RISK AND INVESTMENT OPPORTUNITIES IN PORTFOLIO OPTIMIZATION. Ali Argun Karacabey. Ankara University, Turkey.

Similar documents
Heuristic algorithm for portfolio selection with minimum transaction lots

Barrier Options and a Reflection Principle of the Fractional Brownian Motion

Efficient Frontier - Comparing Different Volatility Estimators

The mean variance portfolio

THE BULLWHIP EFFECT IN SUPPLY CHAINS WITH STOCHASTIC LEAD TIMES. Zbigniew Michna, Izabela Ewa Nielsen, Peter Nielsen

Strategic Decision Making in Portfolio Management with Goal Programming Model

Decomposing the Money-Weighted Rate of Return an Update. Working Paper - Nummer: 21. by: Dr. Stefan J. Illmer. in Journal of Performance Measurement;

Value-Growth Investment Strategy: Evidence Based on the Residual Income Valuation Model

F P = A. PRESSURE In general terms, pressure conveys the idea of a force. Pressure, P, is the force, F that acts on a given area, A:

Capacity Utilization Metrics Revisited: Delay Weighting vs Demand Weighting. Mark Hansen Chieh-Yu Hsiao University of California, Berkeley 01/29/04

Intuitive Understanding of Throughput-Delay Tradeoff in Ad hoc Wireless Networks

Paul M. Sommers David U. Cha And Daniel P. Glatt. March 2010 MIDDLEBURY COLLEGE ECONOMICS DISCUSSION PAPER NO

Homework 2. is unbiased if. Y is consistent if. c. in real life you typically get to sample many times.

The t-test. What We Will Cover in This Section. A Research Situation

Morningstar Investor Return

Discovery of multi-spread portfolio strategies for weakly-cointegrated instruments using boosting-based optimization

Minnesota s Wild Turkey Harvest Fall 2016, Spring 2017

Market Timing with GEYR in Emerging Stock Market: The Evidence from Stock Exchange of Thailand

Math Practice Use Clear Definitions

Bayesian parameter estimation. Nuno Vasconcelos UCSD

Lifecycle Funds. T. Rowe Price Target Retirement Fund. Lifecycle Asset Allocation

Refining i\/lomentum Strategies by Conditioning on Prior Long-term Returns: Canadian Evidence

Making Sense of Genetics Problems

8.5. Solving Equations II. Goal Solve equations by balancing.

Footwork is the foundation for a skilled basketball player, involving moves

Research on Bus Priority Control on Non-coordinated Phase Based on Transmodeler A Case Study of Changjiang Road in Huangdao District of Qingdao

MATHEMATICAL ECONOMICS 10(17)

ICC WORLD TWENTY ( WORLD CUP-2014 )- A CASE STUDY

A SECOND SOLUTION FOR THE RHIND PAPYRUS UNIT FRACTION DECOMPOSITIONS

Representing polynominals with DFT (Discrete Fourier Transform) and FFT (Fast Fourier Transform) Arne Andersson

Available online at ScienceDirect. Procedia Engineering 113 (2015 )

3. The amount to which $1,000 will grow in 5 years at a 6 percent annual interest rate compounded annually is

The Power of 1. One Hope. One Step. One Direction. One Seed One Day. One Choice. One Habit. One voice One Moment. One Dream

CMA DiRECtions for ADMinistRAtion GRADE 6. California Modified Assessment. test Examiner and Proctor Responsibilities

A COMPARISON OF DEGRADATION AND FAILURE-TIME ANALYSIS METHODS FOR ESTIMATING A TIME-TO-FAILURE DISTRIBUTION

Evaluating the Performance of Forecasting Models for Portfolio Allocation Purposes with Generalized GRACH Method

3.10 Convected Coordinates

AEROBIC SYSTEM (long moderate work)

TRACK PROCEDURES 2016 RACE DAY

Contents TRIGONOMETRIC METHODS PROBABILITY DISTRIBUTIONS

Chapter 5. Triangles and Vectors

What the Puck? an exploration of Two-Dimensional collisions

INVESTIGATION 2. What s the Angle?

