Portfolio selection in vietnamese security market a multiple objective optimization approach

Proceedings the 10th students scientific research conference 28 https://www.semanticscholar.org/paper/A-structured-approach-to-the- evaluation-and-of-CASE-BlancKorn/43fc318bbfce41c287119fda8f23f88f9fb094d8 29 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4623860/ 30 https://en.wikipedia.org/wiki/ISO/IEC_9 Portfolio selection in Vietnamese security market:a multiple – objective optimization approach Group sciences: Trần Xuân Đức 202 Proceedings the 10th students scientific research conference Tô Thị Thuỳ Anh Phạm Anh Đức Class: AC2015D Science advisor: Asso Prof Nguyễn Hải Thanh THE NECESSITY OF TOPIC Developing an optimal portfolio, maximizing expected return under given risk, has long been an active research topic among academicians and business practitioners for decades, specifically since the pioneering work of Markowitz in 1956 There have been researchers continuously taking part in investigating this topic since then For very firstly, in Markowitz’s mean-variance model, estimating the pairwise correlations between all stocks are needed as input data to build up the optimal portfolio13 Therefore, between 1992 and 1993, a research on performance of portfolio selection in Hong Kong stock market was also done by some researchers: Kin Lam, Henry M.K Mok, Iris Cheung and H.C Yam14 Recently, in 2012, Colby Wright has 13 Nguyen Hai Thanh and Nguyen Van Dinh, Portfolio Optimization: Some Aspects of Modeling and Computing 14Kin Lam, Henry M.K Mok, Iris Cheung and H.C Yam, Family groupings on performance of portfolio selection in the Hong Kong stock market, published on September 1994 203 Proceedings the 10th students scientific research conference already researched how to perform optimal portfolio problems by using mathematical programming in Excel15 Hence, there are several reasons for researchers to keep going on digging deeply in this topic The most primary one is that this topic plays a very important part in helping practitioners run better in stock market Other motivation is that through investigating, researchers have a good chance to build up knowledge as well as the major in university Therefore, researchers are able to build up knowledge in relevant fields In short, with all the above reason, we have chosen the scientific researching topic, which is expected to contribute to other following researches in other markets: “Portfolio seclection in Vietnamese security market: A multiple-objetive optimization approach”.Method contributes to increase the accuracy in forecasting investment with the effort of saving unexpected costs 15 Colby Wright, Portfolio Optimization in Excel on Jan 11, 2012 on youtube.com 204 Proceedings the 10th students scientific research conference CHAPTER 1: GENERAL DOCUMENTATION AND THEORETICAL FUNDAMENTALS ABOUT EXISTING FORECASTING METHODS 1.1 THEORETICAL FUNDAMENTALS The combination of scientific methods and experience, the qualitative of investors, individuals involving in market in general and Vietnamese market in specific, is one of the important premises creating scientific forecasting or an artistic predict about the variability of investing objectives in future market Overall, all investors have not evaluated one-sided about the growth or reduction of investing objectives in market in a specific condition that every other existing factors are constant Actually, investors generally focus on predict the variability of investing objectives based on dependent variables, which are interdependent The most common interdependent variables are expected return and risk16 However, in practice, almost portfolio selection methods consider the comparison between two objectives Specifically, methods have to continuously repeat regarding to the random pairs of market’s objects Results of forecasting portfolio selection are not totally accurate Even though predicting methods are carefully processed, it still has errors, which is dependent on the actual variability in the future Hence, it is seldom to have an extremely accurate method, compared to actuality Nevertheless, portfolio selection optimization still is one of premising conditions, directly affecting the process and results of investment 16Lawrence J Gitman, Chad J Zutter, Principles of managerial finance 14th edition, published January 2014 205 Proceedings the 10th students scientific research conference 1.2 RESEARCH METHOD 1.2.1 Qualitative method Consult the majority advices of investors and potential investors Personal ideas of investors’ majority are consulted and gathered This represents the market’s investing demand The advantage of this method is that it can catch the majority of expectation of demanding source, then, stimulating