Forecasting Financial Markets Using Neural Networks: An Analysis of Methods and Accuracy

82 Pages·1998·0.41 MB·English

by  Jason E. Kutsurelis

NAVAL POSTGRADUATE SCHOOL
Monterey, California
THESIS
FORECASTING FINANCIAL MARKETS
USING NEURAL NETWORKS: AN ANALYSIS
OF METHODS AND ACCURACY
by
Jason E. Kutsurelis
September 1998
Principal Advisor: Katsuaki Terasawa
Approved for public release; distribution is unlimited.
FORECASTING FINANCIAL MARKETS USING NEURAL
NETWORKS: AN ANALYSIS OF METHODS AND ACCURACY
Jason E. Kutsurelis-Lieutenant, United States Navy
B.S., United States Naval Academy, 1991
Master of Science in Management-September 1998
Principal Advisor: Katsuaki Terasawa, Department of Systems
Management
Associate Advisor: William R. Gates, Department of Systems
Management
sihT hcraeser senimaxe dna sezylana eht esu fo laruen skrowten sa a gnitsacerof .loot
yllacificepS a laru esn'krow tyetniliba ot tciderp erutuf sdnert fo kcotS tekraM secidnIsi
.detset ycaruccA si derapmoc tsniaga a lanoitidart gnitsacerof ,dohtem elpitlumraenil
detalucla cs itcerro cgnie btsacero fs'ledo meh tf oytilibabor peh t,yllani F.sisylan anoisserger
.sei tli algninob iiastbuiodr npeolcihW y l gnynoli fslesa iurkrcurb seo,inwydtreonehstiht
gnitsacero fas askrowte nlarue ngnis uf oytilacitcar pdn aytilibisae feh tsenimrete dhcraeser
draw dyeEbn okdr oe whnto psudli uybdu tssi h.Trotsev nliaudivid ne ihr tolfoot nyiletaG
sih koob larueN skrowteN rof lai c.nganniitFsacero FsihT hcraes esretadilav eht krowfo
yletaG dna sebircsed eht tnempoleved fo a laruen krowten taht deveihca a 3.39tnecrep
tekram a gnitciderp fo ytilibaborp tnecrep 70.88 na dna ,esir tekram a gnitciderp fo ytilibaborp
tsacerof ot ytilibapac eht evah od skrowten laruen taht dedulcnoc saw tI .005P&S eht ni pord
e seu hmto rtfifen edblu orcotsev nliaudivid nei h,tdenia rytlrepo r fp,id nsatekr almaicnanif
.l ogontitsace rs oiffhot
KEYWORDS: Neural Networks, Finance, Time Series Analysis, Forecasting,
Artificial Intelligence
DOD TECHNOLOGY AREA: Modeling and simulation, Artificial
Intelligence, Neural Networks
REPORT DOCUMENTATION PAGE
Form Approved OMB No. 0704-0188
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instruction, searching
existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this
burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services,
Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management
and Budget, Paperwork Reduction Project (0704-0188) Washington DC 20503.
1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED
September 1998 Master’s Thesis
4. TITLE AND SUBTITLE FORECASTING FINANCIAL 5. FUNDING NUMBERS
MARKETS USING NEURAL NETWORKS: AN