Modeling Speed Disturbance Absorption Following State-Control Action- Expected Chains: Integrated Car-Following and Lane-Changing Scenarios

AP Physics 1 Per. Unit 2 Homework. s av

Constructing Absolute Return Funds with ETFs: A Dynamic Risk-Budgeting Approach. July 2008

Stock Return Expectations in the Credit Market

Brand Selection and its Matrix Structure -Expansion to the Second Order Lag-

Why? DF = 1_ EF = _ AC

Time & Distance SAKSHI If an object travels the same distance (D) with two different speeds S 1 taking different times t 1

KEY CONCEPTS AND PROCESS SKILLS. 1. An allele is one of the two or more forms of a gene present in a population. MATERIALS AND ADVANCE PREPARATION

Armstrong PT-516 High Capacity Pump Trap

GENETICS 101 GLOSSARY

A Liability Tracking Portfolio for Pension Fund Management

Spam Message Classification Based on the Naïve Bayes Classification Algorithm

Call To Action. & bb b b b. w w. œ œ œ J. &b b b b b. œ œ œ. œ œ. œ œ œ. ? b b b b b 4. j œ. j œ Ó n. œ j. œ j œ œ Ó Œ. œ j Ó Œ j œœ œ Ó.

Predicting Genotypes

Proportional Reasoning

HERKIMER CENTRAL SCHOOL DISTRICT Herkimer Elementary School 255 Gros Boulevard Herkimer, New York 13350

ANALYSIS OF RELIABILITY, MAINTENANCE AND RISK BASED INSPECTION OF PRESSURE SAFETY VALVES

Considering clustering measures: third ties, means, and triplets. Binh Phan BI Norwegian Business School. Kenth Engø-Monsen Telenor Group

Idiosyncratic Volatility, Stock Returns and Economy Conditions: The Role of Idiosyncratic Volatility in the Australian Stock Market

Lowland Aspen, Maple, Beech, Aspen, Mixed Other Mixed Mixed N Other Mixed

(I Got Spurs That) Jingle Jangle Jingle lyrics Tex Ritter

Oath. The. Life-changing Impact TEACH HEAL DISCOVER. Going Into the Wild to Save Rhinos. Tracking Down Outbreaks page 2. Teaming Up for Nekot page 7

2014 WHEAT PROTEIN RESPONSE TO NITROGEN

Compartment Review Presentation

2. JOMON WARE ROPE STYLES

Lowland Cedar High Density Aspen High Density Aspen, Spruce/Fir High Density Aspen High Density

The Pythagorean Theorem and Its Converse Is That Right?

The Study of Indoor Air Thermal Environment and Energy in Winter

Overview. Do white-tailed tailed and mule deer compete? Ecological Definitions (Birch 1957): Mule and white-tailed tailed deer potentially compete.

A Probabilistic Approach to Worst Case Scenarios

Interpreting Sinusoidal Functions

8/31/11. the distance it travelled. The slope of the tangent to a curve in the position vs time graph for a particles motion gives:

ELIGIBILITY / LEVELS / VENUES

ELIGIBILITY / LEVELS / VENUES

Evaluating Portfolio Policies: A Duality Approach

Bootstrapping Multilayer Neural Networks for Portfolio Construction

Welcome to the world of the Rube Goldberg!

SPH4U Transmission of Waves in One and Two Dimensions LoRusso

Back In The Saddle Again -- by Ray Whitley and Gene Autry --

ELIGIBILITY / LEVELS / VENUES

Headfirst Entry - Diving and Sliding

2017 / 2018 SCORPIONS SOCCER STYLE OF PLAY & TECHNICAL DEVELOPMENT

Grade 6. Mathematics. Student Booklet SPRING 2011 RELEASED ASSESSMENT QUESTIONS. Record your answers on the Multiple-Choice Answer Sheet.

