Are Smart Beta Portfolios Smarter than Market Capitalization

COVENTRY UNIVERISITY

LONDON CAMPUS MSc. Global Financial Trading

M034LON-Individual Consulting Project

Are Smart Beta Portfolios Smarter Than Market

Capitalization Weighted Portfolios?

Hyder Khan

Student ID: 3996827

Supervisor: Dr. Peter Ye

Submitted in fulfillment of requirements for the Master of Science Degree in

global financial trading

Academic Year: 2015/2016

2

M034: INDIVIDUAL CONSULTING PROJECT

Are Smart Beta Portfolios

Smarter than Market

Capitalization-weighted

Portfolios? Back-Testing Fundamental Portfolios.

Hyder Khan

3996827

Word Count: 12,184

EXECUTIVE SUMMARY

PURPOSE

The purpose of this research is to investigate if smart beta (fundamental) portfolios have better

performance than portfolios that are weighted using market capitalization. This research address

the issue that broad indexes should be weighted using fundamentals of the firms’ rather than

market capitalization as it is affected by irrational investors and speculators resulting in noise in

the markets and making them mean-variance inefficient.

DESIGN/METHODOLOGY/APPROACH

This paper used the book value per share, dividend per share, free cash flow, revenue, profit

(loss) and P/E ratio for determination of weights of the stocks to create fundamental portfolios

and compared them to benchmarked cap-weighted portfolio. The research addresses FTSE 100

constituents as the population and uses 10% of the population as samples size. This paper

conducts an experiment and back tests the fundamental portfolios over a period of fifteen years.

Moreover comparing arithmetic returns, risk and Sharpe ratio does the comparison of portfolios.

FINDINGS

Taking a sample of ten securities during the period of 1st January 2001 to 30th November 2015, it

has been observed that fundamental portfolios provide significantly better cumulative returns than

cap-weighted portfolios. The results pertaining to the chosen fundamentals show that there is an

improvement in the portfolios Sharpe ratios, risk and returns. The analysis in terms of mean-

variance efficiency shows that cap-weighted portfolios are not mean-variance efficient portfolios

as perceived by the industry and many masters programs. The research also contradicts the

argument my Malkiel (2014), arguing that additional risk of smart beta portfolios is a result of

taking additional risk. All of the fundamental portfolios were able to achieve better return than

cap-weighted portfolio at same or lower level of volatility with exception of free cash flow

weighted portfolio, which had higher risk as a result of negative free cash flows.

ORIGINALITY/VALUE

The papers contributed to the existing literature of fundamental indexation and alternative

weighting indexations by comparing fundamental portfolios with cap-weighted portfolios using

the FTSE 100 constituents in very resent years.

KEYWORDS

Smart beta, Fundamentals, Sharpe ratio, Mean-variance efficiency, Indexation

PAPER TYPE

Research paper

4

DECLARATION OF ORIGINALITY

This research paper is my own work and not been copied in part or in whole from any other

source except where duly acknowledged. All use of previous work has been recognized and

has been acknowledged within the main body to an entry in the reference list.

I agree that an electronic copy of this report might be stored and used for the purpose of

plagiarism prevention and detection.

I also understand that copying previous work is considered plagiarism and constitute a breach

of Coventry University regulations and will be dealt seriously.

ACKNOWLEDGEMENTS

I would like to express my gratitude to my supervisor Dr. Peter Ye for his feedback, remarks

and engagement throughout this research. Furthermore I would also like to thank him for

introducing me to such an interesting topic and supporting in all the way through.

I would also like to thank Mr. Tao Xue and Miss Ruobing Zhang for giving me the

opportunity to work with their firm over the ten-week period as an intern.

CONTENTS

Executive Summary ................................................................................................................... 3

Purpose ................................................................................................................................... 3

Design/Methodology/Approach ............................................................................................. 3

Findings.................................................................................................................................. 3

Originality/Value ................................................................................................................... 3

Keywords ............................................................................................................................... 3

Paper Type ............................................................................................................................. 3

Declaration of Originality .......................................................................................................... 4

Acknowledgements .................................................................................................................... 5

List of Tables ............................................................................................................................. 8

List of Figures ............................................................................................................................ 9

Introduction .............................................................................................................................. 12

Research Question and Objectives ....................................................................................... 14

Literature Review..................................................................................................................... 15

Justification .......................................................................................................................... 15

Literature .............................................................................................................................. 15

Criticism ............................................................................................................................... 17

Methodology ............................................................................................................................ 19

Justification .......................................................................................................................... 19

Research Design................................................................................................................... 19

Philosophy........................................................................................................................ 19

Approach .......................................................................................................................... 19

Time Horizon ................................................................................................................... 20

Data Collection and Sampling ......................................................................................... 20

Data Analysis and Experiment............................................................................................. 22

Assumptions ..................................................................................................................... 22

Scenario............................................................................................................................ 22

Experimentation ............................................................................................................... 23

Limitations ........................................................................................................................... 29

Budget and Ethics ................................................................................................................ 29

Performance Evaluation ........................................................................................................... 30

Performance Discussion....................................................................................................... 43

Conclusion ............................................................................................................................... 48

Recommendation ..................................................................................................................... 50

7

Further Research .................................................................................................................. 50

Reflective Learning .................................................................................................................. 51

Relationship Betweeen Intership and My Career ................................................................ 51

Learning Outcome ............................................................................................................... 52

Challenges ............................................................................................................................ 55

Conclusion ........................................................................................................................... 55

References ................................................................................................................................ 56

Appendix .................................................................................................................................. 61

Appendix 1: VarCov VBA Function ................................................................................... 61

Appendix 2: FTSE 100 Time Reference .............................................................................. 61

Appendix 3: Mean-Variance Frontier .................................................................................. 62

Appendix 4: Financial data and Weights ............................................................................. 70

Appendix 5: Ethics Approval Checklist .............................................................................. 74

8

LIST OF TABLES

Table 1: Randomly Selected Companies ................................................................................. 20

Table 2: Time Periods .............................................................................................................. 21

Table 3: Fundamental Data (ANTO) ....................................................................................... 21

Table 4: Weights Determined by Market Capitalizastion........................................................ 23

Table 5: Price of Portfolio at T0 .............................................................................................. 23

Table 6: Portfolio's Arithmetic Returns ................................................................................... 30

Table 7: Portfolio Turnover ..................................................................................................... 35

Table 8: Portfolio Risk and Sharpe Ratio ................................................................................ 36

Table 9: Maximum Sharpe Ratio ............................................................................................. 43

Table 10: SWOT Analysis Before Intership ............................................................................ 52

Table 11: SWOT Analysis After Research .............................................................................. 55

9

LIST OF FIGURES

Figure 1: Dot-Com Bubble March 2000 .................................................................................. 13

Figure 2: Chinese Market Crash June 2015 ............................................................................. 13

Figure 3: Smart Beta: Google Trends ...................................................................................... 14

Figure 4: Maximizing Sharpe Ratio Using Solver ................................................................... 25

Figure 5: Cumulative Returns T0-T15 ..................................................................................... 30

Figure 6: Cap-Weighted Portfolio VS Book Value Per Share Portfolio ................................. 31

Figure 7: Cap-Weighted Portfolio VS Dividend per Share Portfolio ...................................... 32

Figure 8: Cap-Weighted Portfolio VS FCF Portfolio .............................................................. 32

Figure 9: Cap-Weighted Portfolio VS Revenue Weighted Portfolio....................................... 33

Figure 10: Cap-Weighted Portfolio VS Profit (Loss) Weighted Portfolio .............................. 34

Figure 11:Cap-Weighted Portfolio VS P/E Ratio Weighted Portfolio .................................... 34

Figure 12: Portfolio Turnover .................................................................................................. 35

Figure 13: Portfolios Risk and Sharpe Ratio: T1 ..................................................................... 37









Figure 22: Portfolios Risk and Sharpe Ratio: T10 ................................................................... 41






10

Figure 28: Market Cap VS Book Volatility ............................................................................. 44

Figure 29: Market Cap VS Dividend Volatility ....................................................................... 44

Figure 30: Market Cap VS FCF Volatility .............................................................................. 45

Figure 31: Market Cap VS Revenue Volatility........................................................................ 45

Figure 32: Market Cap VS Earnings Volatility ....................................................................... 46

Figure 33: Market Cap VS P/E Ratio Volatility ...................................................................... 46

Figure 34: Career Plan ............................................................................................................. 51

Figure 35: Gantt-Chart ............................................................................................................. 54

Figure 36: FTSE 100 Time Reference ..................................................................................... 61

Figure 37: Mean-Variance Frontier: T1 ................................................................................... 62









Figure 46: Mean-Variance Frontier: T10 ................................................................................. 66



Figure 49:Mean-Variance Frontier: T13 .................................................................................. 68



Figure 52: Cap-Weighted Sample Selection ............................................................................ 70

Figure 53: Market Cap Data and Weights ............................................................................... 71

Figure 54: Free Cash Flow Data and Weights ......................................................................... 71

Figure 55: Revenue Data and Weights .................................................................................... 72

11

Figure 56: P/E Ratio Data and Weights ................................................................................... 72

Figure 57: Profit (Loss) Data and Weights .............................................................................. 73

Figure 58: Dividends Data and Weights .................................................................................. 73

12

INTRODUCTION

Recent years have given rise to a new portfolio management strategy called smart beta. Given

the catchy title and promises of active portfolio management, the strategy has already

attracted billions. The long ongoing debate about active versus passive management

strategies has another side to it. What if the investors can get best of both the strategies?

Passive management pursuers believe that the markets are efficient and reflect the fair price

of all securities; hence they try to match the performance of the market. They do this by

holding all or representative sample of all the securities in the index or a fund that closely

follow or tracks the investment index. Investing in an index does not avoid the risk rather it

spreads it widely. Tracking an index will not be affected by decline in one particular security

but it will follow both bull and bear of the market. This requires no special knowledge for

stock picking or timing the market and as securities are not traded on frequent basis the cost

of passive management is low.

On the other hand, we have active management style where the managers claim to achieve

alpha by using advanced techniques and knowledge of stock picking and market timing.

Active managers pick up a subset of securities from an index and try to outperform the

market (Benchmark). This management style requires frequent trading and expertise of a

portfolio manager and as a result the cost of management is high. As these managers are

trying to predict the future, the probability that they will get it right every time is low. There

is a lot of criticism about active management style, or how Sharpe (1991) and Fama and

French (2009) say “active management is a zero sum game before costs and negative sum

game after costs.”

Index funds traditionally use capitalization-weighting mechanism according to which the

weight of the security with higher market capitalization will be higher and vice-versa.

𝑀𝑎𝑟𝑘𝑒𝑡 𝐶𝑎𝑝𝑖𝑡𝑎𝑙𝑖𝑠𝑎𝑡𝑖𝑜𝑛 = 𝑆ℎ𝑎𝑟𝑒 𝑃𝑟𝑖𝑐𝑒 𝑋 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑆ℎ𝑎𝑟𝑒𝑠 𝑂𝑢𝑡𝑠𝑡𝑎𝑛𝑑𝑖𝑛𝑔 [Eq. 1]

According to Eq.1 if we assume that the number of shares outstanding remain constant over

time then, according to capitalization weighting mechanism investors’ would end up with

bunch of over valued securities in their portfolio. According to Graham (1949) investing

principles, investor should buy low and sell high, investing in an index in this case can be

considered unwise. Haugen and Baker (1991) challenged the efficiency of capitalization-

weighted stocks portfolio and found that investment opportunities exist to build equity

portfolio with equal or higher return with significantly less volatility than capitalization-

weighted portfolio. The idea that markets reflect the pair price of securities and cap-weighted

portfolios are mean-variance efficient is highly promoted by the investment industry and

various masters programs. However, numerous arguments have been made by scholars such

as Arnott and Hsu (2008) and Siegel (2006) arguing the presence of noise in the financial

markets. If these arguments hold then cap-weighted indexes or portfolios cannot be

considered mean-variance efficient and can be outperformed using other weighting styles.

Investing in an index fund by definition gives portfolio a beta of 1. Beta can be defined as a

measure of systemic risk that cannot be avoided as it represents the movement in prices of

securities in relation to the movement in the market (NASDAQ 2015; Krause 2001).

Passively investing in an index fund exposes the investor to systematic risk and if the index is

13

capitalization-weighted then it buys more of the stocks that are higher in price. Considering

the markets are news efficient, if the prices keep on rising due to good news effect (Fishe,

Gosnell, Lasser 1993) the capitalization-weighted portfolio will keep on buying stocks that

are getting more and more expensive. If the securities are ‘value stocks’ then the investors’

will yield a high rate of return on the contrary if it’s the news effect then the bullish bubble

will bust and lead to a crash in the market and on the portfolio of the investor at the same

time. This can be justified with the dot-com bubble (Figure 1) and the recent Chinese market

crash (Figure 2).

FIGURE 1: DOT-COM BUBBLE MARCH 2000

FIGURE 2: CHINESE MARKET CRASH JUNE 2015

The downturns of passive investment strategy and expensiveness of active management has

lead to development of new investing strategy called smart beta. Roncalli (2013) puts smart

beta as a marketing term used to refer to alternative-weighted indexing. Malkiel (2014)

simply defines smart beta techniques as, to tilt or flavor the portfolio in some direction such

as value versus growth. He further argues that two or more tilts can be added to a portfolio

and smart beta strategies are related to multi-factor model of asset pricing.

Simply put smart beta is an alternative-weighted index in which the assets are weighted using

any other metrics but market capitalization. There are broadly two forms i.e. fundamental

indexing and risk-based indexing. The underlying idea of smart beta is similar to active

14

managers i.e. to improve the risk-return profile of their portfolio by achieving alpha at similar

or lower level of volatility. Smart beta is a relatively new topic in the industry and we can

assume that more and more investors’ are trying to understand and pursue smart beta

techniques by looking at the Google trends chart for the term ‘Smart beta’ (Figure 3).

FIGURE 3: SMART BETA: GOOGLE TRENDS

According to Bloomberg Intelligence by the end of 2014 there were 400 US-domiciled smart

beta funds managing nearly 20% of all assets in domestic ETF’s compared to nothing in May

2000. Institutional survey outlines that 62% of institutions’ plan to increase the use of smart

beta ETF’s in next three years (Invesco 2015).

Smart beta being relatively new to the industry it has received a lot of criticism and at the

same time significant amount of acceptance. Malkiel (2014) argues that smart beta portfolios

do not consistently outperform the benchmark and when they do they fail the risk test.

