public For exam-ple, investment banks, mutual funds, pension funds, and insurance compa-nies are institutional investors Private or individual investors, sometimescalled retail investors, use brokerage services to invest their own financialresources in various financial instruments For example, individual traders,family offices of high net worth investors, private equities, and hedge fundsare private investors A good example to clarify the criterion of institu-tions versus private investors is to compare mutual funds and hedge funds.Both funds pool financial resources from many people or institutions(e.g., commercial banks) and invest in various financial markets How-ever, the difference between the two kinds of funds is that mutual fundsregister with the U.S Securities and Exchange Commission (SEC), whilehedge funds normally do not register with the SEC This distinction makes

a mutual fund “institutional” or “public” and a hedge fund “individual”

or “private.”

Based on the revenue models and participants that a high-frequencytrading operation has engaged, we may categorize high-frequency tradingoperations as one of four types: the high-frequency trading operation (1)

in an efficient market with institutional investor; (2) in an efficient marketwith individual investors; (3) in an inefficient market with institutional in-vestors; and (4) in an inefficient market with private or individual investors.Table 1.3 shows the categorizations

CONCLUSION

In this chapter, we discussed the following topics:

rThe definition and concept of high-frequency trading.

rReasons why high-frequency trading is important.

rMajor high-frequency trading firms in the United States.

rBasic revenue models of high-frequency trading firms.

rFour types of high-frequency trading operations.

In the next chapter, we trace the roots of high-frequency trading to one

of the eight major functions of investment management, namely programtrading

12

C H A P T E R 2Roots of High-Frequency Trading in Revenue Models of Investment

Management

In order to trace the roots of high-frequency trading, Chapter 2 looks

at the eight functions of investment management, followed by a cussion of program trading, an extension of which is high-frequencytrading

conducted by financial professionals for financial institutions or vate investors It refers to the professional management of institutional

pri-or private investpri-ors’ assets to meet certain goals fpri-or the benefit of theinvestors Financial institutions include commercial banks, investmentbanks and securities firms, insurance companies, pension funds, mutualfunds and hedge funds, and finance companies Investment managers may

be the firms that provide investment management services, or personswho manage investors’ assets Investment managers are sometimes calledmoney managers, portfolio managers, fund managers, or even financialadvisors and planners

In their book Financial Institutions Management, Saunders and

Cornett (2008) organize the activities of the investment banking try for financial institutions and individual investors into eight types ofactivities: investing, investment banking, market making, trading, cashmanagement, mergers and acquisitions, back-office services, and venturecapitals It is a bit misleading to use the word “investment banking indus-try” to capture all the investment management services for financial insti-tutions and individual investors because “investment banking” may have aspecific meanings of underwriting public or private offerings to issue newsecurities Therefore, I use investment management instead to refer to the

indus-activities of the investment banking industry that Saunders and Cornettdiscussed that include the eight investment activities.

In the next section, we’ll discuss eight revenue models of investmentmanagement firms They are investing, investment banking, market mak-ing, trading, cash management, mergers and acquisitions, back-office ac-tivities, and venture capital

REVENUE MODEL 1: INVESTING

Investing refers to a long-term act for investment managers to manage vestors’ asset In this space, financial institutions such as mutual funds andETF funds are commonly used by investment managers to pool the re-sources from investors and diversify the resources with various financialinstruments for long-term gains and minimized risks

in-The general revenue model of investing is the fees that investmentmanagers charge to financial institutions, retail investors, or financial in-struments such as mutual funds For example, a pension fund may hireinvestment managers to choose A+ grade investment instruments for thesafest and stable long-term stream of income for retirees The investmentmanagers may choose fixed income mutual funds The U.S Securities andExchange Commission (SEC) allows mutual funds to pay for the sales anddistribution of the funds with the 12b-1 fees These fees may be part of therevenue model for the investment managers who specialize in investing

Art versus Science: Three Levels of Abstraction

in Investing and Trading

In my opinion, there are three levels of abstraction in investing pertinent

to the goals and the level of certainty on predicting the result of investing

activities Here abstraction means the expected certainty of the outcome

of the investment choice

First, at the level of art, investment managers treat the process of vesting as a black box Knowing the input, which is the starting financialresources, the outcome of the investing bears the maximum uncertainty tothe investment managers a priori In other words, investing as art shows asense that the beauty and appreciation of the activity lies largely in individ-uals’ subjective judgment

in-Second, at the semi-art and semi-science level, investment managersuse only quantitative methods such as regression analysis to analyze theblack box Building on linear regressions, we may choose the capital assetpricing model (CAPM) for asset and portfolio analysis; or we may choose

option pricing models such as the Black-Scholes model for nonlinear ysis for the asset selection or allocation At this level, the certainty level ofthe expected result is quantified For example, value at risk (VaR) providesrisk confidence levels for a portfolio construction When an unexpectedevent occurs, it is part of the anomalies and may be hedged with insurance

anal-or other risk management instruments So it is not guaranteed that at thesecond level, it is a sure win for the investing choice or the chosen invest-ment portfolio