investors attending the market The drawback of method, however, is that investors are directly affected by the market’s variability, so they usually underestimate to reduce risk This method is suitable for initial published Consult financial advisors Each advisor gives independent predict, then, the average predict is calculated; or a group of advisors discuss and lead to an advice This method quickly gives out the predict data, collect a number of objective idea, eliminate relatively subjective ideas The disadvantage is that the classification of objects is not really clear 1.2.2 Quantitative method “Modern Portfolio Selection” was primarily called Mean-Variance Analysis Expected return of portfolio is calculated along with risks represented by standard deviation and the interaction among securities (covariance and correlation) Then, “Portfolio Diversification” was formed Almost important premises of investors are based on two variables: Expected return and risk17 n E(r )=∑ P ᵢ R ᵢ i=1 n σ =∑ ( R ᵢ−E ( r ) ) Pᵢ i=1 17Lawrence J Gitman, Chad J Zutter, Principles of managerial finance 14th edition, published January 2014 206 Proceedings the 10th students scientific research conference 1.3 GENERAL DOCUMENTATION ABOUT OPTIMIZING MULTIPLE OBJECTIVES–PORTFOLIO SELECTION Optimizing multiple objectives-portfolio selection is affected by some factors Indeed, the interaction between Expected return and Risk is the most important Nowadays, in Vietnamese market, predicting optimal portfolio selection has not concentrated on cases of 3-4 objects When investors bring out plans, they often combine historical data and financial advisors’ experience for 1-2 objects Therefore, it is essential to build up an arithmetic and statistical model18 Optimizing multiple objectives-portfolio selection is not able to avoid general characteristics of forecast Although we not expect predict perfectly accurate, we desire errors among predict data not too large Model still gives investor confidence to make investing decisions 18Lawrence J Gitman, Chad J Zutter, Principles of managerial finance 14th edition, published January 2014 207 Proceedings the 10th students scientific research conference CHAPTER 2: THEORETICAL FRAME WORK AND APPLICATION PROCESS 2.1 INVESTIGATING CASE “Researching Strategy, including an actual experiment about a specific object at present context by using diversifying evidence sources, is called investigating case.”19 Investigating case is applied for predicting optimal portfolio selection for two objectives including the four target securities (as decision variables) ACB, FLC, FPT and AAA20 in Vietnamese economy 2017-2018 Forecast is conducted in Vietnamese security market generating premise to develop better model in other markets 2.2 DESIGN RESEARCH CASE The aim of research case Analyze predicting process and clarify characteristics of forecasting optimal multiple objectives portfolio selection along with affected factors Choosing research method Mixed strategy is chosen to be researching method because it has the combination between collecting data method, and quatitatively and qualitatively analyzing method Duration of research There is a limitation in researching duration However, there still exists the length of time factors Then, the development and the variability of objectives through periods are obvious 19Mark Saunders, Philip Lewis & Adrian Thornhill.2010, “Phuong phap nghien cuu kinh doanh”, Translator Nguyen Van Dung, Finance Publisher, page 156 20ACB: Asia Commercial Joint Stock Bank FLC: Finace, Land & Commerce Group FPT:The Corporation for Financing and Promoting Technology AAA: An Phat Green Plastic and Environment Joint Stock Company 208 Proceedings the 10th students scientific research conference 2.3 APPLICATION PROCESS Part 1: Choose samples for portfolio selection using CAN SLIM method to find out security desired to invest21 C-current quarterly earnings per share A-Annual earnings increases over the last five year N-new products, management, and other new event S-Small supply and large demand L- Choose leaders over laggard stocks within the same industry I-Pick stocks who have institutional sponsorship by a few institutions with recent above average performance M- Determining market direction by reviewing market averages daily Part 2: Develop optimal portfolio selection for two objectives Step 1: Using formula of Log return to compute continuously compounded interest over periods based on data collected (Return)22 r=ln ⁡( FV ) PV Step 2: Compute Excess return, Expected return and Standard Deviation Excess Return of Security of time t = Return of time t – Average Log Return of Security Return = Average Log Return of Security Standard Deviation (Using STDEV.P function in excel) Step 3: Covariance Matrix23 