Roncalli (2013) on the other hand concluded that using smart beta portfolios he was able to

reduce the volatility by 30% when compared to market capitalization-weighted portfolios.

Glushkov (2015) in his analysis found that 60% of the smart beta funds outperformed their

raw passive benchmark.

The endless debate and the arguments presented above poses that more research is vital on

the topic of discussion. Through the mode of this research we will try to find out if smart beta

portfolios are smarter than cap-weighted portfolios.

RESEARCH QUESTION AND OBJECTIVES

QUESTION:

Are smart beta portfolios smarter than market capitalization weighted portfolios?

OBJECTIVES:

Identify if smart beta portfolios produce better cumulative returns than cap-

weighted portfolio over the research time period.

Identify weather using smart beta techniques lead to increased volatility levels in

order to achieve better return.

Test weather smart beta portfolios produce better return and risk adjusted return

on year on year basis compared to cap-weighted portfolios.

Determine if the cap-weighted portfolios are more efficient portfolios than

fundamental portfolios.

15

LITERATURE REVIEW

This section will critically discuss the past literature on the research topic to develop a

theoretical framework and deep understanding of smart beta techniques.

JUSTIFICATION

To construct the literature review, a systematic integration of the existing body of knowledge

on portfolio management, indexation and mean-variance efficiency was undertaken. The

approach used in this section was inductive approach implying moving from the general

recognition of the importance and relevance of portfolio management and indexation also the

inconclusive and controversial results and assumptions of the literature. Following the

inductive approach, the author examined various leading journal articles on the topic of

discussions as the primary source of widely accepted knowledge.

The first step was to identify relevant papers on the subject matter, and the author focused on

titles and abstracts with keywords such as smart beta, indexation, fundamental indexing,

alternative approaches in indexing. The databases used to identify the journal articles were

Business Source Complete and Institutional Investor Journals. To maintain the quality of

literature, only the papers that were published in widely accepted journals were picked and

were carefully reviewed. Finally, the articles that were directly related to the research

conducted by author were selected to enhance the understanding of the topic and to construct

the literature review.

LITERATURE

Smart beta is an umbrella term for rule-based strategies that do not use the conventional

market capitalization weights. As discussed in the introduction the two broad bases of smart

beta techniques this research revolves around the fundamental side.

Capital market theory addresses that market portfolio hold all risky assets in the universe.

Market portfolio is based on Markowitz (1952) efficient frontier of risky assets, i.e. a mean-

variance efficient portfolio that provides highest level of return for a given level of risk

(standard deviation). Further research on portfolio selection done by Tobin (1958) presented

the separation theorem arguing that all investors should hold the market portfolio in

combination with the risk-free asset depending on their risk aversion rate. Efficient market

hypotheses (EMH) of Fama (1970) postulate that securities reflect their intrinsic value, which

seems to blend in with the concept of market capitalization weighting methodology.

However, the presence of over-reaction of investors in the market put the mean-variance

efficiency of market capitalization methodology to a hold. Market capitalization weighted

methodology forces the portfolio to over weight the over valued stocks and under weight the

undervalued stocks. Roll (1977; 1978) pointed out the unobservable nature of the market

portfolio; hence most scholars and the industry justify the use of broad indexes as proxies of

market. Broad indexes including FTSE 100, NASDAQ, SENSEX and Russell 2000 are

market-cap weighted, investors’ tracking these indexes are not particularly tracking an

efficient portfolio therefore raises a serious concern.

16

Arnott, Hsu and Moore (2005) concluded that broad indexes should be based on firms’

fundamental values rather than their market capitalization. They argued that the fundamental

indexation is less affected by trading activities of speculators and irrational investors and thus

provide greater mean-variance efficiency compared to capitalization-weighted method.

Schoenfeld (2006) pointed out that due to the nature of capitalization methodology, the

largest companies have greater impact on the overall performance of the index. As

capitalization-weighted methodology works alongside with EMH (Fama 1970), if the

companies are reflecting their intrinsic values then it is justifiable that companies with greater

market share gets more weight in the market portfolio or in this case proxy. Hsu (2006)

concluded high correlation between market capitalization and liquidity and can be further

argued that capitalization-weighted portfolios offer investors high liquidity and moreover it

offers diversity as a result of continuous rebalancing due to change in prices. De Bondt and

Thaler (1985; 1987) argued that investors overreact to new information, which leads to the

mispricing of the stocks. The capitalization-weighted portfolios are as efficient as the prices

and overreaction can jeopardize the efficiency of those portfolios. Whereas, allocating the

weights using fundamental methodology or smart beta techniques helps the portfolio allocate

weights on the fundamental value and on prospects that has already been materialized unlike

market capitalization weighted portfolios that allocate weights according to future prospects

of the firms.

Noise market hypotheses proposed by Siegel (2006) criticized the market capitalization

weighted portfolios and argues that these portfolios are ‘suboptimal’ due to the noise traders

in the market. He points out, “Prices can be influenced by speculators and momentum

traders, as well as by insiders and institutions that often buy and sell stocks for reasons

unrelated to fundamental value, such as for diversification, liquidity and taxes”. He also

argues that fundamental indexation helps the investors capture the mispricing of the

securities. The cap-weighted portfolios work in a way in which if the constituent of the

portfolio becomes over or under valued then the portfolio rebalances its weight accordingly.

Hsu and Campollo (2006) argued that the cap-weighted portfolios are likely to underperform

over time leaving them mean-variance inefficient when overreaction of investors is present.

Hsu (2006) introduced the ‘cap drag’ i.e. the cost of cap weighting as the square of the noise

in stock price. Further, Arnott and Hsu (2008) demonstrated mathematically that size, value

and stock market mean reversion are consequence of the noisy prices.

The most relevant literature to the topic of research is the research done by Arnott, Hsu and

Moore (2005), where they constructed fundamentals weighted index using book value, cash

flow, employment, revenue, sales and dividends for 1000 firms in the U.S. stock markets

from the period 1962 to 2004. The weights in the fundamental index were in accordance to

the average fundamental values of the stocks. The results concluded that the composite index

has a return of 12.47% compared to 10.53% of S&P 500. Sales and revenue-weighted

indexes had the highest levels of returns i.e. 12.91% and 12.87% respectively. The cap-

weighted reference (benchmark) had the lowest return of 10.35% with volatility as high as

15.2% compared to 15.1% of S&P 500 and 14.7% of composite index. Overall conclusion of

the research was that the fundamental-weighted index earned a higher return than S&P 500

and the cap-weighted index at the same or lower level of volatility. This can be further argued

that the cap-weighted index is not mean-variance efficient and lies inside the efficient curve

according to Markowitz (1952).

17

Hemminki and Puttonen (2008) conducted similar research, where they investigated the

performance of fundamental indexes in European stock markets for the period of 10 years i.e.

from 1996 to 2006. Their research focused on the constituents of Dow Jones Euro Stoxx50

index that covers the largest 50 stocks by market capitalization in Europe. The research

concluded that their results were in line with the research conducted by Arnott, Hsu and

Moore in 2005. The portfolios re-weighted according to the fundamental vales were able to

produce consistent higher returns and risk-adjusted returns. Clare, Motson and Thomas

(2013) conducted a comprehensive study on US share data from 1968-2011. They

programmed a computer to pick and weight each of the 1000 stocks in their sample at random

and repeated the exercise ten million times. The results out weighted the cap-weighted index

and almost all of the ten million trails outperformed the benchmarked cap-weighted index.

The concept of Smart beta or fundamental indexing is not only supported in developed

market but also in emerging markets. Research conducted by Arnott and Shepherd (N.D.)

concluded that cap weighting might not be ideal in emerging markets while fundamental

index strategies may well be the right passive solution. FTSE RAFI (Research Affiliates

Fundamental Index) supports their argument as the emerging market index achieved an

annual return of 15.9% when compared to its benchmarks return of 6.9% with similar level of

volatility from the period 1994 to 2009. From 1980s to 2009 the RAFI indexes in Japan,

Europe and global stock markets outperformed their respective benchmarks on a risk-

adjusted basis.

CRITICISM

As much as the evidence supports the existence of fundamental indexing or smart beta

techniques this topic has received a lot of criticism from scholars. Arnott, Hsu and Moore

(2005) opined that the fundamental indexation enjoys the benefits of value stocks and small

firms and avoid the cap drag in fundamentally weighted portfolios. Kaplan (2008) criticized

them by arguing that avoiding the cap drag leads to weighting errors by ignoring the future

prospects of the firms. He further argued that fundamental indexation would automatically

create a bias towards small cap firms in bullish market times. Another criticism received by

Arnott, Hsu and Moore (2005) was from Schoenfeld (2006), He argued that fundamental

indexation is a ‘Naïve multifactor model’ with well document value factors (anomalies). His

research pointed out that from 2000 to 2005 size, style and industry exposures were

accountable for 90% of variation in RAFI returns. He also argues that although RAFI

outperforms the benchmark each year, much of the outperformance comes from the first two

years.

Hsu and Campollo (2006) argued that fundamental indexations reduced the weights of the

stocks that have faster growing share prices than their fundamental values and claimed it to

be far from simple value investing. They contradicted the findings of Schoenfeld (2006)

indicating that U.S. fundamental index 1000 and 2000 both outperformed their respective

benchmarks i.e. S&P 500 and Russell 2000 respectively in both bull market and expansionary

economic environments.

Resent studies that criticized smart beta techniques include the research conducted by Chow

et al. (2011) who constructed alternative equity index for U.S. stock market from 1964 to

2009 and also for global stocks from 1987 to 2009. They found that the portfolios that

18

outperformed their cap-weighted benchmark, was a result of exposure to value and size. They

further claim that any of these strategies can be mimicked and recommended that cost is a

better evaluation criterion than returns. They also concluded that using four-factor model

proposed by Carhart (1997), regression of outperformance on value and size factors the risk-

adjusted alpha to be significantly indifferent from zero. Hsieh (2013) researched fundamental

indexation on emerging markets and concluded results in line with Chow et al. (2011) that

fundamental indexation has significant exposure to size and value in emerging markets. His

research also concluded that in emerging markets fundamental indexation after adjusting for

size and value risk the portfolios earns significantly negative abnormal returns. Dubil (2015)

criticized the smart beta techniques and concluded that these approaches are not long lasting.

He also argues that ‘smart beta techniques outperform cap-weighted funds’ is not supported

by theory, it is a result of statistical experiments.

As there is not enough evidence for both sides of the argument, though this research author

will aim to contribute to the ideology of Arnott, Hsu and Moore (2005) and also try to find

out if cap-weighted portfolios produce better mean-variance efficiency than smart beta

portfolios to contribute towards the efficient market hypothesis by Fama (1970).

19

METHODOLOGY

The methodology followed in this paper is in line with Arnott, Hsu and Moore (2005),

followed by a drift in the sampling and data collection. The methodology also follows the

mean-variance efficiency theory by Markowitz (1952). In this section the paper will outline

in detail the approach, design and various techniques used for finding the answer to the

research question.

JUSTIFICATION

The methodology pursued for the research uses modern portfolio theory’s mean-variance

efficiency by Markowitz (1952) and Sharpe ratio as a measure of return per unit of risk

(Sharpe 1994; 1975). The methodology used for creation of portfolios and measuring the risk

and returns are widely acceptable theories and the research uses pure mathematical models

and acceptable knowledge of statistics to find and justify the results. The study used

secondary quantitative data that is freely available in the market and test it on accepted

mathematical models of modern portfolio theory. To be able to find the answer to the

research question and meet the objectives this approach was best suited for the research, as

pursuing any other form such as interviews, surveys or case studies would not provide any

significant justifiable results when compared to mathematical models. The use of

mathematical models also allows the results of the research to be scientific as if the same data

and methodology is applied same results will be achieved.

The reliability of data is maintained as free market data is used and the author cannot impose

any kind of biases to the data. The validity of the results is also significant as the research

only used mathematical models and statistical knowledge; this also secures the repeatability

of the results.

RESEARCH DESIGN

PHILOSOPHY

The research employs epistemology as research philosophy as the facts are being addressed

by using the available acceptable knowledge (Saunders and Lewis 2012). The research uses

freely available market data with widely accepted knowledge of mathematical models with

no scope of personal biases of the author affecting the outcome.

The research follows realism under epistemology as a philosophical stance as all the findings

are based on pure mathematical models and the significance of the finding will be tested by

acceptable knowledge of mathematics and statistics. Finally, the results will be concluded and

discussed by the author to be able to explain the findings in the context of the research.

APPROACH

The approach of the research can be considered both inductive and deductive, due to the lack

of evidence and no existing theory on the cap-weighted indexes or portfolios are efficient or

optimal portfolios this research will try to find out if fundamental portfolios (smart beta

portfolios) perform better consistently than cap-weighted portfolios. Promising results and

20

enough evidence can be used to formulate a theory proving or disproving the difference

between smart beta and cap-weighted portfolios.

The research uses experiment as the research strategy with mono method as a choice, as

historical data available on FTSE 100 components in October 2015 will be used to study

intensively. The research also only uses one mathematical model to calculate the risk and

return of the portfolios.

TIME HORIZON

The research uses cross-sectional time horizon for the collection of data, as all the data will

be collected at a particular point in time. Further, longitudinal time horizon is also considered

for the research as it looks at the performance difference of smart beta and cap-weighted

portfolios over the period of fifteen years.

DATA COLLECTION AND SAMPLING

The research is targeted towards FTSE 100, which is a cap-weighted index. For the collection

of data we first collected the list of FTSE 100 components ranked in order of their respective

market capitalizations for the year 2015 (Marketcapitalizations 2015). For the research we

created a sample using simple random selection and selected 10 companies at random (10%

of the total population). The purpose of random selection was resolved by using Microsoft

Excel rand function:

= 𝑅𝐴𝑁𝐷𝐵𝐸𝑇𝑊𝐸𝐸𝑁(0,100)

Using this function we got 10 random numbers between 0 and 100 and those numbers were

then matched with the companies in the population. Randomly selected companies are

presented in table 1 (Also See Appendix 4).

S. No Rank/Random number Company Ticker

1 2 HSBC HSBA

2 70 Aberdeen Asset Management ADN

3 55 Antofagasta ANTO

4 72 Babcock Intl Group PLC BAB

5 59 Capita PLC CPI

6 26 Compass Group CPG

7 14 Reckitt Benckiser Group PLC RB.