Third, at the science level, this is where quantitative analysis meetscomputer science High-frequency trading belongs to this level of invest-ing With the requirement for speed and time limit to generate revenue, italso belongs to the trading universe When computer science and statisticalanalysis are combined, the expectation level for the outcome and risk man-agement is lifted, because in computer science, an unexpected event is notcalled an anomaly within the 5 percent or 1 percent significance levels—it

is called a bug or defect that need to be fixed Therefore, at the sciencelevel of investing, investment managers pursue the maximum certainty ofthe expected investment outcome In most cases, the revenue model mayengage countless automated trials with computer algos and as a result asure win is expected as the outcome

The expectation of the third level of abstraction may be the ultimategoal of risk management: The maximum certainty suggests the minimumrisk A financial invention in the past, driven by the probability theory, at-tempted to achieve this with diversifications in investment managementand insurance instruments such as credit default swaps (CDS) for riskmanagement The probability theory comes from the quantitative universe.With the rapid growth of information technology, principles of computerscience should contribute to risk management as well The principle of

“bug-free applications” may be one Its implementation in high-frequencytrading operations would be a novel case to demonstrate the contribution

of information technology to financial inventions and risk management.With the advent of high-frequency trading as the new paradigm fortrading and investing practice, we will see that high-frequency trading maylead to an expanded finance discipline with computer algos being a neces-sity With computer algos, the semi-art and semi-science trading paradigm,solely based on statistics, may be replaced with a new paradigm, namelyinvesting and trading as science engaging computer algos and statistics

Common Investing Vehicles

Mutual funds and hedge funds are common investing vehicles for retail vestors or accredited investors to achieve specified investment goals withrisks minimized or controlled

in-For retail investors, regardless of the amount of capital, mutual fundsmay reduce risk to a minimum by diversifying the pooled capital to maxi-mum independent asset classes Based on the modern portfolio theory andthe CAPM, portfolio managers of mutual funds may minimize the risk oftheir portfolios by expanding the total number of asset classes as much aspossible The statistical explanation for this is the formula to calculate risk

or volatility of a portfolio Assuming the independence of the assets of aportfolio, the risk of the portfolio positively relates to the volatility of an

asset over sqrt(N) N is the total number of assets of the portfolio Hence, when N increases, the portfolio risk decreases When the assets of a port- folio are not correlated in returns, increasing N may not cost the portfolio returns However, when the assets are correlated, increasing N would af-

fect the portfolio returns as shown in a typical efficient portfolio frontier.This is because the expected portfolio return is a function of the portfoliorisk (Markowitz, 1952)

The payment plans for mutual funds reflect the revenue models of tual fund companies On the other hand, the plans outline the cost for in-vestors to invest in mutual funds For most funds, there are three types ofpayment plans through three share classes (A, B, and C) Each share classhas different sales load, management, and 12b-1 fees Sales loads may becharged at the purchase of a fund as a percentage of the price, which iscalled front-end load Back-end loads are charged when the fund is sold(i.e., at redemption) Management fee is also a percentage of the fund value,paying for the expenses to maintain the fund such as portfolio managers’salaries 12b-1 fees, getting the name from the SEC’s rule number, coverthe sales and marketing of the funds to investors Table 2.1 summarizesthe three types of payment plans

mu-Hedge funds are designed for accredited investors who have to meetone of these two criteria according to SEC rules as of 2009: (1) an individualnet worth of 1 million; or (2) annual income of U.S $200,000 or annualhousehold income of U.S $300,000 Hedge funds charge high performancefees (e.g 20 percent) if the net asset value of the fund increases from theprevious assessment (i.e., high water mark), and an agreed absolute returnrate (i.e., the hurdle rate) is achieved

TABLE 2.1 Payment Plans of Mutual Funds

Share Classes Front-End Load Back-End Load Management/12b-1

TABLE 2.2 Comparing Mutual Funds and Hedge Funds

Load Front-end and back-end N/A

Management fees Yes Yes; 2/20 rule

Performance fee No Around 20% if satisfying high

water mark and hurdle rate

We may compare the fee structures of mutual funds and hedge funds

in Table 2.2 to better understand the two types of funds

REVENUE MODEL 2:

INVESTMENT BANKING

Broadly speaking, investment banking may refer to an industry thatinvolves creating and trading securities of equity, debt, and derivatives In-vestment banks and securities firms are the major institutions of the indus-try Narrowly speaking, investment banking refers to a revenue model forinvestment banks to generate profit by underwriting and distributing newissues of debt (corporate or government bonds), equity (e.g., stocks), andderivatives (e.g., options and futures)

The most newsworthy revenue model in investment banking may be

an initial public offering (IPO), the first-time offering of debt and equity

by a company Famous IPO stories include high-tech companies such asGoogle, Yahoo!, and Apple Through IPOs, investment banks earn best-effort fees or profit from firm-commitment practice

In a best-efforts underwriting, the investment banker acts as an agent

of the company issuing the security and receives a fee based on the ber of securities sold With a firm-commitment underwriting, the invest-ment banker purchases the securities from the company at a negotiatedprice and sells them to the investing public at a higher price Thus, the in-vestment banker has greater risk with the firm-commitment underwriting,since the investment banker will absorb any adverse price movements inthe security before the entire new issues are sold

num-Factors concerning the issuing firm include general volatility in themarket, stability, maturity and financial health, and the perceived appetitefor new issues in the marketplace The investment bank may consider thesefactors when negotiating the fees and/or pricing spread in making its deci-sion on the types of underwriting

In order to practice investment banking, let’s try this exercise Let usassume that a company called NewBaidu is an international search enginecompany, seeking IPO with your firm Your team is challenged to make adecision on choosing which type of underwriting for the IPO: best-effortsunderwriting or firm-commitment underwriting This requires you to create

a decision matrix with a scoring mechanism to quantify the risk and reward

of the two options Let us also assume that you would have to present theresult to the NewBaidu executives

REVENUE MODEL 3: MARKET MAKING

Market making creates secondary markets for assets, thus increasing theliquidity of the markets Similar to secondhand cars, financial securities inthe secondary market are not sold by the IPO companies Instead, the orig-inal ownerships of securities traded in the secondary markets frequently

do not exist anymore

One of the main benefits of market making is boosting the liquidity ofthe markets High-frequency trading is perfect for this; thus, one of the rev-enue models of high-frequency trading derives from market making, that

is, to profit from liquidity rebates, an incentive to increase the liquidity ofthe markets

When a security market is frozen (not liquid) due to various reasons,the financial industry may be affected as is the overall economy In thiscase, market making becomes critically important for economic recovery.During the financial and economic crisis of 2008–2009, derivatives mar-kets valued at about $600 trillion globally were frozen, especially for theover-the-counter (OTC) derivatives The OTC derivatives are not publiclytraded; thus they lack transparency and liquidity Note that the U.S GDP

is around $15 trillion in 2008, just a fraction of the size of the derivativesmarkets

A possible solution to make the secondary markets for the derivatives

is increasing the liquidity of the derivatives markets by pushing electronictrading and high-frequency trading for OTC derivatives This has been dis-cussed extensively in regulations by the U.S Congress and the Department

of the Treasury

REVENUE MODEL 4: TRADING

Compared to investing that seeks sustainable stream of income for the longterm with tools like asset allocation and security selection, trading tends to

seek profit on a short-term basis with sophisticated tools or strategies such

as leverages and market timing

Trading properly by the rules is essential to the health of the ties markets, as it provides necessary liquidity and market making As apart of trading, high-frequency trading inherits the characteristics of mar-ket making and the liquidity booster for the markets However, we need

securi-to realize that some of the high-frequency trading practices may threatenthe integrity and fairness of the markets As such, when we build com-puter algos for high-frequency trading systems, abiding by the rules andregulations should be part of the specifications of the system developmentprocess