Aspen, Spruce/Fir High Density Lowland Cedar High Density Aspen High Density Lowland Cedar High Density

Dynamic condensation and selective mass scaling in RADIOSS Explicit

Economics 487. Homework #4 Solution Key Portfolio Calculations and the Markowitz Algorithm

1 What is Game Theory? Game Theory 1. Introduction. Rational Agents. Rational Agents in Game Theory

The structure of the Fibonacci numbers in the modular ring Z 5

Corresponding Author

Stages Written by: Striker. Old West Sayings. Always drink upstream from the herd.-- Will Rogers

A Data Envelopment Analysis Evaluation and Financial Resources Reallocation for Brazilian Olympic Sports

"Pecos Bill Rides a Tornado" Stages Written by: Striker

» WYOMING s RIDE 2013

ELIGIBILITY / LEVELS / VENUES

Introduction to Algorithms 6.046J/18.401J/SMA5503

Research Article Spatial Approach of Artificial Neural Network for Solar Radiation Forecasting: Modeling Issues

CALCULATORS: Casio: ClassPad 300 ClassPad 300 Plus ClassPad Manager TI: TI-89, TI-89 Titanium Voyage 200. The Casio ClassPad 300

Transcription:

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey RISK AND INVESMEN OPPORUNIIES IN PORFOLIO OPIMIZAION Ali Argu Krcbey Akr Uiversiy, urkey. ABSRAC Mrkowiz legedry work bou porfolio opimizio is cceped o be he pioeer of he moder porfolio heory. Corry o is heoreicl repuio, i hs o bee used eesively. Koo d Ymzki (99) proposed ew porfolio opimizio model s lerive o Mrkowiz s me-vrice model. Mrkowiz s me vrice model d he me bsolue deviio models regrd risk i erms of deviios h my be eiher posiive (upwrd) or egive (dowwrd) i relio o he epeced reur. I oher words boh of he models pelize o oly he egive (dowwrd) deviios bu lso he posiive (upwrd) deviios. I his pper, ew model h kes io cosiderio boh risk d beer ivesme opporuiy is proposed. he differece bewee he proposed model d he oher porfolio opimizio models is heir obecives. he proposed model ssumes h ivesor ws o choose porfolio wih higher upside deviios d lower dowside deviios. Key word: porfolio opimizio, MAD, dowside risk JEL Codes: G, G2, G32 Fculy of Poliicl Scieces, 659, Akr, urkey.

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey I. INRODUCION Mrkowiz s legedry sudy of porfolio opimizio is regrded s he pioeerig work of moder porfolio heory. I he Mrkowiz model, risk is sed i erms of he prediced vrice of porfolio reur, fucio h is qudric i he decisio vribles. All oher fucios d cosris re ssumed o be lier (Shrpe, 97). he obecive of he model is o form he efficie porfolios. Corry o is heoreicl repuio, i hs o bee used eesively. he wo impor resos why Mrkowiz s model hs o bee implemeed c be summrized s follows (Elo, Gruber, Pdberg, 976; Koo d Ymzki, 99): (i) he difficuly i esimig he correlio mrices, (ii) he compuiol difficuly of he qudric progrmmig model. Shrpe (97) climed h if he essece of porfolio lysis problem could be dequely cpured i form suible for lier progrmmig mehods, he prospec for prcicl pplicio would be grely ehced. Shrpe (97) d Soe (973) ried o cover he porfolio problem io lier progrmmig model. Koo d Ymzki (99) proposed ew porfolio opimizio model s lerive o Mrkowiz s me-vrice model. hey employed L me bsolue deviio s risk mesure ised of vrice, i order o overcome he problem of compuiol difficuly. he MAD model is sid o be vible lerive becuse (i) i does o require he covrice mri of he reurs, d (ii) MAD porfolios hve fewer sses (Sim, 997). I is lso rgued h s he umber of sses decreses, he rscio coss of he porfolio will lso decrese. he MAD porfolio opimizio model hs 2+2 rows where is he umber of periods. Feisei d hp (993) reformuled he MAD porfolio opimizio model so h he umber of rows decresed o +2, which implies h he mimum umber of socks ivesed i decreses from 2+2 o +2. Chg (25) modified Feisei d hp s model so h his model hs fewer vribles d he sme umber of cosris. 2