8 29 SSE PLC SSE

9 6 Vodafone Group VOD

10 56 Schroders PLC SDR TABLE 1: RANDOMLY SELECTED COMPANIES

The daily prices of these selected companies were downloaded from yahoo finance from 31st

December 2000 to 30th November 2015. We then allocated them in the following manner:

21

Period Time

31/12/00 T0

01/01/01-31/12/01 T1

01/01/02-31/12/02 T2

01/01/03-31/12/03 T3

01/01/04-31/12/04 T4

01/01/05-31/12/05 T5

01/01/06-31/12/06 T6

01/01/07-31/12/07 T7

01/01/08-31/12/08 T8

01/01/09-31/12/09 T9

01/01/10-31/12/10 T10

01/01/11-31/12/11 T11

01/01/12-31/12/12 T12

01/01/13-31/12/13 T13

01/01/14-31/12/14 T14

01/01/15-30/11/15 T15 TABLE 2: TIME PERIODS

Where T0 was the starting point of all the portfolios and last date of every year was the date

when the weights of the stocks were rebalanced. We will further discuss this in the scenario

section.

For the purpose of creating smart beta portfolios we downloaded historical yearly

fundamental data from year ending 2000 to 2014 using Bloomberg terminal. The measures of

company size used in the research were as follow:

Free cash flow

Revenue

Book value per share

Dividend per share

Earnings (Profit/Loss)

P/E ratio

Table 3 shows the fundamental data collection snapshot of Antofagasta.

Antofagasta (ANTO) Year Market cap. Free Cash Flow Revenue P/E ratio Net profit (loss) Dividends/Share Book Value/Share

2014 7,418,571.39 128.24 3,213.79 25.16 279.32 0.13 4.02

2013 8,123,459.31 239.64 3,819.13 20.40 421.85 0.61 4.12

2012 13,052,742.21 1,255.18 4,253.19 20.44 654.50 0.13 4.44

2011 11,978,158.37 1,136.04 3,789.72 15.03 771.29 0.12 4.05

2010 15,892,010.24 446.04 2,963.96 23.55 681.11 0.10 4.01

2009 9,779,698.20 -193.25 1,898.94 23.66 427.98 0.06 3.35

2008 4,194,820.16 450.85 1,840.85 3.58 931.45 0.05 3.67

2007 7,068,592.37 811.73 1,913.31 10.14 691.04 0.04 2.08

2006 5,018,010.78 1,009.13 2,101.10 7.25 735.28 0.04 1.65

2005 3,685,131.49 603.36 1,345.79 8.73 399.45 0.04 1.21

2004 2,210,290.20 620.87 1,060.14 7.31 316.33 0.04 0.78

2003 2,080,157.20 231.61 597.98 18.84 110.49 0.04 0.50

2002 1,232,320.59 161.12 574.17 19.17 64.40 0.04 0.60

2001 1,039,092.74 56.10 534.30 24.17 43.00 0.03 0.63

2000 872,610.00 -16.60 505.40 9.64 90.70 0.03 0.62

All Values in Million GBP except for per share values and ratios TABLE 3: FUNDAMENTAL DATA (ANTO)

22

In the similar fashion the data for other nine remaining companies was downloaded and

arranged (See Appendix 4).

DATA ANALYSIS AND EXPERIMENT

ASSUMPTIONS

For analyzing the data and conducting the experiment the research accompanies the following

assumptions:

The research assumes that the market only consists of ten securities and our sample

represents them.

Normal distribution of stock returns and log normal distribution of stock prices is also

assumed supported by vast literature (Cootner 1962; Fama 1965; Kendall 1953;

Narayan and Smyth 2006).

We also assume that the markets are efficient and there exists a market portfolio of

combination of securities where all rational investors should invest combined with

weight in a risk free asset according to their risk aversion.

The research does not account for any transaction fees, commission and taxes and

assumes them to be zero.

As the research keeps in mind that through these portfolios we are trying to replicate a

stock index we assumes that in the market short selling is prohibited.

The report also uses UK 10 year Gilt for risk free rate of 1.81% p.a. (Bloomberg

2015).

SCENARIO

For conducting this experiment to find the answer to the research question the report creates a

scenario. The report assumes that T0 dated 31st December 2000 is the starting point of all the

portfolios and from T1 to T15 we will combine the selected ten securities using weights

determined by market capitalization and fundamental values that are discussed in data

collection and sampling section. To determine the weights the report uses the following

formula:

𝑀𝑒𝑎𝑠𝑢𝑟𝑒 𝑜𝑓 𝐶𝑜𝑚𝑝𝑎𝑛𝑦 𝑠𝑖𝑧𝑒

Σ 𝑀𝑒𝑎𝑠𝑢𝑟𝑒 𝑜𝑓 𝑎𝑙𝑙 𝑐𝑜𝑚𝑝𝑎𝑛𝑦 𝑠𝑖𝑧𝑒 𝑋 100

Table 4 shows the snap shot of weights determined using market capitalization. In the similar

manner weights were determined for other measures of company size (See Appendix 4).

The rebalancing of the portfolios only took place once a year i.e. 31st December of every

year. The reason for only rebalancing once a year was after trying rebalancing quarterly and

semiannually we found no significant advance for returns over annual rebalancing and

moreover some of the fundamental data was also only available on annual basis.

For the purpose of experiment we determined the price of the portfolio at the starting point

i.e. T0 as the sum of prices of all ten securities. Then the returns of portfolios would be

calculated using the weights determined at T0 used in time T1 and so on and so forth. The

returns for each year for every portfolio then would be recorded and the value of the portfolio

23

would be then determined by multiplying the value of starting portfolio by the return for

every year’s portfolio.

𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑉𝑎𝑙𝑢𝑒 = 𝑃𝑥 × (1 + 𝑟𝑥)

Where 𝑃𝑥 is the price of portfolio at T-1 and 𝑟𝑥 is the return of portfolio at time T.

Weights using market capitalization

Time HSBA ADN ANTO BAB CPI CPG RB. SSE VOD SDR Total Weight

T15 42.44% 1.91% 2.69% 1.77% 2.60% 6.05% 13.58% 5.20% 21.13% 2.63% 100.00%

T14 40.15% 1.46% 2.62% 1.27% 2.20% 4.94% 11.11% 4.61% 29.39% 2.26% 100.00%

T13 41.62% 1.25% 4.55% 1.00% 1.72% 4.42% 9.72% 4.37% 29.77% 1.59% 100.00%

T12 35.49% 0.80% 4.84% 0.90% 1.56% 4.00% 9.37% 4.78% 36.82% 1.44% 100.00%

T11 42.75% 0.68% 5.90% 0.51% 1.58% 3.71% 9.50% 3.77% 29.70% 1.88% 100.00%

T10 49.38% 0.60% 3.91% 0.39% 1.88% 2.83% 9.67% 4.08% 25.78% 1.47% 100.00%

T9 38.08% 0.43% 1.99% 0.62% 2.17% 3.01% 8.68% 5.80% 38.09% 1.12% 100.00%

T8 43.66% 0.50% 3.10% 0.37% 1.86% 2.56% 9.10% 5.82% 31.39% 1.63% 100.00%

T7 47.53% 0.45% 2.22% 0.30% 1.70% 2.53% 7.46% 4.29% 32.09% 1.43% 100.00%

T6 45.46% 0.30% 1.58% 0.13% 1.17% 1.91% 5.98% 3.24% 39.02% 1.20% 100.00%

T5 45.66% 0.10% 1.03% 0.08% 1.14% 2.22% 5.13% 2.75% 40.86% 1.03% 100.00%

T4 47.94% 0.06% 1.04% 0.07% 0.81% 3.78% 4.46% 2.71% 38.23% 0.91% 100.00%

T3 36.45% 0.08% 0.69% 0.09% 0.93% 3.30% 4.75% 3.28% 49.59% 0.84% 100.00%

T2 34.05% 0.25% 0.47% 0.06% 1.46% 0.93% 2.84% 2.41% 56.39% 1.13% 100.00%

T1 29.27% 0.27% 0.26% 0.04% 0.99% 0.00% 1.76% 1.36% 64.88% 1.16% 100.00% TABLE 4: WEIGHTS DETERMINED BY MARKET CAPITALIZASTION

EXPERIMENTATION

The experiment uses the close price of the securities, for determining the price of the

portfolio at time T0 we used the closing prices of all ten securities on 31st December 2000.

Table 5 shows the calculation of the price.

Securities Price (T0)

HSBA 8.58

ADN 4.01

ANTO 4.43

BAB 0.92

CPI 4.97

CPG 7.29

RB. 9.01

SSE 6.20

VOD 1.40

SDR 13.21

Price of Portfolio (T0) 60.01

All prices in GBP TABLE 5: PRICE OF PORTFOLIO AT T0

After determining the price for the portfolio the daily close prices of the securities were

changed into returns using the logarithmic change from T1 to T15.

24

𝐷𝑎𝑖𝑙𝑦 𝑟𝑒𝑡𝑢𝑟𝑛 = 𝐿𝑁 (𝑃𝑡

𝑃𝑡−1⁄ )

The average of daily returns were multiplied by the number of trading days in that year,

which differed from 260-263 except for T15 where the trading days were 242 as we only took

data till 30th November 2015. This gave us the returns on the stock for every period from T1

to T15. These returns were then multiplied by their respective weights and added together to

give the return for the portfolio. We calculated the returns for each portfolio from T1 to T15

using the excel inbuilt function shown below:

𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑅𝑒𝑡𝑢𝑟𝑛 = 𝑆𝑈𝑀𝑃𝑅𝑂𝐷𝑈𝐶𝑇(𝑊𝑒𝑖𝑔ℎ𝑡𝑠, 𝑅𝑒𝑡𝑢𝑟𝑛𝑠)

For smart beta portfolio during time of construction the weights for free cash flow and profit

(loss) were sometimes negative for those instances the weights were adjusted to 0% as we

assume that there is no short selling of securities allowed.

For calculation of risk i.e. standard deviation, we used matrix multiplication on excel. The

daily returns were first used to create variance-covariance matrix. We created a VBA

function to do so and named it VarCov (Look Appendix 1 for function). The variance-

covariance matrix was constructed for every time period, then using the different weights in

different time period and portfolios we first multiplied the weights as a row matrix with

variance-covariance matrix and then multiplied weights as a column matrix to the result to

give us the variance of the portfolio.

[𝑊𝑒𝑖𝑔ℎ𝑡𝑠] [𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒

𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒𝑀𝑎𝑡𝑟𝑖𝑥

]

[ 𝑊𝑒𝑖𝑔ℎ𝑡𝑠 ]

= 𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒

The variance of the portfolio was then multiplied with the number of trading days to give the

variance for the time period and then the square root of the variance was taken to give the

measure of risk i.e. the standard deviation.

𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝐷𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 = √𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒 × 𝑁𝑜 𝑜𝑓 𝑡𝑟𝑎𝑑𝑖𝑛𝑔 𝑑𝑎𝑦𝑠

After the calculation of risk and return of the portfolios the experiment further undertakes the

Sharpe ratio (Sharpe 1994) for each portfolio from T1 to T15 to find out the return per unit of

risk for the portfolios. The UK 10 year Gilt and the following formula was used:

𝑆ℎ𝑎𝑟𝑝𝑒 𝑅𝑎𝑡𝑖𝑜 =𝑅𝑝 − 𝑅𝑓

𝜎𝑝

Where 𝑅𝑝 is the return on the portfolio, 𝑅𝑓 is the risk free rate and 𝜎𝑝 is the standard

deviation of the portfolio. After getting the Sharpe ratios of the portfolios we used the Solver

add-in on excel to find out the maximum Sharpe ratio portfolio with added constraints of no

negative weights and compared the maximum Sharpe ratio and our portfolios Sharpe ratio.

Snapshot of Solver are shown below:

25

FIGURE 4: MAXIMIZING SHARPE RATIO USING SOLVER

After finding out all the data the experiment further plots the mean-variance efficient

frontiers by Markowitz (1952) for all the portfolios and also plots the risk and returns of

maximum Sharpe ratio, Cap-weighted and smart beta portfolios. For plotting the efficient

frontier the report uses the Lagrange optimization technique to create minimum variance

frontier for n assets.

LAGRANGE OPTIMIZATION FOR N ASSETS

Mathematically the variance of portfolio can be calculated using the formula:

𝜎𝑝2 = Σ𝑖=1

𝑛 Σ𝑗=1𝑛 𝑥𝑖𝑥𝑗𝜎𝑖𝑗

Where 𝑥𝑖 𝑎𝑛𝑑 𝑥𝑗 are the weights of securities 𝑖 and 𝑗.

Since we are trying to create a minimum variance frontier, the objective was set to

𝑚𝑖𝑛𝑖𝑚𝑖𝑧𝑒 [𝜎𝑝2 = Σ𝑖=1

𝑛 Σ𝑗=1𝑛 𝑥𝑖𝑥𝑗𝜎𝑖𝑗] … Eq. (i)

The objective was then added with two constraints i.e. the desired return of the investor is

equal to the expected return and that the sum of the weights of all securities is 1. These can be

mathematically expressed as:

Σ𝑖=1

𝑛

𝑥𝑖𝑒 − 𝑑 = 0 … Eq. (ii)

Σ𝑖=1

𝑛

𝑥𝑖 − 1 = 0 …Eq. (iii)

Where, 𝑒 is the expected return and 𝑑 is the desired return. The objective and the two

constraints can then be used in the Lagrangian function and can be presented as:

𝑦 = Σ𝑖=1

𝑛Σ𝑗=1

𝑛𝑥𝑖𝑥𝑗𝜎𝑖𝑗 + 𝜆1 (Σ𝑖=1

𝑛𝑥𝑖𝑒 − 𝑑) + 𝜆2 (Σ𝑖=1

𝑛𝑥𝑖 − 1) … From (i), (ii) and (iii)

26

Where, 𝜆1 and 𝜆2 are Lagrangian multipliers used with two constraints. As we are trying to

minimize the variance of the portfolio we will take the partial derivative of the Lagrangian

function 𝑦 with respect to 𝑥𝑖, 𝜆1and 𝜆2 and set them to be equal to zero.

For making the minimum variance portfolio for ten assets we will demonstrate the

mathematical equations for two assets and then extrapolate them to be used for ten assets.