In general, there are four types of trading; high-frequency trading tends one of them The four types of trading are: position trading, purearbitrage, risk arbitrage, and program trading High-frequency trading ex-tends program trading, involving computer algos to find opportunities andautomate the trading process Position trading takes positions over timewith the hope that the price movement would go the way as expected.Pure arbitrage trading requires the buying and selling of securities at thesame time for a profit, exploiting market inefficiency Risk arbitrage en-gages the time or ownership of the security; thus the risk of loss is associ-ated with the transactions

ex-REVENUE MODEL 5: CASH MANAGEMENT

Before 1999, investment management is separate from commercial ing, which manages consumers’ cash with deposits and loans With the Fi-nancial Services Modernization Act of 1999, the investment managementindustry may collect deposits from consumers Thus cash management be-comes another revenue model of investment banks and securities firms inaddition to creating and trading securities

bank-Cash management blurs the distinction between investment ment and traditional commercial banks Large banks such as Bank ofAmerica and Morgan Stanley handle both investment banking and commer-cial banking operations This has streamlined the securitization process ofmortgage-backed securities

manage-Mortgage-backed securities for special-purpose entities (see ure 12.1) are created by investment banks that buy packages of loansfrom the commercial banks that interface with mortgage debt owners Bycombining the commercial banking function with the investment bankingfunction, the creation of mortgage-backed securities becomes part of theintradepartmental transactions within a large bank

Fig-REVENUE MODEL 6: MERGERS

AND ACQUISITIONS

Mergers and acquisitions (M&A) has been a lucrative revenue model forinvestment banks Similar to the idea of creating new securities of debtequity of a single new company, mergers and acquisitions form new com-panies by combining two or more existing companies, most times followed

by issuing new securities or converting existing securities

For a fee, analysts of investment banks may seek opportunities forcompanies to merge by proposing expected benefit of the merger Theymay also be hired by a company to explore the disadvantages of an M&Aproposal so that the company may argue against the proposal

With a merger and acquisition, similar functions of the companies may

be combined or consolidated for efficiency concerns As a result of themerger, duplicate positions may appear which would give rise to layoffs orworkforce reductions Hence, news on mergers and acquisitions may bene-fit equity prices Yet it is normally not good news for most of the employees

of the affected companies

REVENUE MODEL 7: BACK-OFFICE

ACTIVITIES

The back-office functions of an investment management firm may be ceived as expenses for the firm Yet these functions are necessary to oper-ate the firm’s day-to-day business These functions include clearance andsettlement, escrow services, and IT services such as new account setupand management

per-Automating back-office activities with information technology hasproved to be a long-term investment for the firms The payoff may comefrom the savings of manual engagements and additional quality services of-fered to investors For example, for querying his or her account balances,

as opposed to scheduling appointments on weekdays to come to investorcenters, an investor may use a Web-based account management service toaccess the accounts anytime at a convenient location

REVENUE MODEL 8: VENTURE CAPITAL

Venture capital firms are a special type of investment bank or securitiesfirm They pool money from individual investors and other financialinstitutions (e.g., hedge funds, pension funds, and insurance companies)

to fund relatively small and new businesses (e.g., in social media andbiotechnology).

Normally, the revenue model of venture capital firms is to invest instartup companies for a part of the ownership of the companies They mayalso provide cash to the companies as corporate loans When the compa-nies become successful and even go IPO, the venture capital firm wouldget extremely lucrative returns For example, Sequoia Capital is a success-ful venture capital firm that has excellent returns on investing in high-techfirms such as Google, Apple, and YouTube

CREATING YOUR OWN REVENUE MODEL

Creating your own revenue models has been the main task for trepreneurs or corporate leaders With brand-new technology or newproduct, a paradigm shift could happen to an existing consumer market

en-to create a new market As a result, the entrepreneurs or leaders may have

a temporary monopoly with the new revenue model to produce a profit inthe new market When competition comes in, the revenue model may becopied by the competitors and the profit may decline

As part of a case study for a financial engineering class, I ask graduatestudents to form groups of three to start new financial institutions Eachstudent in the group is assigned an executive role as chief executive officer(CEO), chief technology officer (CTO), or chief marketing officer (CMO).The setting of the case could be: “You and two friends have a meeting atStarbucks, talking about why Warren Buffett and Bill Gates can amass somuch fortune and then give back to society All inspired, you decide to start

a new financial institution together with major financial institutions as theprimary clients Here are the questions to ask: (1) How will my businessgenerate revenue? (2) How will I launch my business plan? (3) How can Iget help and make connections?” Therefore, the first question for any newcorporation is: What is our revenue model?