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey All of hese opimizio models cosider risk s he deviio of reurs from he epeced or me reur. However, hese models ssume h here is o differece bewee posiive d egive deviio. O he oher hd, for ivesor posiive deviio is desirble, while egive deviio is o. So, i his pper ew model h differeies posiive d egive deviios is proposed. Accordig o his proposed model ivesor simuleously wishes o mimize posiive deviios d miimize egive deviios. II. REVIEW OF HE MEAN VARIANCE AND MAD PORFOLIO OPIMIZAION MODELS Mrkowiz (952) cosiders wo rules while formulig he porfolio opimizio model. Firs, he ivesor does (or should) mimize epeced reurs, d secodly, he ivesor does (or should) cosider epeced reur desirble hig d vrice of reur udesirble hig. he cocep of efficie porfolio hs emerged i ccordce wih hese wo rules. he Mrkowiz porfolio opimizio model employs vrice s he mesure of risk, d he obecive of he model is o fid ou he weighigs of sses h miimizes he vrice of porfolio d esure reur equl o or bigger h he epeced reur. Accordigly he mhemicl model for sses is s follows: Miimize subec o i= = = = r ρm =... Where: σ i = covrice bewee sses i d, = he mou ivesed i sse, r = he epeced reur of sse per period, σ i i () 3

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey ρ = prmeer represeig he miiml re of reur required by ivesor, M = ol mou of he fud, d u = mimum mou of moey which c be ivesed i sse. Koo d Ymzki (99) iroduced he L risk fucio (me bsolue deviio-mad) w () = ER E R ised of = he L 2 risk (vrice) fucio where R is rdom vrible represeig he re of reur per period of he sse. hey proved h hese wo mesures re he sme if (R R ) re mulivrie ormlly disribued. So he Koo-Ymzki MAD porfolio opimizio model becomes s follows: Miimize subec o w() = E = = E[R ] R ρm E = R =... Koo d Ymzki ssumed h he epeced vlue of he rdom vrible c be pproimed by he verge from he d. So: r = E[R ] = r = Where r is he relizio of rdom vrible R durig period (where = ). hus, w() is pproimed by = = (r r ). Deoig =r -r (= d = ), model (2) c be epressed s follows. (2) 4

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Miimize subec o = = = r = ρm =... Ali Argu Krcbey Koo d Ymzki replced he model wih model (4) which is equivle o model (3). Miimize subec o = y y y = - = + r = = ρm =... =... =... Accordig o Koo d Ymzki he MAD porfolio opimizio model s dvges over he Mrkowiz s model re (i) his model does o use he covrice mri which herefore does o eed o be clculed, (ii) solvig his lier model is much esier h solvig qudric model, (iii) he mimum umber of sses h re ivesed i is 2+2 (if u = ) while Mrkowiz s model my coi s my s sses, d (iv) c be used s corol vrible o resric he umber of sses. Feisei d hp (993) modified he MAD porfolio opimizio model d proposed ew model h is equivle o Koo d Ymzki s bu hs limi of +2 o he umber of ozero sses i he opiml porfolio o he ssumpio h here is o upper limi o he ivesme u = - i sse. (3) (4) 5

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey hey subrc o-egive surplus vribles 2v d 2w from ech of he cosris i problem 4 i order o replce he iequliies wih equliies: y y + = = 2v - 2w = = I order o elimie y, hey subrc (6) from (5) d divide by 2. hus he porfolio opimizio model becomes: (5) (6) Miimize subec o = v (v = = r + w ) - w ρ M (7) = = =... =... v, w, =... Compred wih Koo d Ymzki s opimizio model uder he u = ssumpio, model (7) llows ivesme i mos +2 sses i he opiml porfolio. Accordig o Koo d Ymzki his mes h he opiml porfolio produced by (7) should hve lower rscio coss compred wih he opiml porfolio obied by model (4). Chg (25) reformule Feisei d hp s model by iroducig coiuous vrible d ; 6