The 𝑦 function for two assets can be written as:

𝑦 = [𝑥12𝜎11 + 𝑥2

2𝜎22 + 2𝑥1𝑥2𝜎12] + 𝜆1[𝑥1𝑒1 + 𝑥2𝑒2 − 𝑑] + 𝜆2[𝑥1 + 𝑥2 − 1]

Where,

𝑥1 𝑎𝑛𝑑 𝑥2 are weights of the two securities respectively.

𝜎11 𝑎𝑛𝑑 𝜎22 are variance of the two securities respectively.

𝜎12 𝑖𝑠 𝑡ℎ𝑒 𝑐𝑜𝑣𝑎𝑟𝑎𝑖𝑛𝑐𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡ℎ𝑒 𝑡𝑤𝑜 𝑠𝑒𝑐𝑢𝑟𝑖𝑡𝑖𝑒𝑠.

𝑒1 𝑎𝑛𝑑 𝑒2 𝑎𝑟𝑒 𝑡ℎ𝑒 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑟𝑒𝑡𝑢𝑟𝑛𝑠 𝑜𝑓 𝑡𝑤𝑜 𝑠𝑒𝑐𝑢𝑟𝑖𝑡𝑖𝑒𝑠 𝑟𝑒𝑠𝑝𝑒𝑐𝑡𝑖𝑣𝑒𝑙𝑦.

As we have to minimize the variance we will have to take the partial derivative of the 𝑦

function with respect to 𝑥1, 𝑥2, 𝜆1, 𝜆2 and set them equal to zero. Equations can be written as:

1. 𝜕𝑦

𝜕𝑥1= 2𝑥1𝜎11 + 2𝑥2𝜎12 + 𝜆1𝑒1 + 𝜆2 = 0

2. 𝜕𝑦

𝜕𝑥2= 2𝑥2𝜎22 + 2𝑥1𝜎12 + 𝜆1𝑒2 + 𝜆2 = 0

3. 𝜕𝑦

𝜕𝜆1= 𝑥1𝑒1 + 𝑥2𝑒2 − 𝑑 = 0

4. 𝜕𝑦

𝜕𝜆2= 𝑥1 + 𝑥2 − 1 = 0

These four linear equations can solved by using matrix algebra, and can be written in a 𝐶𝑥 =𝑘 format. We can now extrapolate these equations and write the 𝐶 matrix for ten securities as:

𝐶 𝑀𝑎𝑡𝑟𝑖𝑥:

[ 2 × [

𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒

𝑀𝑎𝑡𝑟𝑖𝑥]

𝑒1

…𝑒10

111

1 1 1 0 0𝑒1 … 𝑒10 0 0]

Where, C matrix is a 12 × 12 square matrix.

𝑥 𝑉𝑒𝑐𝑡𝑜𝑟:

[ 𝑥1𝑥2…𝑥10

𝜆1

𝜆2 ]

27

𝑘 𝑉𝑒𝑐𝑡𝑜𝑟:

[ 00…01𝑑]

Where, 𝑥 𝑎𝑛𝑑 𝑘 are 12 × 1 column matrix. Now using these matrices we need to solve for

weights i.e. the x matrix.

𝐶𝑥 = 𝑘

𝐶𝑥 × 𝐶−1 = 𝑘 × 𝐶−1

𝐼𝑥 = 𝐶−1𝑘

To solve for x we multiplied both sides by 𝐶−1. Now solving 𝐶−1𝑘 we will get ten equations

for all ten weights for the stocks and we can choose the desired return to solve for the weight.

Below we show an example of all matrices and equations for revenue-weighted portfolio for

T2.

C Matrix:

0.0006 0.0003 0.0000 0.0002 0.0006 0.0005 -0.0001 0.0002 0.0007 0.0007 -0.1605 1

0.0003 0.0096 0.0002 0.0000 0.0004 0.0007 -0.0001 0.0002 0.0002 0.0006 -1.7263 1

0.0000 0.0002 0.0004 0.0001 0.0002 0.0002 0.0001 0.0000 0.0001 0.0001 0.1706 1

0.0002 0.0000 0.0001 0.0006 0.0003 0.0002 0.0000 0.0000 0.0002 0.0004 0.1120 1

0.0006 0.0004 0.0002 0.0003 0.0022 0.0008 -0.0001 0.0002 0.0011 0.0011 -0.6835 1

0.0005 0.0007 0.0002 0.0002 0.0008 0.0014 -0.0001 0.0002 0.0009 0.0009 -0.4451 1

-0.0001 -0.0001 0.0001 0.0000 -0.0001 -0.0001 0.0006 0.0000 -0.0002 -0.0002 0.1865 1

0.0002 0.0002 0.0000 0.0000 0.0002 0.0002 0.0000 0.0004 0.0003 0.0003 0.1086 1

0.0007 0.0002 0.0001 0.0002 0.0011 0.0009 -0.0002 0.0003 0.0025 0.0014 -0.4618 1

0.0007 0.0006 0.0001 0.0004 0.0011 0.0009 -0.0002 0.0003 0.0014 0.0022 -0.5030 1

1 1 1 1 1 1 1 1 1 1 0 0

-0.1605 -1.7263 0.1706 0.1120 -0.6835 -0.4451 0.1865 0.1086 -0.4618 -0.5030 0 0

x Matrix:

HSBA ADN

ANTO BAB CPI CPG

RB. SSE VOD SDR

For making 𝐶−1 matrix we used the inbuilt excel formula:

28

= 𝑀𝐼𝑁𝑉𝐸𝑅𝑆𝐸(𝐶 𝑀𝑎𝑡𝑟𝑖𝑥)

𝐶−1 Matrix:

3,296 -8 -174 -433 -313 -353 -384 -836 -376 -420 0.20 0.13

-8 68 39 49 -60 -99 -8 48 9 -38 0.04 -0.24

-174 39 2,500 -857 -65 -74 -767 -865 22 241 0.16 0.56

-433 49 -857 1,819 31 18 -252 -261 118 -232 0.10 0.15

-313 -60 -65 31 624 -220 59 243 -134 -164 0.04 -0.39

-353 -99 -74 18 -220 1,138 17 -120 -145 -164 0.01 -0.25

-384 -8 -767 -252 59 17 1,478 -361 119 99 0.25 -0.15

-836 48 -865 -261 243 -120 -361 2,277 -125 0 0.23 0.40

-376 9 22 118 -134 -145 119 -125 754 -243 0.00 -0.10

-420 -38 241 -232 -164 -164 99 0 -243 920 -0.02 -0.11

0.13 -0.24 0.56 0.15 -0.39 -0.25 -0.15 0.40 -0.10 -0.11 0.00 0.00

0.20 0.04 0.16 0.10 0.04 0.01 0.25 0.23 0.00 -0.02 0.00 0.00

Now the equations to find the minimum variance weights can be written as:

𝐻𝑆𝐵𝐴 = 0.20 + 0.13𝑑

𝐴𝐷𝑁 = 0.04 − 0.24𝑑

𝐴𝑁𝑇𝑂 = 0.16 + 0.56𝑑

𝐵𝐴𝐵 = 0.10 + 0.15𝑑

𝐶𝑃𝐼 = 0.04 − 0.39𝑑

𝐶𝑃𝐺 = 0.01 − 0.25𝑑

𝑅𝐵.= 0.25 − 0.15𝑑

𝑆𝑆𝐸 = 0.23 + 0.40𝑑

𝑉𝑂𝐷 = 0.001 − 0.10𝑑

𝑆𝐷𝑅 = −0.02 − 0.11𝑑

Now, substituting the desired return we can find the minimum variance weights and plot

them on to a graph to get minimum variance frontier. We used the similar technique for all

the portfolios and in the next section we will present our finding.

29

LIMITATIONS

The research methodology is accompanied by limitations such as the methodology assumes

that there are no commissions, transaction costs or taxes. The research is also only concerned

with stock included in FTSE 100 index for a period from 2001 to 2015 and moreover

overnight price changes are ignored. The methodology we have used only accounts for

certain measures for the size of the company, there are lot more measure that could have been

used such as profit margins, total assets, return on equity etc. The results we present in this

paper are only restricted to the selected fundamental and are only justifiable for these metrics.

Creating an index generally sets the starting price of the index to be 1000 by creating a

divisor:

𝑑𝑖𝑣𝑖𝑠𝑜𝑟 =Σ (𝑝𝑟𝑖𝑐𝑒𝑠 𝑜𝑓 𝑎𝑙𝑙 𝑠𝑒𝑐𝑢𝑟𝑖𝑡𝑖𝑒𝑠)

1000

Then the price is simply calculated by dividing the sum of prices of all securities divided by

the divisor as it gives the starting price of 1000. This research has ignored this concept and

just used the sum of all share prices as the starting price of the portfolio at T0.

The research methodology also only uses ten stocks for only a period of fifteen years due to

the limitation of time; larger data set would help to check the significance of the results. The

methodology also doesn’t account for any market proxy’s as we just compared portfolios

using different weighting styles, we did however created a portfolio using market

capitalization compare our portfolios. Another limitation is that only mathematical and

statistical measures for calculations were used and this was criticized by Dublin (2015).

BUDGET AND ETHICS

The research only uses secondary data moreover no interviews or any kind of human

interaction was involved in any form. All the data collected was freely available from yahoo

finance and Bloomberg so the research is not exposed to any unethical practices. The author

also ensures that no data manipulations were done during the research.

The research does not involved any direct costs as most of the journal articles, books and e-

books were available via locate and resource center (Library) at Coventry University London

Campus. Science Direct granted access to some articles and the author also enrolled for free

7-day access to institutional investor journals.

30

PERFORMANCE EVALUATION

Performance evaluation of the portfolios is not only done by Sharpe ratio but also with

arithmetic return, cumulative return and standard deviation. The risk adjusted performance

measure i.e. Sharpe ratio and other performance evaluation measure are calculated for each

portfolio over the period of examination from 1st January 2001 (T1) to 30th November 2015

(T15).

Firstly we present the arithmetic return of all our portfolios in a tabular form:

Time Market Cap Book Dividend FCF Revenue Earnings P/E

T1 -26.76% -21.73% -13.35% -8.88% -18.63% -18.64% -19.34%

T2 -33.46% -31.22% -17.61% -14.71% -24.19% -12.50% -51.11%

T3 20.13% 16.88% 12.90% 22.60% 18.95% 20.67% 15.70%

T4 1.44% 9.08% 13.00% 3.70% -1.94% 2.83% 5.80%

T5 0.08% -5.73% 15.53% 0.79% -0.08% 9.07% 15.32%

T6 6.43% 9.66% 18.89% 1.57% 9.04% 4.43% 19.95%

T7 7.47% 12.03% 9.78% 26.63% 4.81% -3.20% 10.51%

T8 -24.93% -28.09% -22.77% -29.02% -22.88% -23.70% -19.63%

T9 14.31% 29.47% 18.65% 19.06% 15.20% 15.39% 19.83%

T10 3.42% 15.99% 9.06% 13.48% 4.78% 5.93% 11.54%

T11 -12.68% -15.42% -7.88% -21.22% -8.35% -9.50% -8.93%

T12 9.11% 17.63% 17.81% 20.39% 11.52% 10.99% 19.56%

T13 16.72% 15.03% 17.84% 31.57% 16.01% 18.71% 12.91%

T14 13.50% 6.02% 6.70% 34.52% 13.63% -3.03% 3.29%

T15 -5.69% -2.66% 1.97% -6.66% -6.32% -2.65% -6.08%

Cumulative -28.19% 1.38% 90.98% 92.42% -3.46% 2.27% -5.47% TABLE 6: PORTFOLIO'S ARITHMETIC RETURNS

Comparing the portfolios using the arithmetic return we found that 64 out of 90 portfolios

weighted according to their fundamental values performed better than the portfolio that was

weighted according to the market cap over the time period. Given this, we found that 71% of

the smart beta portfolios outperformed the cap-weighted portfolio. Moreover, cumulative

returns for time period T0 to T15 all smart beta portfolios outperformed the cap-weighted

portfolio with significant difference. We show this graphically below:

FIGURE 5: CUMULATIVE RETURNS T0-T15

-40% -20% 0% 20% 40% 60% 80% 100%

Market Cap

Book

Dividend

FCF

Revenue

Earnings

P/E

Return

Cumulative Return

31

The cap-weighted portfolio over the time period lost 28.19% of its values whist in the same

time portfolios that were created using dividends and free cash flow (FCF) had returns of

90.98% and 92.42% respectively. These portfolios almost doubled the initial starting amount

i.e.T0. The portfolios that used Revenue and P/E ratio for weights, even though they had

negative returns they did not cumulatively lost as much value as cap-weighted portfolio.

Now we will present our findings comparing each of the smart beta portfolio returns with

cap-weighted portfolio returns.

FIGURE 6: CAP-WEIGHTED PORTFOLIO VS BOOK VALUE PER SHARE PORTFOLIO

There was no identifiable pattern between the two portfolios; the book value per share

portfolio does however has some significant return difference in T4, T9, T10 and T12 where

on average the returns were 3.74 times greater than the cap-weighted portfolio. T5 was the

only year where the book value portfolio has a negative return when the cap-weighted

portfolio had little but positive return. Cumulatively over the research time period book value

per share portfolio has outperformed the cap-weighted portfolio by 29.57-percentage point

(pp).

Except for two occasions i.e. T3 and T14 the dividend per share-weighted portfolio

outperformed the cap-weighted portfolio every year. The dividend-weighted portfolio

cumulatively outperformed the cap-weighted portfolio by substantial 119.17 pp. During the

research time period the dividend weighted portfolio always performed better when the

returns were negative. In fact, T1 and T2 were times during the financial crisis where the

dividend weighted portfolios only lost half as much as the cap-weighted portfolio. During the

time T4 and T5 FTSE 100 was recovering from the downfall (See Appendix 2), combining

these stocks using market capitalization during these time period only gave a return of 1.44%

in T4 and 0.08% in T5 whereas using dividends for weights had much greater return of 13%

and 15.53% respectively.

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15Ret

urn

s

Time Period

Cap-Weighted VS Book Value per Share

Portfolios

Market Cap

Book

32

FIGURE 7: CAP-WEIGHTED PORTFOLIO VS DIVIDEND PER SHARE PORTFOLIO

These similar results can be seen again after T8 (2008 crisis) as in T9 the dividend weighted

portfolio outperformed cap-weighted portfolio by 4.34 pp and in T10 by 5.64 pp. We observe

the same results again after T11 (2011 crisis); in T12 the dividend weighted portfolio

produced almost double the return by cap-weighted portfolio.