I have developed a conceptual framework to guide students throughthe process of creating new ideas and revenue models for newcorporations

The goal for entrepreneurs is to achieve success and happinessthrough a sound revenue model The core of this conceptual frameworkstresses the use of novel means and approaches to deliver success (i.e., rev-enue and customer satisfaction) through innovation The framework hasthree stages: goal, strategy, and execution The goal is to deliver “painkiller-like” customer and investor satisfaction The strategy is to expand the ideapool and process the selected idea The idea pool comprises expandedknowledge, skills, and attitude in technology and business through learningand training These ideas will be filtered as one and only one opportunity

FIGURE 2.1 Generic Entrepreneurial Modeling

to enter a revenue model The revenue model takes in the business tunity and then converts it to a product or a service Using processors such

oppor-as price, promotion, and place, the revenue model will deliver profit andcustomer satisfaction as the output Figure 2.1 illustrates the approach tocreating a new revenue model

HOW TO ACHIEVE SUCCESS: FOUR

Outcome Above

Equal 10

Below

Joyful

Satisfied

Unhappy Expectations

FIGURE 2.2 Relationship between Satisfaction and Expectations

a comparison between two entities: expectation and outcome Say a dent takes a final exam If the outcome of the grade (e.g., C) is below hisexpectations (e.g., B), then he would be unhappy If the outcome of thegrade (e.g., B) meets his expectations, then he would be satisfied If thegrade outcome (e.g., A) exceeds his expectations, then he would be joyful.Figure 2.2 shows the relationship between satisfaction and expectations

stu-It is very important to manage reasonable expectations for investors.Otherwise, it would set up the business for failure even though the businessmay actually do very well Look at this example On February 1, 2006, theBBC published an analysis of Google’s stock prices and its business perfor-mance Google shares fell 7 percent the day after its earnings fell short ofWall Street expectations for the first time

Why did Google shares fall? Compared with an average profit increase

of 10 to 20 percent in the industry, Google is doing very well by delivering

an outcome of 82 percent profit increase However, as investors set veryhigh expectations in the first place, this stellar business performance stilldoes not meet the “very high” expectations In other words, many investorsare not satisfied and start to sell the stock This results in a 7 percent drop

in Google’s stock price Therefore, in order to make investors happy, sonable expectations should be set as a proper reference point for them tocompare to

rea-Knowing this principle, we may find a way to make people happy Let

us first start with ourselves After we work hard and work smart, a able lower expectation should be used to expect an uncertain outcome Ifthe outcome cannot be changed anymore, a relatively lower expectationwould better prepare us to accept and be happy with the outcome Theprinciple is applicable to making others happy The others could be ourparents, partners, customers, and so forth When others are happy, we will

reason-be more likely to reason-be happy Consciously framing the expectations of theoutcome to properly set the expectations may give rise to a satisfactoryresponse to the outcome, even if the actual outcome may not be that good.The other two values related to business success are intelligence andtruth The more intelligent a person or a team is, and the more truth the

person or the team knows, the more successful the person or team willlikely be Hence, intelligence and truth are the foundation for achievingsuccess with a sound revenue model.

From a personal perspective, the four drivers of success with a soundrevenue model are concentration, confidence, motivation, and persistence(CCMP) To achieve success, one has to concentrate on the objective of therevenue model If you have too many ideas, then choose one after carefulconsideration and stick to that one Concentrate on the one with all yourenergy and effort with 100 percent devotion because your competitors arestrong and committed If you invest just 50 percent of the effort, maybebecause you have two projects going on at the same time, then you areusing 50 percent of your strength to compete with others who are working

at 100 percent strength Unless you are super strong, you are unlikely tosucceed in both projects

A person’s confidence on the subject matter comes from what he orshe can offer in the competition Usually a person’s confidence is based

on his or her knowledge, skills, and attitude (KSA) Knowledge meanshow much the person knows about the subject matter, the strengths andweaknesses of competitors, and so forth Skills means what a person can

do to deliver on expectations and promises, inclusive of technical skillssuch as developing a computer program or managerial skills such as lead-ing a team of subject matter experts to success Attitude means that aperson should always stay positive even if things are not going as ex-pected This is especially crucial in a team-based environment One teammember’s attitude can influence the others’ motivation and effort in thecompletion of the group project This is the foundation of leadership andteamwork skills

To illustrate the relationship between success, confidence, and KSA,

we look at a competitive analysis with KSA, competitors, and the needsand wants of customers (the market) Figure 2.3 shows three circles; theone on the left is for the market, which represents the needs and wants ofcustomers or clients The customers may be individuals who are using theproducts or services you are offering, or business customers as the clients

or business partners of your company, or employers who are hiring talentwith specific qualifications

The circle on the right represents the KSA that we offer to customers Itmay take the form of products or services that are built on the knowledge,skills, and attitude, or our qualifications for employers or clients For ex-ample, one of the KSA constructs is the incremental creativity that a busi-ness student or entrepreneur may demonstrate in the offerings to clients.Incremental creativity suggests that the products or services of the offer-ings are relevant to the market’s needs and wants, and depart from the