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Miimize subec o d d (2d = = - = = r ρm ) =... =... Ali Argu Krcbey (8) = =... Chg (25) proved h model (8) is equivle o model (7) while he umbers of ddiiol coiuous vribles d uiliry sig cosris re hlf of he Feisei d hp s model. So he CPU ime d he umber of ierios eeded o fid he opiml soluio re decresed. III. PROPOSED OPIMIZAION MODEL Mrkowiz s me vrice model d he me bsolue deviio models regrd risk i erms of deviios h my be eiher posiive (upwrd) or egive (dowwrd) i relio o he epeced reur. I oher words boh of he models pelize o oly he egive (dowwrd) deviios bu lso he posiive (upwrd) deviios. However, he differece bewee egive d posiive deviios is crucil. Negive deviio is regrded s udesirble for mos ivesors while, posiive deviio is desirble. Followig he work of Mrkowiz (959), i which he proposed h semivrice replce vrice s he mesure of risk, dowside risk hs bee he subec of umerous sudies [Grooveld d Hllebch (999), Michlowski d Ogryczk (2)]. he geerl coclusio cocerig dowside risk is h he lef-hd side of reur disribuio ivolves risk while he righ-hd side cois he beer ivesme opporuiies Grooveld d Hllebch (999). 7

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey Some of hese sudies proposed porfolio opimizio models which employ dowside risk s he mesure of risk. A porfolio opimizio model which icorpores dowside risk s he mesure of risk oly pelizes dowside deviios bu does o ke upside deviios io cosiderio. hese models re similr o he me vrice or MAD models s ll of hem shre he obecive of miimizig risk, he lef-hd side of reur disribuio. Perhps he mos impor deficiecy of hese models is h hey do o ke io cosiderio he beer ivesme opporuiies, he righ hd side of reur disribuio. I his pper ew model h kes io cosiderio boh risk d beer ivesme opporuiies re proposed. he differece bewee he proposed model d he oher porfolio opimizio models is heir obecives. he proposed model ssumes h ivesor ws o choose porfolio wih higher upside deviios d lower dowside deviios. I oher words he/she hs wo simuleous obecives. Firs he/she ws o mimize upside deviios d secod ws o miimize dowside deviios. hese wo obecives c be merged d resed s he sigle obecive of mimizig he differece bewee he upside d dowside deviios. For securiies (= ) durig periods (= ) he dowside risk (NMAD) d beer ivesme opporuiies (PMAD) c be show i his wy: NMAD = N = mi [, r -r ] (9) PMAD = P = m [, r -r ] () Accordig o hese defiiios, he obecive of he proposed model c be sed s follows: Mimize subec o = = = r = P. ρm - = = N. =... () 8

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey his model is similr o model (2) if we ssume h = P. / = Y d = Mimize = subec o Y = = = P = _ N r N. / = Z. = Z _ Y _ Z ρm = = =... =... =... (2) his ew model hs 2+ vribles d 2+2 cosris. As cosequece of developmes i compuer echology, he umber of vribles d cosris re o loger crucil problem. Bu his ew model differs from oher porfolio opimizio models sice i cosiders risk i erms of egive deviios from he me d lso kes he beer ivesme opporuiies io cosiderio. Accordig o oher opimizio models, he securiy wih he lowes deviio mog securiies wih he sme reur is regrded s he mos desirble wheres his model regrds securiy which hs higher posiive d lower egive deviios s more desirble. I oher words, his ew model ries o fid equilibrium poi bewee risk d beer ivesme opporuiies. III. COMPARISON OF HE OPIMIZAION MODELS I his secio, i order o evlue he performce of he porfolio opimizio models, porfolios developed ccordig o he proposed model, me vrice d me bsolue deviio models re compred for differe ivesme horizos. he dbse cosiss of socks icluded i he ISE-, he well-kow ide of he Isbul Sock Echge o he srig de of he lysis, Jury 2. 9 of he socks re ecluded becuse of missig d. 9