Σ=1/n(XTX) Nis the number of scores in each set of data X is excess return XTis the transpose of X Cov(X, Y) = Σ ( Xi - X ) ( Yi - Y ) / N = Σ xiyi / N 21A system for selecting stocks, created by Investor's Business Daily founder William J O'Neil Mehdi Majafi, Farshid Asgari,“Using CANSLIM Analysis for Evaluating Stocks of the Companies Admitted in Tehran Stock Exchange”, Journal of American Science 2013 22Wai-Sum Chan, Yiu – Kuen Tse (2013) Financial mathematics for actuaries, Mc Graw Hill Education (Asia) 23Wai-Sum Chan, Yiu – Kuen Tse (2013), Financial mathematics for actuaries, Mc Graw Hill Education (Asia) 209 Proceedings the 10th students scientific research conference N is the number of scores in each set of data X is the mean of the N scores in the first data set Xi is the ith raw score in the first set of scores xi is the ith deviation score in the first set of scores Y is the mean of the N scores in the second data set Yi is the ith raw score in the second set of scores yi is the ith deviation score in the second set of scores Cov(X, Y) is the covariance of corresponding scores in the two sets of data Step 4: Compare Expected Return and Standard Deviation based on Sharp rate Sharp Rate = Expected Return / Standard Deviation24 Step 5: Compute Expected return and Standard Deviation based on Weight of objectives, then, compute new Sharp rate25 Step 6: Using Excel Solver to compute weights in case Mean and Sharp rate are maximized and Standard Deviation is minimized26 The weights already had found out is an optimal portfolio CHAPTER 3: COLLECTING AND PROCESSING DATA 3.1 COLLECTING DATA 2015 2016 2017 3.1.1, C (Current quarterly earnings per share)27 ACB 1,145 1,483 2,148 FLC 1,952 1,795 599 We can see that almost stocks have EPS FPT 5,209 4,643 5,908 increase at least 29% from the same quarter’s AAA 962 earnings reported in the prior year 2,792 3,767 Current quarterly earnings per share table (unit: EPS) 3.1.2, A (Annual earning over last five years– from 1/1 to 31/12 each year )28 24, 18, 19 Portfolio Optimization in Excel - Colby Wright Published on Jan 11, 2012 on youtube.com 25 26 2720, 21 Information on finance.vietstock.vn 28 210 Proceedings the 10th students scientific research conference Object 74 Annual earning over last five years chard Year 2012 2013 2014 ACB: 2015 ACB 737534000 000 825596000 000 922249000 000 101208600 0000 FLC FPT AAA 48297751 171746851 43575265 85 7764 498 80225652 101624562 43810669 012 2188 080 26711820 178228095 36270141 6298 8870 145 93411415 177710688 54758545 3469 2168 610 In 2016 Income before tax were VND 1.667 billions, increased 27% from 2015 During 2016, net income raised 17% Service revenue also increased 27% almost focus on personal customer and financial service, while reduced risk in credit Almost expenses were used for improving information technology, maintaining branding system however always controlled closely In balance sheet, total asset, debit balance increased 16% and 21% FLC: In 2016, Sales return increased 0.78%, income before tax and profit after tax decreased 1.72% and 0.72% FPT: Profit after tax in Audit Separate Financial Statement of 2017 reached VND 2.987 billions, increase 161% compared with 2016 thanks to gain on the divestment of two companies: FPT Digital Retail Join Stock Company and FPT Trading company Limited 211 Proceedings the 10th students scientific research conference Profit after tax in Audited Consolidated Financial Statement of 2017 reached VND 3.528 billions, increased 37% compared with 2016 thanks to good performance of its business lines and gain divestment of two companies FPT Digital Retail Join Stock Company and FPT Trading Company Limited AAA: Profit after tax increased 92% compared with 2016 thanks to good performance of new factory, factory and and gain divestment of VBC and Yen Bai plastic and mineral corporation 3.1.3, N (New product, management and other new event) ACB: + Privilege Banking Project: Project is holding the lines of supplying products, services and/or activities suitable to privileged customer segment to keep contact with customer + Transaction Banking Project: To raise non-term deposits, increasing non-credit income collected from services fee, developing outstanding loan balance of individual customers and SMEs by sponsoring distributors + Developing business procedure ACMS We can see that ACB is focusing on improving banking services and customer services more than credit FLC: + Continuing raise resources so as to exploit some real estate project such as FLC Ha Long Bay Golf & Luxury Resort, Quang Binh Beach & Golf Resort + Carrying out some real estate home : FLC Twins Tower, Star Tower, Garden City, Eco-house Long Bien + Developing some potential field such as agriculture, industry +Participating