ANALYSIS OF METHODS AND ACCURACY
6. AUTHOR(S) Kutsurelis, Jason E.
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING
Naval Postgraduate School ORGANIZATION
Monterey CA 93943-5000 REPORT NUMBER
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10.SPONSORING/MONITORING
AGENCY REPORT NUMBER
11. SUPPLEMENTARY NOTES The views expressed in this thesis are those of the author and do not reflect
the official policy or position of the Department of Defense or the U.S. Government.
12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE
Approved for public release; distribution is unlimited.
13. ABSTRACT (maximum 200 words)
yallacifi c.e lpo Sogtnitsace r ssoaakfrowt elnaru efe nose uhstezyla nd ansaenima xhecraes esrihT
derapmo cs iycaruc c .Adetse ts isecidn Itekra Mkcot Sf osdner terutu ftcider po tytilib as'krowte nlaruen
eh tf oytilibabor peh t,yllan i .Fsisylan anoisserge rraeni lelpitlu m,dohte mgnitsacero flanoitidar tatsniaga
gnissucs iydlfei rybl neol i. hsWeitilibabo rlpanoitidn ogcni sduetalucl a scticerr ogcni etbsacer osf'ledom
s askrowte nlarue ngnis uf oytilacitcar pdn aytilibisae feh tsenimrete dhcraese rsih t,yroeh tkrowte nlaruen
draw d yEebn okdr oew hnto psudli uybdu tssi h.Trotsev nliaudivid nei hrt olfo ogtnitsacer oaf n iyletaG
sih koob larueN skrowteN rof laicnaniF .gnitsaceroF sihT hcraeser setadilav eht krow fo yletaGdna
sebircs etednhetmpole vfeod a lar ukernowte nta hdteveihca a 3.3 9yttnielcirbeapb o frgopnitciderpa
dedulcnoc saw tI .005P&S eht ni pord tekram a gnitciderp fo ytilibaborp tnecrep 70.88 na dna ,esir tekram
taht laruen skrowten od evah eht ytilibapac ot tsacerof laicnanif stekram ,dna fi ylreporp ,deniarteht
.lo ogtnitsacer osfi h fteo seu hmto rtfifen edblu orcotsev nliaudividni
14. SUBJECT TERMS Neural Networks, Finance, Time Series Analysis, 15. NUMBER OF
PAGES *123
Forecasting, Artificial Intelligence.
16. PRICE CODE
17. SECURITY CLASSIFI- 18. SECURITY CLASSIFI- 19. SECURITY CLASSIFI- 20. LIMITATION OF
CATION OF REPORT CATION OF THIS PAGE CATION OF ABSTRACT ABSTRACT
Unclassified Unclassified Unclassified UL
NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89)
Prescribed by ANSI Std. 239-18 298-102
i
Approved for public release; distribution is unlimited.
FORECASTING FINANCIAL MARKETS USING NEURAL
NETWORKS: AN ANALYSIS OF METHODS AND ACCURACY
Jason E. Kutsurelis
Lieutenant, United States Navy
B.S., United States Naval Academy, 1991
Submitted in partial fulfillment
of the requirements for the degree of
MASTER OF SCIENCE IN MANAGEMENT
from the
NAVAL POSTGRADUATE SCHOOL
September 1998
Author:
Jason E. Kutsurelis
Approved by:
Katsuaki Terasawa, Principal Advisor