FIGURE 8: CAP-WEIGHTED PORTFOLIO VS FCF PORTFOLIO

Using FCF for weighting the portfolio created highest cumulative return of 92.42% for the

time period. There was no identifiable particular pattern between the cap-weighted and FCF

weighted portfolios but from T12 to T14 the FCF weighted portfolio continuously

outperformed the cap-weighted portfolio for three consecutive years and that enhanced the

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15

Ret

urn

s

Tme Period

Cap-Weighted VS Dividend per Share

Portfolios

Market Cap

Dividend

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15Ret

urn

Time Period

Cap-Weighted VS FCF Portfolios

Market Cap

FCF

33

turnover of the portfolio. Moreover, the FCF portfolio on average produced 2.27 times the

return on cap-weighted portfolio. The FCF weighted portfolio also was very volatile due to

the fact that, when the securities had negative FCF the weights for them were adjusted to be

zero.

FIGURE 9: CAP-WEIGHTED PORTFOLIO VS REVENUE WEIGHTED PORTFOLIO

The revenue weighted portfolio had a cumulative return of -3.46% compared to -28.19% of

cap-weighted portfolio. Even though the revenue-weighted portfolio cumulatively had a

negative return it performed 0.7 times or 24.72 pp better than cap-weighted portfolio over the

research time. During T4 and T5 i.e. when FTSE 100 started recovering the revenue-

weighted portfolio had negative returns whereas the cap-weighted portfolio had positive

returns (we will discuss this further in the analysis section). During bearish market times the

revenue-weighted portfolio didn’t lose as much as cap-weighted portfolio, which is why the

cumulative return of revenue-weighted portfolio is better when compared to cap-weighted

portfolio. Also, cap-weighted portfolio dropped to half of its value in T2 (this is presented

later in Table 7).

The Earnings-weighted portfolio outperformed the cap-weighted portfolio throughout the

time of experiment except for T6, T7 and T14. Out of these three time periods the earning-

weighted portfolio had a negative return when the cap-weighted portfolio had a positive

return in T7 and T14. Despite of having those two major downturns the earning-weighted

portfolio outperform the cap-weighted portfolio over research time by 30.46 pp. and on

average performed 8.32 times better than the cap-weighted portfolio. During bearish times

except for those two occasions (T7 and T14) the earnings-weighted portfolio always

outperformed the cap-weighted portfolio.

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%


Ret

urn

s

Time Period

Cap-Weighted VS Revenue Weighted Portfolio

Market Cap

Revenue

34

FIGURE 10: CAP-WEIGHTED PORTFOLIO VS PROFIT (LOSS) WEIGHTED PORTFOLIO

P/E ratio weighted portfolio had a negative cumulative return of -5.47%, which is the least of

all the fundamentally weighted portfolios. Despite of this the P/E ratio weighted portfolio

outperformed the cap-weighted portfolio over the research time. The P/E ratio weighted

portfolio lost 61% of its value by T2 but as it had some significant returns during T4, T5 and

T6. The portfolio ended up performing 14.24 times better than cap-weighted portfolio on

average.

FIGURE 11:CAP-WEIGHTED PORTFOLIO VS P/E RATIO WEIGHTED PORTFOLIO

We will now present the value (turnover) of the portfolios both graphically and in tabular

form.

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%


Ret

urn

s

Time Period

Cap-Weighted Portfolio VS Profit(loss)

weighted Portfolio

Market Cap

Earnings

-60.00%

-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%


Ret

urn

s

Time Period

Cap-Weighted Portfolio VS P/E Ratio

Weighted Portfolio

Market Cap

P/E

35

Portfolio Turnover (£’s) Market Cap Book Dividend FCF Revenue Earnings P/E

T0 (Starting Value) 60.03 60.03 60.03 60.03 60.03 60.03 60.03

T1 43.96 46.98 52.01 54.70 48.85 48.84 48.42

T2 29.25 32.31 42.85 46.65 37.03 42.73 23.67

T3 35.14 37.77 48.38 57.19 44.05 51.57 27.39

T4 35.65 41.20 54.67 59.30 43.19 53.02 28.98

T5 35.67 38.84 63.16 59.77 43.16 57.84 33.42

T6 37.97 42.59 75.09 60.71 47.06 60.40 40.08

T7 40.81 47.71 82.44 76.88 49.32 58.47 44.29

T8 30.63 34.31 63.67 54.57 38.04 44.61 35.60

T9 35.01 44.42 75.54 64.97 43.82 51.48 42.66

T10 36.21 51.52 82.39 73.73 45.91 54.53 47.58

T11 31.62 43.58 75.90 58.08 42.08 49.36 43.33

T12 34.50 51.27 89.41 69.93 46.92 54.78 51.81

T13 40.27 58.97 105.36 92.00 54.44 65.03 58.50

T14 45.71 62.52 112.42 123.75 61.86 63.06 60.42

T15 (Ending Value) 43.11 60.86 114.64 115.50 57.95 61.39 56.75 TABLE 7: PORTFOLIO TURNOVER

Table 7 presents our findings for returns in terms of portfolio turnover; we can see that the

value of portfolios that were weighted using fundamental values outperformed the cap-

weighted portfolio. Moreover, using dividends per share and FCF the value of portfolio

almost doubled in fifteen-year period of research.

FIGURE 12: PORTFOLIO TURNOVER

Figure 11 shows that only the P/E ratio weighted portfolio underperformed the cap-weighted

portfolio from T2 to T5 other than that all the smart beta portfolios outperformed the cap-

weighted portfolio. Over the fifteen-year period cap-weighted portfolio underperformed all

the fundamental portfolios in terms of arithmetic returns. We will now present our finding for

risk and Sharpe ratio of these portfolios.

0.00

20.00

40.00

60.00

80.00

100.00

120.00

140.00

T0 T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15

Turn

ove

r

Time Period

Portfolio Turnover

Market Cap Book Dividend FCF Revenue Earnings P/E

Time Market Cap Book Dividend FCF Revenue Earnings P/E ratio

𝑹𝒇: 1.81% Risk MSR Risk MSR Risk MSR Risk MSR Risk MSR Risk MSR Risk MSR

T1 40.22% -0.11 27.62% -0.07 21.11% -0.03 20.13% -0.02 31.64% -0.06 31.95% -0.07 32.03% -0.07

T2 39.04% -0.14 29.35% -0.10 20.10% -0.04 26.05% -0.04 28.56% -0.07 23.28% -0.03 33.28% -0.18

T3 19.38% 0.95 18.58% 0.81 14.52% 0.76 17.48% 1.19 18.24% 0.94 16.20% 1.16 22.21% 0.63

T4 13.45% -0.03 12.00% 0.61 9.61% 1.16 16.97% 0.11 13.39% -0.01 12.26% 0.08 14.20% 0.28

T5 10.48% -0.17 14.18% -0.01 9.01% 1.52 10.44% -0.10 9.99% 0.00 9.14% 0.79 10.51% 1.29

T6 14.75% 0.31 16.58% 0.47 11.62% 1.47 13.47% -0.02 13.23% 0.55 15.11% 0.17 15.18% 1.19

T7 16.46% 0.34 18.57% 0.55 14.81% 0.54 23.09% 1.07 16.26% 0.18 16.68% -0.01 18.73% 0.46

T8 36.96% -0.10 40.72% -0.12 34.14% -0.08 41.57% -0.13 37.40% -0.09 40.15% -0.10 37.28% -0.08

T9 28.40% 0.44 24.00% 1.15 16.70% 1.01 46.79% 0.37 31.99% 0.42 25.43% 0.53 18.18% 0.99

T10 18.36% 0.09 23.11% 0.61 19.08% 0.38 18.17% 0.64 16.05% 0.19 18.19% 0.23 18.04% 0.54

T11 20.83% -0.03 22.02% -0.04 17.55% -0.02 24.32% -0.06 18.43% -0.02 19.88% -0.02 21.60% -0.02

T12 13.74% 0.53 14.66% 1.08 12.63% 1.27 17.62% 1.05 13.06% 0.74 14.14% 0.65 14.32% 1.24

T13 14.59% 1.02 14.09% 0.94 13.34% 1.20 16.70% 1.78 13.45% 1.06 15.23% 1.11 13.23% 0.84

T14 22.08% 0.53 12.83% 0.33 11.97% 0.41 44.20% 0.74 18.92% 0.62 13.32% -0.01 11.85% 0.12

T15 17.27% -0.01 17.28% -0.01 16.43% 0.01 18.01% -0.02 16.46% -0.01 20.60% -0.01 16.69% -0.01 TABLE 8: PORTFOLIO RISK AND SHARPE RATIO

The time of research covers two major downturns in the financial markets. FTSE 100 saw strong bearish market times during T1, T2, T7 and T8.

These time periods had negative returns and since the research uses Sharpe ratio to compare the performance of the portfolios the Sharpe ratio

formula by Sharpe (1994) produces biased results at these times. Consider the time period T1; the cap-weighted portfolio (A) had a standard

deviation of 40.22% and a return of -26.76% (See Table 6 for returns) and the same year book value per share weighted portfolio (B) had

standard deviation of 27.62% and a return of -21.73%. Using the Sharpe ratio formula gives -0.71 and -0.85 respectively for portfolios A and B.

But given the data portfolio B not only had a lower risk but lost 5.03 pp less than portfolio A. Using the Sharpe (1994) formula gives portfolio A

advantage over portfolio B. Israelsen (2005) provides with similar example and provides a better way to scale the results when the excess returns

are negative. For the calculation of Sharpe ratio we used the modified Sharpe ratio formula (Israelsen 2005; Grable and Chatterjee 2014),

mathematically can be expressed as:

𝑀𝑜𝑑𝑖𝑓𝑖𝑒𝑑 𝑆ℎ𝑎𝑟𝑝𝑒 𝑅𝑎𝑡𝑖𝑜 (𝑀𝑆𝑅): 𝑅𝑖 − 𝑅𝑓

𝜎𝑖

(𝑅𝑖

𝐴𝐵𝑆(𝑅𝑖))

Where, 𝑅𝑖 is the return on asset and 𝑅𝑓 is the risk free rate. 𝜎𝑖 Is the standard deviation of

𝑅𝑖’s returns and 𝐴𝐵𝑆 (𝑅𝑖) is the absolute value of 𝑅𝑖.

The MSR gives the same result if the returns are positive as the term 𝑅𝑖

𝐴𝐵𝑆 (𝑅𝑖) equals 1, but if

the returns are negative then the same term scales the results and gets a value of -1. Now, we

will present our findings for the portfolios in graphical manner from T1 to T15

FIGURE 13: PORTFOLIOS RISK AND SHARPE RATIO: T1

Cap-weighted portfolio performed the worst in T1; it had the lowest level of return and MSR

with highest level of volatility.


-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

Risk MSR Risk MSR Risk MSR Risk MSR Risk MSR Risk MSR Risk MSR

Market Cap Book Dividend FCF Revenue Earnings P/E ratio

Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios

Portfolios Risk and Sharpe Ratio: T1

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


38

With the exception of P/E ratio weighted portfolio in T2, the cap-weighted portfolio had the

second lowest risk to return ratio. Cap-weighted portfolio however still maintained highest

level of volatility in T2.


Cap-weighted portfolio had the second highest standard deviation and comparatively

performed better than the P/E ratio weighted portfolio. Even though the FCF and earnings

weighted portfolios had lower level of risk and better return and Sharpe ratio.


0.00%

20.00%

40.00%

60.00%

80.00%

100.00%

120.00%

140.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


-20.00%

0.00%

20.00%

40.00%

60.00%

80.00%

100.00%

120.00%

140.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


39

Cap-weighted portfolio had second highest level of volatility whereas, dividend weighted

portfolio produced significant amount of returns compared to cap-weighted portfolio at lower

volatility level.


Cap-weighted portfolio produced the lowest risk adjusted return in T5. P/E ratio, Earning and

dividend weighted portfolios outperformed at a significant level on the basis of risk-adjusted

return with almost maintaining same volatility level.


Cap-weighted portfolio had a better risk adjusted return than two of the fundamentally

weighted portfolios i.e. FCF and earnings weighted portfolio. Smart beta portfolios using

-40.00%

-20.00%

0.00%

20.00%

40.00%

60.00%

80.00%

100.00%

120.00%

140.00%

160.00%

180.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


-20.00%

0.00%

20.00%

40.00%

60.00%

80.00%

100.00%

120.00%

140.00%

160.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


40

dividends, book value per share, P/E ratio and revenue performed much more efficiently

maintaining similar level of volatility.


In T7, the cap-weighted portfolio had a better risk adjusted return than earnings and revenue-

weighted portfolio. Other fundamental portfolios were however more efficient.



-20.00%

0.00%

20.00%

40.00%

60.00%

80.00%

100.00%

120.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


0.00%20.00%40.00%60.00%80.00%

100.00%120.00%140.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


41

In T8, the cap-weighted portfolio placed itself in between and had better risk adjusted return

than book, FCF and earnings weighted portfolio. While in T9 it outperformed revenue and

FCF weighted portfolios.




0.00%10.00%20.00%30.00%40.00%50.00%60.00%70.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


-10.00%

-5.00%

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


0.00%

20.00%

40.00%

60.00%

80.00%

100.00%

120.00%

140.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


42

All other smart beta portfolios on a risk-adjusted return basis in T10 and T12 outperformed

cap-weighted portfolio. While in T11 the cap-weighted portfolio outperformed FCF and book

value weighted portfolios but all other portfolios had a better risk-adjusted returns than cap-

weighted portfolio.


Cap-weighted portfolio in T13 had a better risk-adjusted return than book and P/E ratio

weighted portfolios. Also, in T13 all of the portfolios almost had same level of volatility.


T14 is the only time period where the cap-weighted portfolio outperformed four of the smart

beta portfolios. FCF and revenue weighted portfolios were the only two that were more

efficient than cap-weighted portfolio and had a better risk-adjusted return.

In T15, the risks and returns of the portfolios were very similar, dividend weighted portfolio

outperformed all other portfolios and had a positive Sharpe ratio.

0.00%

20.00%

40.00%

60.00%

80.00%

100.00%

120.00%

140.00%

160.00%

180.00%

200.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%



Sta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


43


To provide a more visual illustration of our finding we created mean-variance frontiers using

the theory of modern portfolio theory by Markowitz (1952). These illustrations are presented

in Appendix 3 for all our time periods.