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey I he firs sge of he lysis, 48 differe d ses A,A2 A48- which iclude d for 9 socks over 2 mohs re prepred. A icludes d from Jury 2 o December 2 (- 2 mohs); A2 icludes d from Februry 2 o Jury 2 (2-3 mohs) d so o. For every d se, he me-vrice, MAD d proposed opiml porfolios re pu ogeher usig models, 7 d 2. I he secod sge, i order o ke io cosiderio he effec of he porfolio horizo, he ssumpio ivolvig modificio of he porfolios every moh is replced wih modificio of he porfolios i every 3 mohs, 6 mohs, yer d 2 yers. For ech ssumpio, he reurs of he 3 porfolios developed ccordig o he models, 7 d 2, re clculed, d he hese reurs re compred. Porfolios modified more h oce yer show h he proposed model s reurs defiiely eceed he reurs of he oher wo models (Figure ). Whe he ivesme horizo eceeds yer, he me bsolue deviio models produce higher reurs. Figure : Reurs of he models for differe ivesme horizos,35,3,25,2,5,,5 moh 3 mohs 6 mohs yer 2 yer MV MAD PRO Eve hough ordiry ivesor geerlly evlues his/her ivesme performce oly ccordig o he reurs of he porfolio, for vlid evluio, clculig he mou of he reur

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey for every ui of risk bore would be more pproprie. o clcule he rewrd per risk equio 3 is employed: Porfolio Performce = me reur/risk (3) I is obvious h porfolio performce clculed usig equio (3) will differ ccordig o he risk mesure chose. I his pper 3 differe risk mesures re used. hese re sdrd deviio, dowside risk d me bsolue deviio. For every porfolio d for every ivesme horizo, 3 performce mesures re clculed usig hese 3 differe risk mesures. Besides, well kow porfolio performce mesure, Shrpe rio is clculed. he performce of he porfolios which were modified every moh is show i Figure 2. Compred wih he me reurs i Figure, he differece bewee he performces of differe porfolios is smller. his idices h, idepede of he risk mesure seleced, porfolios composed ccordig o he proposed model re riskier h he oher porfolios. Figure 2: Differe Risk Mesures of he Porfolios MV MAD PRO,45,4,35,3,25,2,5,,5 Sdrd Deviio Dowside Risk MAD Usig he proposed model, he icresed risk for porfolios is see o oly i hose porfolios revised mohly bu lso i ll ivesme horizos. ble summrizes some sisics of he porfolios. Aoher coclusio of he ble is h porfolios usig he proposed model ouperform he oher porfolios for ivesme horizos shorer h yer.

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. ble : Summry Sisics of he Porfolios Ali Argu Krcbey Moh 3 Mohs 6 Mohs Opimizio Model Me Reur Sdrd Deviio Dowside Risk MAD MV,2,26,64,93 MAD,5,9,52,84 PRO,33,82,82,4 MV,6,32,67,99 MAD,2,32,73,92 PRO,8,78,82,32 MV,3,27,67,96 MAD,6,4,75,3 PRO,27,73,79,33 Whe compred o me vrice d me bsolue deviio models, he ew proposed model geeres more risky porfolios. Bu for idividul ivesors who geerlly mke ivesme decisios bsed o reur proposed model cosiues more rcive porfolios. ble 2 summrizes he performce mesures of he porfolios. For d 6 moh periods, proposed model ouperforms he oher porfolios. For 3 moh period, me vrice porfolio shows beer performce lhough proposed model s me reur is he highes. 2