William R. Gates, Associate Advisor

Reuben T. Harris, Chairman
Department of Systems Management
ii
iii
ABSTRACT
.loo tgnitsacero f as askrowte nlarue nf oes ueh tsezylan adn asenimax ehcraese rsihT
y llacific eapS l asr'uk erynotwitleinba ot tcid eerrput usfdnert fo kcot Stek rsaeMcidnIsi
.detset ycaruccA si derapmoc tsniaga a lanoitidart gnitsacerof ,dohtem elpitlumraenil
detalucla cs itcerro cgnie btsacero fs'ledo meh tf oytilibabor peh t,yllani F.sisylan anoisserger
.sei tli algninob iiastbuiodr npeolcihW y l gnynoli fslesa iurkrcurb seo,inwydtreonehstiht
gnitsacero fas askrowte nlarue ngnis uf oytilacitcar pdn aytilibisae feh tsenimrete dhcraeser
draw dyeEbn okdr oe whnto psudli uybdu tssi h.Trotsev nliaudivid ne ihr tolfoot nyiletaG
sih koob larueN skrowteN rof lai c.nganniitFsacero FsihT hcraes esretadilav eht krowfo
yletaG dna sebircsed eht tnempoleved fo a laruen krowten taht deveihca a 3.39tnecrep
tekram a gnitciderp fo ytilibaborp tnecrep 70.88 na dna ,esir tekram a gnitciderp fo ytilibaborp
tsacerof ot ytilibapac eht evah od skrowten laruen taht dedulcnoc saw tI .005P&S eht ni pord
e seu hmto rtfifen edblu orcotsev nliaudivid nei h,tdenia rytlrepo r fp,id nsatekr almaicnanif
.l ogontitsace rs oiffhot
iv
v
ST NEE LTFBNOAOTC
I. INTRODUCTION ................................ ................................ ............................... 1
.A SLAOG ................................ ................................ ................................ .... 1
1. dnuorgkcaB ................................ ................................ .................. 2
2. sevitcejbO ................................ ................................ ..................... 6
3. noits ehucQrae seehRT ................................ ................................ . 6
4. snoitpmus sdA nsanoitatim i,LepocS ................................ ............. 7
5. ygolodoh t dewnMeai veerRutaretiL ................................ ............. 7
6. ydu tfnSooitazinagrO ................................ ................................ ... 8
II. KROWEM ALRAFCITERO EDWHNETAI VEERRUTARETIL .................... 11
.A ERUTARWEETIIVLER ................................ ................................ ....... 11
B. KROWEMA RLFACITEROEHT ................................ .......................... 13
III. YGOLODOHTEM ................................ ................................ ............................ 18
IV. NDOEI TTF CAOAETTLNALEDOSCERP ................................ .................... 33
.A KROW TEESNOLC ................................ ................................ .............. 33
B. K RTONWETCERNEP ................................ ................................ ......... 39
C. EL P NIROTAILESUNSLMIEELRDGOEMR ................................ ..... 47
D. YTILIBA BLOARNPOITIDNOC ................................ .......................... 53
V. NOITATERPR ESDTINNSAIY LAATNAAD ................................ ................. 59
VI. SNOITADNE MS MNDOONCIAESRULCNOC ................................ .............. 63
.A SNOISULCNOC ................................ ................................ .................... 63
B. SNOITADNEMMOCER ................................ ................................ ....... 63
vi
XIDNE.PAPA LNTAOUII PTSD TRSRDUAEINOPRLHAGATETRIV
PLOTS (NOT INCLUDED IN ELECTRONIC VERSION)
XIDNEPPA.B DESSECO RG PNFWDEIOANR TRAPLSAIILTRAP
TAUTPANDI (NOT INCLUDED IN ELECTRONIC VERSION)
XIDNEPPA.C KRO WET SE ERONEDOLCOFCRCUOS (NOT INCLUDED
IN ELECTRONIC VERSION)
TSIL FOSECNEREFER ................................ ................................ .............................. 65
YHPARGOILBIB ................................ ................................ ................................ ......... 67
INITIAL DISTRIBUTION LIST (NOT INCLUDED IN ELECTRONIC VERSION)
vii
SERU GTFISOFIL
.e1rugiF noru elNacigoloiB ................................ ................................ .................... 3
.e2rugiF ledo Mnorue NlaicifitrA ................................ ................................ ............ 4
.e3rugiF noitcn unFoitavit ccAitsigoL ................................ ................................ ..... 5
.e4rugiF noitcn unFoitavit cnAaissuaG ................................ ................................ ... 5
.e5rugiF gnis Unorue NlaicifitrA gninr aneoLitagaporpkcaB ................................ . 14
.e6rugiF tgenSini akrrTow te es rNono lorgCrnEiniarT ................................ ....... 34
.e7rugiF gniniarT rorrE no esolC krowteN tseTteS ................................ ............. 34
.e8rugiF tup nnIoitubir tsnrooCtcaF esolCkrowteN ................................ ............. 36
.e9rugiF l akurt ocewAstoelNC es o0l0C5 Pd&eStcid esrvP ................................ . 37
.e0r1ugiF Close Network Error ................................ ................................ .............. 38
.e1r1ugiF gntie nSkira o rt wrTngto enernciNrornEeiParT ................................ .... 40
.e2r1ugiF Training Error on Percent Network Test Set ................................ ........... 40
.e3r1ugiF Input Contribution Factors Percent Network ................................ .......... 42
.e4r1ugiF dlk eartt uonc.tweisctcFdvAereNerPP eg ntanh eCe cr0yru1aetDPu
005P & fSeoci rgPniso ldCna ................................ ................................ . 43
.e5r1ugiF Percent Network Error ................................ ................................ ........... 44
.e6r1ugiF d 0eS 0Ct&5cPid e .rlsPavu tsclAe dkorMow tlelNA ........................... 46
.e7r1ugiF ro rlre EdkorMow tleaNr uneoNitcid eeru PlganVis o0l0C5P&S
trahC ................................ ................................ ................................ ...... 47
.e8r1ugiF eso l0C05P &lSaut c. Asdvetcide rlPed onMoissergeR ........................... 50
.e9r1ugiF nlrtoeoridrasorhsMECergeR ................................ ................................ 51
viii