PERFORMANCE DISCUSSION

Cumulatively, all the selected fundamental portfolios had better return than the cap-weighted

portfolios over the research time period (refer to Figure 4). Smart beta portfolios

outperformed the cap-weighted portfolios 71% of the times, moreover four out of fifteen

years all of the fundamental portfolios had better returns than cap weighted portfolio i.e. in

T1, T9, T10 and T12.

Time Period Maximum Sharpe Ratio Portfolio

T1 -0.02 FCF

T2 -0.03 Earnings

T3 1.19 FCF

T4 1.16 Dividend

T5 1.52 Dividend

T6 1.47 Dividend

T7 1.07 FCF

T8 -0.08 Dividend

T9 1.15 Book

T10 0.64 FCF

T11 -0.02 Dividend

T12 1.27 Dividend

T13 1.78 FCF

T14 0.74 FCF

T15 0.01 Dividend TABLE 9: MAXIMUM SHARPE RATIO

Table 9 shows that there was always one fundamental portfolio that outperformed the cap

weighted portfolio on a risk adjusted return basis in each time period. The portfolio using

-5.00%

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%


Market Cap Book Dividend FCF Revenue Earnings P/E ratioSta

nd

ard

Dev

iati

on

an

d M

SR

Portfolios


44

dividends for weighting outperformed all the other portfolios including the cap-weighted

portfolio seven out of fifteen times followed by FCF six out of fifteen times. We will now

compare the volatilities of fundamental portfolios with cap-weighted portfolio graphically.

FIGURE 28: MARKET CAP VS BOOK VOLATILITY

The book value per share-weighted portfolio maintained almost same levels of volatility

during the research time period but performed much better in terms of returns. Book value

per share portfolio did however have higher volatility level between T5 and T12 but the

difference in volatilities is significantly low.

FIGURE 29: MARKET CAP VS DIVIDEND VOLATILITY

Dividend per share-weighted portfolio almost doubled the initial value of portfolio while

maintaining lower level of risk all through the research period. Also Dividend weighted

portfolio maintained the lowest volatility in comparison with all other portfolios 73% of the

times. This fundamental portfolio during the research time period does proves to be more

mean-variance efficient than the cap-weighted portfolio and also contradicts the argument by

Malkiel (2014) about failing the risk test. Using dividends for weighting can although be

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

40.00%

45.00%


Ris

k

Time Period

Volatility: Market Cap VS Book

Market Cap

Book

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

40.00%

45.00%


Ris

k

Time Period

Volatility: Market Cap VS DividendMarket Cap

Dividend

45

tricky, as companies are not obligated to pay dividends and some companies like Google

never pay dividends accounting for those companies using dividends can be complicated.

FIGURE 30: MARKET CAP VS FCF VOLATILITY

The FCF weighted portfolio did however increased the overall risk as result of significant

difference between the free cash flow of the firms during the years and also FCF for the firms

could be negative and those negative weights were selected to be zero. Moreover, FCF for the

firms is not a stable unit of measure as companies use capital for expansion and other

operating expenses. Using FCF as a unit of company size measure can lead to a lot of

fluctuation in the weights of companies over time leading to increased volatility.

FIGURE 31: MARKET CAP VS REVENUE VOLATILITY

Revenue weighted portfolio also maintains lower level of volatility compared to cap-

weighted portfolio with exception of T9 and T10, where it had slightly higher volatility level.

This portfolio did have negative returns in T4 and T5 when cap-weighted portfolios portfolio

had positive returns. This can be argued that as companies having low revenues during the

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

40.00%

45.00%

50.00%


Ris

k

Time Period

Volatility: Market Cap VS FCFMarket Cap

FCF

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

40.00%

45.00%


Ris

k

Time Period

Volatility: Market Cap VS RevenueMarket Cap

Revenue

46

recession. However, This portfolio did perform better than cap-weighted portfolio and

supports our argument against failing the risk test.

FIGURE 32: MARKET CAP VS EARNINGS VOLATILITY

Earnings-weighted portfolio does contradict the findings of Malkiel (2014) as it maintains

lower volatility levels with the exception in T8 and T15.

FIGURE 33: MARKET CAP VS P/E RATIO VOLATILITY

Similar to earnings portfolio the P/E ratio weighted portfolio maintains lower volatility level

but does perform better than cap-weighted portfolio. Even though P/E ratio does considers

the price factor into account it still performed better than the cap-weighted portfolio.

Our findings point out that fundamental portfolios do produce better return, risk and Sharpe

ratios compared to cap-weighted portfolios. The mean-variance frontiers (Appendix 3)

clearly present that cap-weighted portfolios are not always the worse performing portfolio on

year on year basis but then again they are not the best performing portfolios either. In all the

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

40.00%

45.00%


Ris

k

Time Period

Volatility: Market Cap VS Earnings

Market Cap

Earnings

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

40.00%

45.00%


Ris

k

Time Period

Volatility: Market Cap VS P/E ratio

Market Cap

P/E ratio

47

time periods the cap-weighted portfolio was at least outperformed by two of the

fundamentally weighted portfolios. We can also argue that the fundamental portfolios are

able to successfully reduce the noise in the markets by considering the aspects that have

already been realized unlike market capitalization, which also looks at the future aspects that

might or might not come true.

The assumption promoted by the industry and various masters programs that’s cap-weighted

portfolios are mean-variance efficient is contradicted by this research as investors can have

much better mean-variance performance using the smart beta techniques. The mean-variance

frontiers (Appendix 3) do point out that alternative weighting techniques such as smart beta

can be used passively to outperform cap-weighted portfolios.

CONCLUSION

This research investigates whether smart beta portfolios represent better performance than

market capitalization weighted portfolios. Fama (1970) argued that the prices of securities

represent their fair and intrinsic values then, according to capital market theories investors

should invest in cap-weighted portfolios or indexes as market capitalization in that case

accurately reflects the intrinsic value of the stocks’ worth. A countless number of arguments

have been presented in the academic literature arguing about the presence of noise in

financial markets, major ones related to the concerned research are by Arnott and Hsu (2008)

and Seigel (2006). Arnott, Hsu and Moore (2005) argued that cap-weighted indexes don’t

stand on being mean-variance efficient in presence of noise in the markets. Investor over and

under reaction to certain news is present in the market, which leads to inefficiency in the

prices of securities, and so does the market capitalization. The investment industry and a

countless number of master’s level programs have proposed and promoted the assumption

that the cap-weighted indexes are mean-variance efficient, if we accept this assumption then

it simplifies the complication of constructing the optimal portfolio. We through this research

challenged the idea of cap-weighted portfolios being mean-variance efficient and conducted

an experiment back testing smart beta portfolios and compared them to cap-weighted

portfolio.

This research uses FTSE 100 constituents to construct fundamentals based market portfolios

whose construction method is based on weighting using fundamental metrics of company size

other than market capitalization. However, the research does include construction of market

portfolio using market capitalization for the purpose of comparing the portfolios. The size

metrics used in the research are book value per share, free cash flow, revenue, dividend per

share, P/E ratio and earnings. The research used 10% of the population as the sample and

tested the fundamentally weighted portfolios over fifteen-year period. The period of research

was filled with boom and bust cycles, FTSE 100 over the time period had a return of -8.28%.

Although, we use mono method in the research i.e. experiment (back-testing), we use a

number of measures to understand and analyze the performance of the portfolios. The

research uses arithmetic returns, risk, modified Sharpe ratio and mean-variance frontiers to

understand the findings of the performance of each and every portfolio. The findings of the

research were robust across the bullish and bearish market times as, cumulatively the returns

of fundamentally weighted portfolios outperformed the cap-weighted benchmark during the

research time period. Smart beta portfolios, over the research time outperformed the

benchmarked cap-weighted portfolio 71% of the times and moreover there was always a

fundamental portfolio that had a better risk-adjusted return than the cap-weighted portfolio in

each time period. The experiment also presented four years in the research time scale where

all the smart beta portfolios had better returns than cap-weighted portfolios i.e. T1, T9, T10

and T12.

Malkiel (2014) argued in his paper ‘Is smart beta really smart?’ that the excess returns

achieved by fundamental portfolios are a result of taking additional risk. The findings of this

research pointed out that smart beta portfolios 70% of the times had better risk adjusted

return than cap-weighted portfolios. This can be argued that the FCF weighted portfolio had

the highest volatility levels, but FCF of the chosen companies was some of the times

negative, those time the weights of securities were selected to be zero. That increased the

volatility of the portfolios. At the same time the dividend weighted portfolio had the lowest

49

level of volatility eleven times out of fifteen-year period i.e. 73% of the times. The dividend-

weighted portfolio, given had the least level of volatility, almost doubled the value of the

portfolios over the fifteen-year period. Two of our six smart beta portfolios i.e. revenue and

P/E ratio weighted portfolios, did have a negative return cumulatively, but when compared to

the cap-weighted benchmark even these portfolios had better returns over the years. Where

cap-weighted portfolio loss 28.19% of its values over the research time period the revenue

weighted portfolio only lost 3.46% and P/E ratio weighted portfolio lost 5.47%. Not to

mention that other portfolios did outperform the cap-weighted benchmark with a significant

level of return. The most significant returns were earned by dividends and FCF weighted

portfolios i.e. 90.98% and 92.42% respectively. These portfolios almost doubled in the

portfolio value over time.

The numerous arguments presented above conclude that this research does present findings in

line with the research of Arnott, Hsu and Moore (2005). The argument by Malkiel (2014) is

however contradicted by this research, as we found no evidence of excess return being earned

due to addition risk acquisition of the portfolios. Moreover the dividend-weighted portfolio

almost doubled its value while having least levels of volatility. The research also concludes

with challenging the mean-variance efficiency of the cap-weighted portfolios as pursued by

many in the industry and various masters programs. The research found that the cap-weighted

portfolios are not optimal portfolios and moreover they are not mean-variance efficient in

presence of noise in the market. Even though cap-weighted portfolio did outperform some of

the fundamental portfolios in efficiency but there was always a smart beta portfolio that

earned a better return at lower risk level. This research adds to the findings of Arnott, Hsu

and Moore (2005), Hemminki and Puttonen (2008) and Clare, Motson and Thomas (2013)

and stresses on the belief that broad indexes should be based on fundamental values of the

firms and not on prices.

“In short run, the market is a voting machine, but in the long run, it is a weighting machine.”

Benjamin Graham

50

RECOMMENDATION

After conducting the research and analyzing the results the research can be concluded to

provide advice to potential investors. The research indicates that use of fundamentals is a

better way to invest passively than cap-weighted indexes. The use of fundamental weighted

portfolios reduces the impact of noise in the market, enhances return and reduces the risk.

The research shows that fundamental portfolios are better mean-variance efficient than cap-

weighted portfolios over time and they are less affected by irrational traders and speculators.

On this basis the research suggest that broad indexes should be fundamentally weighted

rather than weighting them using market capitalization.

FURTHER RESEARCH

This research is solely depended on the fundamental metrics selected for the research over a

comparatively short period of time. There are many more metrics for calculation of

companies’ value and size. Moreover, the research is also only conducted on FTSE 100

constituents, which is only large-cap based index. We would like to pursue a further research

using all of the FTSE-All share index constitutes with similar fundamental metrics used in

this research and more such as, year on year growth of the companies’, debt to equity ratios,

profit margins etc. Also using different combinations of these fundamental metrics. We

would also like to possibly create a standardize form of combination of these fundamental

metrics where all the companies could be represented by size.

A similar further research using an inductive approach and all of the constituents and

ideology of fundamental metrics could be conducted in the future for a longer period of time

to be able to find more robust findings and creation of a strong theoretical framework.

51

REFLECTIVE LEARNING

Work place is one of the significant parts of education. It helps students discover the hidden

skills and also develop new ones. The experience gained can more effective after reflecting

on what has been achieved during that time period. This section will carry out my reflection

of the experiences and focus on my SWOT analysis before and after the internship and the

research. This section will also present a clear picture of my self-analysis and the skills I have

learned and skills that can be achieved.

Choosing and being selected to be a part of a internship program was a great opportunity for

me to develop my skills and working knowledge for the financial industry and be ready to hit

the ground running. My internship was with Shepherd Capital Holding International Ltd.,

which was a new start up firm looking at developing their business in the UK with their

Chinese clientele. I was asked to work alone and research on “Smart Beta” as my project for

internship.

RELATIONSHIP BETWEEEN INTERSHIP AND MY CAREER

I want to be an investor and do restructuring in next ten years time. I believe that I have a

good sense of judging the things that are holding the company to grow to its full potential. I

want to start up my own firm that invests in good and potential investments and also buy

companies, restructure them and sell them when they are profitable.

FIGURE 34: CAREER PLAN

This internship helps me follow my career path, as I will be working in the financial industry

and, not just the internship but also the research I was asked to do will help me achieve my

ultimate goal.

52

LEARNING OUTCOME

This internship was a great opportunity for me as I not only developed new skills but also

polished some of existing skills. This section will present my SWOT analysis before my

internship period and the skills developed in the ten-week internship.

Strengths

Determined to achieve goals

Well organized

Enthusiastic about technology and

financial markets

Good verbal communication

Good team player

Weaknesses

Proactive

Distractions

Writing skills

Opportunities

First class masters degree

Attractive in job market

Internship

Becoming a successful investor

Threats

Extremely high competition in job

markets especially due to visas.

TABLE 10: SWOT ANALYSIS BEFORE INTERSHIP

Working the project and with the company, I have learned many things about myself from a

different perspective and I will discuss them here.

Motivation

I discovered that little things, such as the atmosphere of the work, attitudes of my colleges,

motivate me. I got motivated to work when I found out my topic for research, also when I

conducted the research and I started to get results. The biggest motivation was when I spoke

to my first client as a professional and discussed his potential collaboration working with us.

Brooks (2009) presented leadership styles; the organization I was working for clearly

followed a democratic leadership style. We had meetings with the CEO and COO every other

week where we would discuss our progress and share ideas for the business. One of the key

motivations was that everyone was allowed to pitch their ideas and then everyone would give

feedback. Moreover, you could directly get in touch with the CEO and the COO with any

problems and they would reply as quickly as they could. This was on the key motivation as

you could get direct feedback from the CEO.