in social activities in order to improve corporation’s fame FPT: + Promoting global development + Be a global partner of famous corporation in the world + Expanding customer list in Forbes 500 + Improving international competing capability AAA: 212 Proceedings the 10th students scientific research conference +Intensifying management, supervise manufacturing in order to raise worker’s responsibility in each position, purpose is ensure safety, quality and saving for company + Operating factories, Factory and +Building Factory + Raising market share in Japanese, Australia, America market 3.1.4, S (Supply and Demand)29 In our opinion, we choose stock in which Supply and Demand increase at least 50% because it shows that they have possibilityof increasing in price unit : stoc k 201 201 201 201 unit : stoc k 201 201 201 FPT Deman AAA supply Deman supply 74,064,560 71,017,490 90,861,800 87,413,200 128,999,220 121,215,670 133,860,700 139,396,700 427,971,320 454,171,030 187,557,300 194,206,500 ACB Deman FLC supply Deman supply 209,782,600 220,178,700 210,391,100 196,742,000 163,111,900 155,412,500 1,338,736,620 1,332,745,380 80,618,400 86,602,500 6,110,923,380 6,137,052,350 3.1.5, L (lead) 30 We choose stock which will take the lead in stock market in future 3.1.6, I (investor)31 ACB: 1, Standard Chartered APR Ltd 2, Connaught Investors Ltd 3, Dragon Financial Holdings Limited 29, 24 Information on finance.vietstock.vn 3023Mehdi Majafi, Farshid Asgari, “Using CANSLIM Analysis for Evaluating Stocks of the Companies Admitted in Tehran Stock Exchange” Journal of American Science 2013 31 213 Proceedings the 10th students scientific research conference 4, Standard Chartered Bank (Hong Kong) Ltd FPT: Although they don’t have big investor however they have many partner such as Microsoft, Amazon Web Service, AAA: Almost their partner is distribution channels both domestic and foreign 3.1.7, M (market orientation)32 At the end of 2016 many professor consider that VN-Index will fluctuate at 800 point, even the others see that it will reach 850 point According to Sate Security Commission of Vietnam’s data, in December 19,2017 ,VN-Index reached 951.42 point, increase 43% compared with 2016 HNX-Index reached 111,61point raised 41.5% compared with 2016 In 2017, many famous company and banking jointed in stock market and helped expand Vietnamese stock market in size In December 19, Capital stock reached 3.360 billion Vietnam dong, increased in 73% compared with 2016 equal to 74.6% GDP About transaction, the value of transaction was 13.870 billion Vietnam dong in which stock reached 4.981 billion Vietnam dong increase 63% Beside that in 2017, business result in many corporation have many positive signal especially banking, real estate, technologies and environmental protection product According to market movement, we believe that 2018 will be a successful year of Vietnamese stock market 3.2 Data processing and outcome analysis Mathematical model For the sake of simplify maths to jude and comments about portfolios selection actions It’s necessary to have some conventions about relative variables and to make clear for some mathematics symbols33: Ri = (R1, R2, R3, …, Rn) (matrix of n x return variables) W = (w1, w2, w3, …, wn) (matrix of n x weights of portfolios) μ = (μ1, μ2, μ3, …, μn) (matrix of n x portfolios’s expected return) 32 Information on finance.vietstock.vn 33 All formular and constrains base on Colby Wright method to optimize portfolios selection Portfolio Optimization in Excel - Published on Jan 11, 2012 on youtube.com 214 Proceedings the 10th students scientific research conference μp = W1μ1 + W2μ2 +W3μ3 + …+Wnμn = WTμ (expected return corresponding to n weights) σ = (σ1,σ2,σ3, …,σn) (matrix of n x variances) √ W T ΣW σp = (variance corresponding to n weights) (variance, co-variance matrix) D= σσT(the matrix of paired products of the standard deviations) D= [ σ 1σ σ 2σ ⋮ σn σ1 σ1 σ2 σ1 σn … σ2 σ2 σ2 σn ⋮⋱ ⋮ σn σ2 … σn σn ] C= Σ/D (correlation matrix) Objectives and their constrains: Minimize the risk (with μp max) W1+ W2+ W3+ …+Wn= σp ≤ β (β is σn min) W1, W2, W3, …, Wn ≥ Maximize rate of return (with σp min) W1+ W 2+ W3+ … +Wn= μp ≥ α (α is µn max) W1, W2, W3, …, Wn ≥0 Maximize sharp rate (µp/σp max) 215 Proceedings the 10th students scientific research conference W1+ W2+ W3+ …+Wn= W1, W2, W3, …, Wn ≥ RACB 2.4002 24 3.2659 551 Data processing RFLC 0.301 391 RFPT 0.019 496 0.006 23 RAAA 0.110 84 0.099 95 0.114 29 0.047 DateWithACB FLChad collected FPT AAA all the data above Q1/20 has table.Its 0.403 13been shown 430in quarterly 489 EPS 5,689 3,841 951 Q2/20 starts in the first quarter of 2013 and ends 13 39 661 5,801 3,438 0.1518 0.233 0.014 inQ3/20 the third quarter of 2017 Thus we have 194 466 13 1,022 990 5,765 3,111 19 