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey ble 2: Performce Mesures of he Porfolios Me Reur / MAD Me Reur / Dowside Risk Shrpe Rio Opimizio Model Moh 3 Mohs 6 Mohs MV,96,88,29 MAD,26,288,79 PRO,82,42,234 MV,8,239,62 MAD,9,64,3 PRO,4,22,36 MV,,94,35 MAD,6,23,55 PRO,59,342,23 Whe compred o me vrice d me bsolue deviio models, he ew proposed model geeres more risky porfolios. Bu for idividul ivesors who geerlly mke ivesme decisios bsed o reur proposed model cosiues more rcive porfolios. IV. CONCLUDING REMARKS Mos of he reserch bou porfolio opimizio fer Mrkowiz focused o overcomig he compuiol burde of he me vrice model. Koo d Ymzki sugges ew model which cceps MAD s risk mesure ised of vrice. hey lso proved h miimizig MAD is similr o miimizig vrice if he reurs of he socks re mulivrie ormlly disribued. Mrkowiz s me vrice model d he me bsolue deviio models ccep risk s he deviio h c be posiive (upside) or egive (dowside) from he epeced reur. I oher words boh of he models puish o oly he egive (dowside) deviios bu lso he posiive (upside) deviios. However, he differece bewee egive d posiive deviios is crucil. 3

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey Negive deviio is cceped o be udesirble for mos of he ivesor while posiive deviio is desirble. I his pper ew model h kes io cosiderio boh risk dowside deviios d beer ivesme opporuiies upside deviios-, is proposed. he differece bewee he proposed model d he oher porfolio opimizio models is heir obecives. he proposed model ssumes h ivesor ws o choose porfolio wih higher upside deviios d lower dowside deviios. I his pper his heoreiclly righ issue is esed wih rel d of emergig mrke. he resuls of he lysis showed h he porfolios of he proposed model re riskier bu geere higher reurs. So his model c be useful for ivesors whose ivesme evluio mosly depeds o reur. REFERENCES Chg, C. (25). A Modified Gol Progrmmig Approch for he Me Absolue Deviio Porfolio Opimizio Model. Applied Mhemics d Compuio, (Aricle I Press). Elo, E. J., Gruber, M.J. d Pdberg, W. (976). Simple Crieri for Opiml Porfolio Selecio. he Jourl of Fice, Vol. 3, No. 5, pp. 34-357. Feisei, C.D. d hp, M.N. (993). Noes: A Reformulio of Me- Absolue Deviio Porfolio Opimizio Model. Mgeme Sciece, Vol. 39, No. 2, pp. 552-553. Grooveld, H. d Hllerbch, W. (999). Vrice vs Dowside Risk: Is here Relly h Much Differece? Europe Jourl of Operiol Reserch, Vol. 4, Issue 2, pp. 34-39. Koo, H. d Ymzki, H. (99). Me-Absolue Deviio Porfolio Opimizio Model d IS Applicios o okyo Sock Mrke. Mgeme Sciece, Vol. 37, No. 5, pp. 59-53. Mrkowiz, H. (952). Porfolio Selecio. he Jourl of Fice, Vol. 7, No., pp. 77-9. Michlowski, W. d Ogryczk, W. (2). Eedig he MAD Porfolio Opimizio Model o Icorpore Dowside Risk Aversio. Nvl Reserch Logisics, Vol. 48, İssue 3, pp.85-2. Shrpe, W. (97). A Lier Progrmmig Approimio for he Geerl Porfolio Alysis Problem. Jourl of Ficil d Quiive Alysis, December, pp. 263-275. Sim, Y. (997). Esimio Risk i Porfolio Selecio: he Me Vrice Model Versus he Me Absolue Deviio Model. Mgeme Sciece, 4

Europe Jourl of Fice d Bkig Reserch Vol.. No.. 27. Ali Argu Krcbey Vol. 43, No., pp. 437-446. Soe, B. (973). A lier Progrmmig Formulio of he Geerl Porfolio Selecio Problem. Jourl of Ficil d Quiive Alysis, Vol. 8, pp. 62-636. 5