Learning Style

Using Honey and Mumford learning styles (Rosewell 2005), I discovered that I have qualities

of all four learning types but predominantly I am an activist. I learn the best when I do stuff

myself. A great example is when I learned that the company was planning to look for client

for portfolio management. I started looking for clients myself and also started looking for the

procedures, industry average fees and charges etc. I finally found a client who was ready to

invest with the company and wanted us to manage his funds. This experience not only taught

53

me my learning style but also more about the investment and portfolio industry. This

experience was very useful for my research as it was based on Smart beta portfolios and

indexes.

Learning from Research

Doing an internship provides steep learning curve for skills however, most of my work in the

internship was to do research and that is where I developed and enhanced some of my skills.

Excel

The research I conducted required a lot of work to be done on Excel, the time I started

conducting my research I faced some difficulties using some of the models and as the data set

was comparatively bigger I had to find smarter ways to conduct research. To make the

research a little bit easier I started learning VBA and started designing my functions to

calculate the answers to my problems. I finally succeeded and created two very useful

functions that made conducting my research comparatively easier.

Bloomberg Terminal

We were taught how to use Bloomberg during the first three terms, this was the time to use

Bloomberg terminal to actually find results to the research question. I started watching the

Bloomberg news on a daily basis. The advantage of this was during the news they would tell

some new Bloomberg functions that came to be very useful for the research. These functions

first of all helped me find accurate data for the research and also I ran simulations of my

research on Bloomberg to test my results.

As I was working alone on the whole project myself the only feedback I could receive was

from my supervisor. To keep myself motivated and to make sure I was moving in the right

direction I made a development plan for the research using Tuckman 1965 model (Wilson

2010).

FORMING STORMING NORMING PERFORMING

I was assigned

with the topics I

had to research

for the final

work, but there

was no clear

direction of

work and I had

to find the

research

question and

objectives.

I started doing

research in the

topic and started

looking for

previous

researches done

on the same

topic. As the

topic was new to

the industry not

many researches

were available.

After carefully

researching the

topic I decided

my research

question and

objectives. I

used the wall of

my room to

stick some white

sheets and

started writing

all ideas on it.

After writing all

the ideas I

finally

discovered my

methodology

and the direction

to which I

wanted my

research to go. I

also used this

same technique

to analyze my

finings.

54

Writing and Research

Through the mode of this research the writing and research skills were one of the most

affected skills. Firstly, I had to review a significant amount of well-documented research

literature. Reading and analyzing these research paper was hard in the beginning but after I

started reading and understanding those papers the idea of my research started becoming

clear. I was able to critically analyze and review those papers and also develop my theoretical

framework. Not only it enhanced by analytical and reading skills reviewing those papers also

helped me understand how to pursue the writing of my own research. I learned how to create

links while writing and use more professional and academic writing style.

Time Management

Time management is one of the most important skills in any business. I was doing an

internship and I had to write a report. I also have a part time job, managing all this work was

very important to be able to achieve my goals. I managed my work and internship timings by

speaking to my manager at my part time job so they don’t conflict with the timings of my

internship dedicated hours. For doing the research I used the Gantt chart to spread out my

work over ten week period and I aimed to strictly follow it. Making a Gantt chart helped a lot

and using that I was able to achieve all the targets in time.

Figure 28 shows my Gantt chart that I created to conduct this research.

FIGURE 35: GANTT-CHART

12-Oct-15 22-Oct-15 1-Nov-15 11-Nov-15 21-Nov-15 1-Dec-15 11-Dec-15 21-Dec-15

Pre-reseach of topic

Reviewing the research

Collection of data

Creating market portfolios

Creating fundamental portfolios

Analysis of data

Planing the structure of the report

Wrting the report

Making the power point presentation

Research presentation

Implemeting the feedback from presentation

Review

Formatting

Proof Reading

Pre-reseachof topic

Reviewingthe research

Collection ofdata

Creatingmarket

portfolios

Creatingfundamental

portfolios

Analysis ofdata

Planing thestructure ofthe report

Wrting thereport

Making thepower pointpresentation

Researchpresentation

Implemetingthe feedback

frompresentation

ReviewFormattingProof

Reading

Start date 12-Oct-1514-Oct-1526-Oct-1531-Oct-155-Nov-1512-Nov-1517-Nov-1519-Nov-159-Dec-1511-Dec-1512-Dec-1514-Dec-1516-Dec-1518-Dec-15

days to complete 2125575220212223

Gantt Chart

55

CHALLENGES

During this research and internship period I was faced by two major challenges that was

multi-taking and demotivation. I had to work at two places while at the same time do the

research. This led to a distraction and was followed by demotivation. To concur these

challenges I planned group study sessions with my fellow classmate and my best friend, this

helped us a lot as we were both doing research in finance. We did various brain storming

sessions that helped us both focus and get an outside perspective for out thoughts.

CONCLUSION

Overall the experience of internship and conducting this research was one of the best

experiences at the university. I learned about various new topic related to my field of study

and enhanced a number of skills. I created a SWOT analysis after this research, which is

presented below.

Strengths

Strong verbal skills

Determination to achieve goals

Organization

Time management

Analytical skills

Team player

Learning quickly

Eager to learn new skills (Technology)

Weaknesses

Distraction

Weak commercial knowledge

Programming skills need to be

improved

Opportunities

Achieving a first class masters degree

Becoming a successful investor

Improved CV

Start Investing

Threats

High competition due to visa

Lack of funds to invest

TABLE 11: SWOT ANALYSIS AFTER RESEARCH

REFERENCES

Arnott, R., Hsu, J. (2008) “Noise, CAPM And The Size And Value Effects”. Journal of

Investment Management [online] 6 (1) 1-11. Available from

<https://www.researchaffiliates.com/Production%20content%20library/JOIM_First_Quarter_

2008_Noise_CAPM_and_the_Size_and_Value_Effects.pdf> [1 Nov 2015]

Arnott, R., Shepherd, S.D. (N.D.) “The Fundamental Index ® Concept in Emerging

Markets”. [online] Available from

<http://www.invescopowershares.co.uk/ps/global/UK/literature/FIinEmergingMarkets.pdf>

[20 Nov 2015]

Arnott, R.D., Hsu, J., Moore, P. (2005) “Fundamental Indexation”. Financial Analysts

Journal [online] 61 (2) 83-99. Available from

<https://www.researchaffiliates.com/Production%20content%20library/FAJ_Mar_Apr_2005

_Fundamental_Indexation.pdf> [2 Nov 2015]

Bloomberg (2015) “United Kingdom Government Bonds” [online] available from

<http://www.bloomberg.com/markets/rates-bonds/government-bonds/uk> [10 Dec 2015]

Brooks,I. (2009) ‘Organizational Behavior: Individual, Group and Organisation.’ 4th edn.

Essex: Pearson Education Limited.

Carhart, M.M. (1997) “On Persistence in Mutual Fund Performance”. Journal of Finance

[online] 52 (1) 57-82. Available from

<http://faculty.chicagobooth.edu/john.cochrane/teaching/35150_advanced_investments/Carh

art_funds_jf.pdf> [19 Nov 2015]

Chow, T., Hsu, J., Kalesnik, V., Little, B. (2011) “A Survey of Alternative Equity Index

Strategies”. Financial Analysts Journal. [online] 67 (5) 37- 57.Avialable from

<https://www.researchaffiliates.com/Production%20content%20library/FAJ_SeptOct_2011_

A_Survey_of_Alternative_Equity_Index_Strategies.pdf> [19 Nov 2015]

Clare, A., Motson, N., Thomas, S. (2013) “An evaluation of alternative equity indices”

[online] available from < http://www.cass.city.ac.uk/news-and-

events/news/2013/april/monkeys-beat-market-cap-indices> [15 Oct 2015]

Cootner, P.H. (1962) “Stock Prices: Random vs. Systematic Changes”. Industrial

Management Review [online] 3 24-45. Available from < http://www.e-m-h.org/Coot62.pdf>

[30 Oct 2015]

De Bondt, W.F.M., Thaler, R. H. (1985) “Does the Stock Market Overreact?”. The Journal of

Finance [online] 40 (3) 793-805. Available from

<http://efinance.org.cn/cn/fm/Does%20the%20Stock%20Market%20Overreact.pdf> [1 Nov

2015]

De Bondt, W.F.M., Thaler, R. H. (1987) “Further Evidence on Investor Overreaction and

Stock Market Seasonality”. The Journal of Finance [online] 42 (3) 557-581. Available from

< https://driehaus.depaul.edu/about/centers-and-institutes/driehaus-center-for-behavioral-

finance/publications-and-resources/Documents/233.pdf> [2 Nov 2015]

57

Dubil, R. (2015) “How Dumb Is Smart Beta? Analyzing the Growth of Fundamental

Indexing.” Journal of Financial Planning [online] 28 (3) 49-54. Available from

<http://web.b.ebscohost.com/bsi/pdfviewer/pdfviewer?sid=dd4bd88e-6eca-465a-a288-

3385e7faad19%40sessionmgr111&vid=20&hid=107> [21 Nov 2015]

Fama, E.F. (1965) “Random Walks in Stock Market Prices”. Financial Analysts Journal


<http://web.b.ebscohost.com/bsi/pdfviewer/pdfviewer?sid=1b22a059-1864-4798-ac30-

1acf224cf8c4%40sessionmgr115&vid=36&hid=105> [3 Dec 2015]

Fama, E.F. (1970) “Efficient Capital Markets: A review of Theory and Empirical Work”.

Journal of Finance [online] 25 (2) 383-417. Available from < http://www.e-m-

h.org/Fama70.pdf> [21 Oct 2015]

Fama, E.F., French, K.R. (2009) Why Active Investing Is a Negative Sum Game. [online]

Available from <https://www.dimensional.com/famafrench/essays/why-active-investing-is-a-

negative-sum-game.aspx> [5 Oct 2015]

Fishe, R.P.H., Gosnell, T.F., Lasser, D.J. (1993) “Good News, Bad News, Volume, And The

Monday Effect.” Journal of Business Finance & Accounting [online] 20 (6) 881-892.

Available from < http://web.a.ebscohost.com/bsi/pdfviewer/pdfviewer?sid=cbcac316-f11b-

4abd-b6cb-981a2081b2e0%40sessionmgr4001&vid=96&hid=4107> [6 Oct 2015]

Glushkov, D. (2015) “How Smart are “Smart Beta” ETFs? Analysis of Relative Performance

and Factor Exposure.” [online] available from

<http://ecgi.ssrn.com/delivery.php?ID=3320921170880031061041071170901051210260110

680810170860050051070780921220780921021080630550590520580250620750270720220

810040890410360130810920910690131211031241110310460851070310921211181151200

65100081005085017110115111092002003001115127024123103000&EXT=pdf> [16 Oct

2015]

Grable, J.E., Chatterjee, S. (2014) “The Sharpe Ratio and Negative Excess Returns: The

Problem and Solution.” Journal of Financial Service Professionals [online] 68 (3) 12-13.

Available from < http://web.a.ebscohost.com/bsi/pdfviewer/pdfviewer?sid=4d3c7af3-c99e-

45fe-9396-6d52684fd8f1%40sessionmgr4001&vid=16&hid=4109> [11 Dec 2015]

Graham, B. (1949) ‘The Intelligent Investor’. Revised edn. New York: HaperCollins

Haugen, R.A., Baker, N.L. (1991) “The efficient market inefficiency of capitalization-

weighted stock portfolios.” “Journal of Portfolio Management” [online] 17 (3) 35-40.

Available from < http://www.efalken.com/LowVolClassics/HaugenBaker991.pdf> [3 Oct

2015]

Hemminki, J., Puttonen, V. (2008) “Fundamental Indexation in Europe”. Journal of Asset

Management [online] 8 (6) 401-405. Available from



58

Hsieh, H. (2013) “Unlocking The Secrets of Fundamental Indexes: Size Effect or Value

Effect? Evidence From Emerging Stock Markets” Investment Management and Financial

Innovations [online] 10 (4) 48-63. Available from

<http://businessperspectives.org/journals_free/imfi/2013/imfi_en_2013_04_Hsieh.pdf> [23

Nov 2015]

Hsu, J.C. (2006) “Cap-Weighted Portfolios Are Sub-Optimal Portfolios”. Journal of

Investment Management [online] 4 (3) 1-10. Available from

<https://www.researchaffiliates.com/Production%20content%20library/JOIM_Third_Quarter

_2006_Cap-Weighted_Portfolios_are_Sub-Optimal_Portfolios.pdf> [2 Nov 2015]

Hsu, J.C., Campollo, C. (2006) “New Frontiers In Index Investing”. Journal of Indexes

[online] January/February 32-58. Available from

<https://www.researchaffiliates.com/Production%20content%20library/JOI_Jan_Feb_2006_

New_Frontiers_In_Index_Investing_An_Examination_of_Fundamental_Indexation.pdf> [1

Nov 2015]

Invesco (2015) ‘Study: Smart Beta ETFs Poised for Continued Growth’. [online] available

from

<https://www.invesco.com/static/us/investors/contentdetail?contentId=8a6b021101cc3410Vg

nVCM100000c2f1bf0aRCRD&dnsName=us> [14 Oct 2015]

Israelsen, C.L. (2005) “A refinement to the Sharpe ratio and information ratio.” Journal of

Asset Management [online] 5 (6) 423-427. Available from

<http://web.a.ebscohost.com/bsi/pdfviewer/pdfviewer?sid=4d3c7af3-c99e-45fe-9396-

6d52684fd8f1%40sessionmgr4001&vid=13&hid=4109> [10 Dec 2015]

Kaplan, P.D. (2008) “Why Fundamental Indexation Might--or Might Not--Work”. Financial

Analysts Journal. [online] 64 (1) 32-39. Available from



Kendall, M.G. (1952) “The Analysis of Economic Time-Series – Part 1: Prices”. Journal of

Royal Statistical Society [online] 116 (1) 11-25. Available from < http://ww.e-m-

h.org/KeHi53.pdf> [3 Dec 2015]

Krause, A. (2001) “An Overview of Asset Pricing Models”. [online] Available from <

http://people.bath.ac.uk/mnsak/Research/Asset_pricing.pdf> [3 Oct 2015]

Malkiel, B.G. (2014) “Is Smart Beta Really Smart?” The Journal of Portfolio Management


<http://www.iijournals.com/doi/pdfplus/10.3905/jpm.2014.40.5.127> [15 Oct 2015]

Marketcapitalizations (2015) “UK/FTSE 100 Companies – Historical Market Caps” [online]

available from < http://marketcapitalizations.com/historical-data/market-caps-uk-

companies/> [23 Oct 2015]

Markowitz, H. (1952) “Portfolio Selection”. The Journal of Finance [online] 7 (1) 77-91.