numerical data for each portfolio(ACB, Q4/20 3435 13 FPT, AAA) 878 FLC, 1,250 5,849 2,775 0.0707 0.143 0.008 84 581 851 Q1/20 for the5,901 sake 2,646 of 14 Furthermore, 818 1,443 0.1193 0.260 0.071 0.040 Q2/20 compound interest rate, we can computing 12 817 28 368 14 726 1,873 5,495 2,755 not computing return in the first quarter in Q3/20 0.145 0.021 0.056 14 581 1,620 5,379 2,915 0.2227 2013 due to missing variable in formula of 99 12 34 452 Q4/20 0.5578 14 1,015 1,402 5,024 continuously compound interest Its can1,898 not 931 0.144 0.068 0.429 Q1/20 measures rate of return in the4,919 first quarter 53 28 07 15 1,048 1,310 1,581 ofQ2/20 2013 unless we have data of the fourth quarter of 2012 To sum up, we are going 15 1,012 1,380 4,816 1,095 have 18 returns from the second quarter of 2013 to third quarter of 2017 (n= 18) Q3/20 15 1,033 1,540 4,642 1,073 Quarterly rate of return table is the result of the first step of data processing, Q4/20 15 number 1,099 1,733 5,032 every represent return in each1,124 quarter They could compute through this Q1/20 formula: 16 1,130 1,649 4,663 1,684 Q2/20 16 1,197 2,032 3,660 2,164 FV (36) Q3/20 R= PV 16 1,250 1,659 3,711 2,404 Q4/20 Where: 16 1,483 1,527 3,459 2,736 Q1/20 R: rate of return 17 1,626 1,581 3,605 3,208 FV: future value Q2/20 17present value 1,747 885 4,449 3,265 PS: Q3/20 Rate in each quarter been3,379 distributed to expected return (μ) and variance 17 of return 1,927 885 have 4,093 EPS of portfolios table for eachQuarterly (σ) variable Result in table 3.2b Despite the expected return and variance, 34 ACB: Asia Commercial Joint Stock Bank FLC: Finace, Land & Commerce Group FPT: The Corporation for Financing and Promoting Technology AAA: An Phat Green Plastic and Environment Joint Stock Company 35 Information on Finance.vietstock.vn 36Financial mathematics for actuaries, Mc Graw Hill Education (Asia) Wai-Sum Chan, Yiu – Kuen Tse (2013) 216 Proceedings the 10th students scientific research conference there is a third measure to helps better judements toporfolios and better helps investing decisions, it’s the revert of coefficient of variance (sharp rate - μ/σ) Sharp rate measures return in Var-cov matrix FLC FPT ACB AC B 0.93170 0.02090 FL C 0.02090 0.04914 AAA 0.00443 0.00807 a unit of risk 0.01136 0.00258 The next step to compute variance matrix and Its covariance helpful in when its mathematics necessary to measure means and variances of weights On the other hands, the matrix itself can not be uses to make some judement activities Thus to make things Correlation matrix C=Σ /D ACB FLC clearly, correlation matrix is useful ACB FLC C= Σ/D = σxy/σxσy FPT Where: AAA 0.0976 85 0.0528 126 0.0607 137 C: FPT AAA 0.0528 0.0607 126 137 0.0602 T 0.4187 Σ=1/n(X X) 912 073 Where:0.4187 0.1856 Σ: variancecovariance 912 041 matrix.0.0602 0.1856 n: is073 number 041 of samples 0.0976 85 taken X: excess return XT: transpose of excess return Expected returns, Standard deviations and sharpe rate of portfolios correlation matrix Σ: variance- covariance matrix D: the matrix of paired products of the standard deviations Analysis outcome All the objectives above had been simplified by setting all weights to be positive This constrain had limited another application of this method to measure short sale Though this is not affected to the first purpose of the team (application of maxi-mizing 217 Proceedings the 10th students scientific research conference portfolios selections in Vietnam’s stock market), cause Vietnam government does not allowing to short sale at this time of stock market (4/2018) However, the application to other market that has alowed short sale is still possible Thus, short sale will not be mentioned in this report cause our team want to focus to Vietnam’s stock market C=Σ/D ACB FLC FPT AAA Unfortunately, the result in Correlation matrix ACB FLC FPT 1.00 0.10 -0.05 0.10 1.00 -0.42 -0.05 -0.42 1.00 -0.06 -0.06 -0.19 Correlation matrix table AAA -0.06 -0.06 -0.19 1.00 correlation matrix let us know there is no correlation between four of our portfolios Obviously when all of portfolios have corre-lation measures are negative or very close to zero, this means these four portforlios are independence from each other business Maybe, in the greater of scale, they have related or in another words this interpretation of nearest four years that they are independence As the result of method had mentioned above, optimization selections table gives us very specific number to jude or study.Base on the result, the first comment should be on the high level of risk that four fortfolios are giving us Even with very Equal weight constraining variable value of constraint ACB FLC FPT AAA Σwi μp σp None N/a 25.00% 25.00% 25.00% 25.00% 100.00 % 2.2719 % 25.1081 % Portfolios Max return ratio at σ