Available from

<https://www.math.ust.hk/~maykwok/courses/ma362/07F/markowitz_JF.pdf> [20 Oct 2015]

59

Narayan, P.K., Smyth, R. (2006) “ Random Walk Versus Multiple Trend Breaks in Stock

Prices: Evidence From 15 European Markets”. Applied Financial Economics Letters [online]

2 (1) 1-7. Available from

<http://web.b.ebscohost.com/bsi/pdfviewer/pdfviewer?sid=1b22a059-1864-4798-ac30-

1acf224cf8c4%40sessionmgr115&vid=45&hid=105> [5 Dec 2015]

NASDAQ (2015) Beta. [online] Available from

<http://www.nasdaq.com/investing/glossary/b/beta> [4 Oct 2015]

Roll, R. (1976) “A Critique of The Asset Pricing Theory’s Tests Part 1: On Past and Potential

Testability of The Theory”. Journal of Financial Economics [online] 4 (2) 129-176.

Available from < http://schwert.ssb.rochester.edu/f532/JFE77_RR.pdf> [14 Oct 2015]

Roll, R. (1978) “Ambiguity When Performance Is Measured By The Security Market Line”.

Journal of Finance [online] 33 (4) 1051-1069. Available from <

http://web.b.ebscohost.com/bsi/pdfviewer/pdfviewer?sid=bb773f0d-e1a5-4170-9c8f-

7e74879aade3%40sessionmgr120&vid=6&hid=107> [20 Oct 2015]

Roncalli, T. (2014) ‘Introduction to Risk Parity and Budgeting’. Oxfordshire: CRC Press,

Taylor & Francis Group

Rosewell, J. (2005) “ Learning Styles” [online] available from

<http://www.open.edu/openlearnworks/pluginfile.php/69355/mod_page/content/1/learning_st

yles.pdf> [18 Dec 2015]

Saunders, M., Lewis, P. (2012) Doing Research in Business & Management: An Essential

Guide to Planning Your Project. England: Pearson Education Limited

Schoenfeld, S.A. (2006) “Are Alternatively Weighted Indexes Worth Their Weights?”

[online] available from < http://www-

ac.northerntrust.com/content/media/attachment/data/white_paper/0605/document/altweighted

_indexes.pdf> [20 Oct 2015]

Sharpe, W.F. (1975) “Adjusting for risk in portfolio performance measurement”. The journal

of Portfolio Management [online] 1 (2) 29-34. Available from

<http://www.iijournals.com/doi/pdfplus/10.3905/jpm.1975.408513> [25 Oct 2015]

Sharpe, W.F. (1991) "The Arithmetic of Active Management." Financial Analysts Journal


<http://web.a.ebscohost.com/bsi/pdfviewer/pdfviewer?sid=cbcac316-f11b-4abd-b6cb-

981a2081b2e0%40sessionmgr4001&vid=58&hid=4107> [2 Oct 2015]

Sharpe, W.F. (1994) “The Sharpe Ratio”. The Journal of Portfolio Management [online] 21

(1) 49-58. Available from

<http://www.iijournals.com/doi/pdfplus/10.3905/jpm.1994.409501> [20 Oct 2015]

Siegel, J.J. (2006) “The ‘Noisy Markey’ Hypothesis”. The Wall Street Journal [online]

available from

<http://www3.nd.edu/~mbamicro/datafiles/articles/The%20'Noisy%20Market'%20Hypothesi

s.pdf> [20 Oct 2015]

60

Tobin, J. (1958) “Liquidity Preference as Behavior Towards Risk”. The Review of Economic

Studies [online] 25 (2) 65-86. Available from < http://web.uconn.edu/ahking/Tobin58.pdf>

[15 Oct 2015]

Wilson, C. (2010) “Bruce Tuckman’s Forming, Storming, Norming & Performing Team

Development Model” [online] available from

<http://www.sst7.org/media/BruceTuckman_Team_Development_Model.pdf> [18 Dec

2015]

61

APPENDIX

APPENDIX 1: VARCOV VBA FUNCTION

Function VarCov(rng As Range) As Variant

Dim i As Integer

Dim j As Integer

Dim column As Integer

Dim matrix() As Double

column = rng.Columns.Count

ReDim matrix(column - 1, column - 1)

For i = 1 To column

For j = 1 To column

matrix(i - 1, j - 1) = Application.WorksheetFunction.Covar(rng.Columns(i), rng.Columns(j))

Next j

Next i

VarCov = matrix

End Function

APPENDIX 2: FTSE 100 TIME REFERENCE

FIGURE 36: FTSE 100 TIME REFERENCE

0.00

1,000.00

2,000.00

3,000.00

4,000.00

5,000.00

6,000.00

7,000.00

8,000.00

T0 T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15

Pri

ce

Time Period

FTSE 100 Time Reference

FTSE 100

62

APPENDIX 3: MEAN-VARIANCE FRONTIER

T1

FIGURE 37: MEAN-VARIANCE FRONTIER: T1

T2


-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 10.00% 20.00% 30.00% 40.00% 50.00%

Re

turn

Risk

Mean-Variance frontier: T1

Market Cap Book Dividend FCF

Revenue Earnings P/E Max Sharpe ratio

-60.00%

-40.00%

-20.00%

0.00%

20.00%

40.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00% 35.00% 40.00% 45.00%Re

turn

Risk

Mean-Variance Frontier: T2


Revenue Earnings P/E Max Sharpe

63

T3


T4


-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00% 35.00%

Re

turn

Risk



Revenue Earnings P/E Max Sharpe Ratio

-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00%

Re

turn

Risk




64

T5


T6


-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00%

Re

turn

Risk




-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00%

Re

turn

Risk




65

T7


T8


-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00%

Re

turn

Risk




-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00% 35.00% 40.00% 45.00% 50.00%

Re

turn

Risk




66

T9


T10


-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00% 35.00% 40.00% 45.00% 50.00%

Re

turn

Risk

Mean-VAriance Frontier: T9



-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00%

Re

turn

Risk




67

T11


T12


-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00%

Re

turn

Risk




-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00%

Re

turn

Risk




68

T13

FIGURE 49:MEAN-VARIANCE FRONTIER: T13

T14


-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 2.00% 4.00% 6.00% 8.00% 10.00% 12.00% 14.00% 16.00% 18.00% 20.00%

Re

turn

Risk




-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00% 35.00% 40.00% 45.00% 50.00%

Re

turn

Risk

MEan-Variacne Frontier: T14



69

T15


-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

0.00% 5.00% 10.00% 15.00% 20.00% 25.00%

Re

turn

Risk




70

APPENDIX 4: FINANCIAL DATA AND WEIGHTS

FIGURE 52: CAP-WEIGHTED SAMPLE SELECTION

71

FIGURE 53: MARKET CAP DATA AND WEIGHTS

FIGURE 54: FREE CASH FLOW DATA AND WEIGHTS

72

FIGURE 55: REVENUE DATA AND WEIGHTS

FIGURE 56: P/E RATIO DATA AND WEIGHTS

73

FIGURE 57: PROFIT (LOSS) DATA AND WEIGHTS

FIGURE 58: DIVIDENDS DATA AND WEIGHTS

74

APPENDIX 5: ETHICS APPROVAL CHECKLIST

Low Risk Research Ethics Approval Checklist

Applicant Details

Name: Hyder Ali Khan E-mail: [email protected]

Department: Finance Date: 19 December 2015

Course: MSc. Global Financial Trading Title of Project: Are Smart Beta Portfolios Smarter than Market Capitalisation-Weighted Portfolios?

Project Details

The research bask tests fundamentally weighted portfolios and compares them to market-capitalization weighted portfolio to check the efficiency and any enhancement in returns.

Research Objectives:

Identify if smart beta portfolios produce better cumulative returns than cap-weighted portfolio over the

research time period.

Identify weather using smart beta techniques lead to increased volatility levels in order to achieve

better return.

Test weather smart beta portfolios produce better return and risk adjusted return on year on year basis

compared to cap-weighted portfolios.

Determine if the cap-weighted portfolios are more efficient portfolios than fundamental portfolios.

Research Design: Experiment (Back-testing)

Methods of Data Collection: Bloomberg-terminal, Yahoo finance.

Participants in your research

Will the project involve human participants? Yes No

If you answered Yes to this questions, this may not be a low risk project.

If you are a student, please discuss your project with your Supervisor.

If you are a member of staff, please discuss your project with your Faculty Research Ethics Leader or use the Medium to High Risk Ethical Approval or NHS or Medical Approval Routes.

75

Risk to Participants

Will the project involve human patients/clients, health professionals, and/or patient (client) data and/or health professional data?

Yes No

Will any invasive physical procedure, including collecting tissue or other samples, be used in the research?

Yes No

Is there a risk of physical discomfort to those taking part? Yes No

Is there a risk of psychological or emotional distress to those taking part? Yes No

Is there a risk of challenging the deeply held beliefs of those taking part? Yes No

Is there a risk that previous, current or proposed criminal or illegal acts will be revealed by those taking part?

Yes No

Will the project involve giving any form of professional, medical or legal advice, either directly or indirectly to those taking part?

Yes No

If you answered Yes to any of these questions, this may not be a low risk project.

If you are a student, please discuss your project with your Supervisor.

If you are a member of staff, please discuss your project with your Faculty Research Ethics Leader or use the Medium to High Risk Ethical Approval or NHS or Medical Approval Routes.

Risk to Researcher

Will this project put you or others at risk of physical harm, injury or death? Yes No

Will project put you or others at risk of abduction, physical, mental or sexual abuse?

Yes No

Will this project involve participating in acts that may cause psychological or emotional distress to you or to others?

Yes No

Will this project involve observing acts which may cause psychological or emotional distress to you or to others?

Yes No

Will this project involve reading about, listening to or viewing materials that may cause psychological or emotional distress to you or to others?

Yes No

Will this project involve you disclosing personal data to the participants other than your name and the University as your contact and e-mail address?

Yes No

Will this project involve you in unsupervised private discussion with people who are not already known to you?

Yes No

Will this project potentially place you in the situation where you may receive unwelcome media attention?

Yes No

Could the topic or results of this project be seen as illegal or attract the attention of the security services or other agencies?

Yes No

Could the topic or results of this project be viewed as controversial by anyone? Yes No

If you answered Yes to any of these questions, this is not a low risk project. Please:

If you are a student, discuss your project with your Supervisor.

If you are a member of staff, discuss your project with your Faculty Research Ethics Leader or use the Medium to High Risk Ethical Approval route.

76

Informed Consent of the Participant

Are any of the participants under the age of 18? Yes No

Are any of the participants unable mentally or physically to give consent? Yes No

Do you intend to observe the activities of individuals or groups without their knowledge and/or informed consent from each participant (or from his or her parent or guardian)?

Yes No

If you answered Yes to any of these questions, this may not be a low risk project. Please:


If you are a member of staff, discuss your project with your Faculty Research Ethics Leader or use the Medium to High Risk Ethical Approval route.

Participant Confidentiality and Data Protection

Will the project involve collecting data and information from human participants who will be identifiable in the final report?

Yes No

Will information not already in the public domain about specific individuals or institutions be identifiable through data published or otherwise made available?

Yes No

Do you intend to record, photograph or film individuals or groups without their knowledge or informed consent?

Yes No

Do you intend to use the confidential information, knowledge or trade secrets gathered for any purpose other than this research project?

Yes No

If you answered Yes to any of these questions, this may not be a low risk project:


If you are a member of staff, discuss your project with your Faculty Research Ethics Leader or use the Medium to High Risk Ethical Approval or NHS or Medical Approval routes.

Gatekeeper Risk

Will this project involve collecting data outside University buildings? Yes No

Do you intend to collect data in shopping centres or other public places? Yes No

Do you intend to gather data within nurseries, schools or colleges? Yes No

Do you intend to gather data within National Health Service premises? Yes No

If you answered Yes to any of these questions, this is not a low risk project. Please:


If you are a member of staff, discuss your project with your Faculty Research Ethics Leader or use the Medium to High Risk Ethical Approval or NHS or Medical Approval routes.

Other Ethical Issues

Is there any other risk or issue not covered above that may pose a risk to you or any of the participants?

Yes No

Will any activity associated with this project put you or the participants at an ethical, moral or legal risk?

Yes No

If you answered Yes to these questions, this may not be a low risk project. Please:


If you are a member of staff, discuss your project with your Faculty Research Ethics Leader.

77

Principal Investigator Certification If you answered No to all of the above questions, then you have described a low risk project. Please complete the following declaration to certify your project and keep a copy for your record as you may be asked for this at any time.

Agreed restrictions to project to allow Principal Investigator Certification

Please identify any restrictions to the project, agreed with your Supervisor or Faculty Research Ethics Leader to allow you to sign the Principal Investigator Certification declaration.

Participant Information Leaflet attached. N/A

Informed Consent Forms attached. N/A

Risk Assessment Form attached.

Principal Investigator’s Declaration

Please ensure that you:

Tick all the boxes below and sign this checklist.

Students must get their Supervisor to countersign this declaration.

I believe that this project does not require research ethics approval. I have completed the checklist and kept a copy for my own records. I realise I may be asked to provide a copy of this checklist at any time.

I confirm that I have answered all relevant questions in this checklist honestly.

I confirm that I will carry out the project in the ways described in this checklist. I will immediately suspend research and request a new ethical approval if the project subsequently changes the information I have given in this checklist.

Signatures

If you or your supervisor do not have electronic signatures, please type your name in the signature space. An email sent from the Supervisor’s University inbox will be accepted as having been signed electronically.

Principal Investigator

Signed Hyder Ali Khan (Electronically Signed) ........... (Principal Investigator or Student)

Date 19 December 2015 ..............................

Students storing this checklist electronically must append to it an email from your Supervisor confirming that they are prepared to make the declaration above and to countersign this checklist. This-email will be taken as an electronic countersignature.

Student’s Supervisor

Countersigned Dr Z. Ye (Peter) Email: [email protected] ....... (Supervisor)

Date 21 December 2015 ............................

I have read this checklist and confirm that it covers all the ethical issues raised by this project fully and frankly. I also confirm that these issues have been discussed with the student and will continue to be reviewed in the course of supervision.

Economy & Finance

Are Smart Beta Portfolios Smarter than Market Capitalization