Liu, Yadong; Na hee Nak nasukanjn; Tamp asi , Anukul; Ra anadam ongakso n,
Tana a
A icle
Do c ude oil, gold and he US dolla con ibu e o Bi coin
in es men decisions? An ANN-DCC-GARCH app oach
Asian Jou nal o Economics and Banking (AJEB)
P o ided in Coope a ion wi h:
Ho Chi Minh Uni e si y o Banking (HUB), Ho Chi Minh Ci y
Sugges ed Ci a ion: Liu, Yadong; Na hee Nak nasukanjn; Tamp asi , Anukul;
Ra anadam ongakso n, Tana a (2024) : Do c ude oil, gold and he US dolla con ibu e o Bi coin
in es men decisions? An ANN-DCC-GARCH app oach, Asian Jou nal o Economics and Banking
(AJEB), ISSN 2633-7991, Eme ald, Leeds, Vol. 8, Iss. 1, pp. 2-18,
h ps://doi.o g/10.1108/AJEB-10-2023-0106
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/334110
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/4.0/
Do c ude oil, gold and he US dolla
con ibu e o Bi coin in es men
decisions? An ANN-DCC-
GARCH app oach
Yadong Liu, Na hee Nak nasukanjn, Anukul Tamp asi and
Tana a Ra anadam ongakso n
In e na ional College o Digi al Inno a ion, Chiang Mai Uni e si y,
Chiang Mai, Thailand
Abs ac
Pu pose –Bi coin (BTC) is signi ican ly co ela ed wi h global inancial asse s such as c ude oil, gold and he
US dolla . BTC and global inancial asse s ha e become mo e closely ela ed, pa icula ly since he ou b eak o
he COVID-19 pandemic. The pu pose o his pape is o o mula e BTC in es men decisions wi h he aid o
global inancial asse s.
Design/me hodology/app oach –This s udy sugges s a mo e accu a e p edic ion model o BTC ading
by combining he dynamic condi ional co ela ion gene alized au o eg essi e condi ional he e oscedas ici y
(DCC-GARCH) model wi h he a i icial neu alne wo k (ANN). The DCC-GARCH model o e s signi ican inpu
in o ma ion, including dynamic co ela ion and ola ili y, o he ANN. To analyze he da a e ec i ely, he
s udy di ides i in o wo pe iods: be o e and du ing he COVID-19ou b eak. Each pe iod is hen u he di ided
in o a aining se and a p edic ion se .
Findings –The empi ical esul s show ha BTC and gold ha e he highes posi i e co ela ion compa ed wi h
c ude oil and he USD, while BTC and he USD ha e a dynamic and nega i e co ela ion. Mo e impo an ly, he
ANN-DCC-GARCH model had a cumula i e e u n o 318% be o e he ou b eak o he COVID-19 pandemic and
can dec ease loss by 50% du ing he COVID-19 pandemic. Mo eo e , he isk-a e se can u n a loss in o a p o i
o abou 20% in 2022.
O iginali y/ alue –The empi ical analysis p o ides echnical suppo and decision-making e e ence o
in es o s and inancial ins i u ions o make in es men decisions on BTC.
Keywo ds Co ela ion, Vola ili y, Po olio, ANN, BTC
Pape ype Resea ch pape
1. In oduc ion
In Janua y 2009, Bi coin (BTC) was in oduced as he wo ld’s pionee c yp og aphic digi al
cu ency (Bashe and Sado sky, 2022). BTC o e s lowe ansac ion expenses compa ed o
o he c yp ocu encies (Kayal and Rohilla, 2021) and ope a es independen ly om cen al
banks (Bau and Dimp l, 2021). In addi ion, BTC allows indi iduals o mine, pu chase, sell, o
ecei e i , while ensu ing use anonymi y du ing ansac ions (Chen, 2023). Consequen ly,
inancial ins i u ions and in es o s a e inc easingly a ac ed o he BTC ma ke . The e is
also he ac ha BTC is a high- isk, high- e u n inancial asse (Baek and Elbeck, 2015;
Huang e al., 2019;Cheah e al., 2022). Du ing he COVID-19 pandemic, which is 2021, he BTC
AJEB
8,1
2
© Yadong Liu, Na hee Nak nasukanjn, Anukul Tamp asi and Tana a Ra anadam ongakso n.
Published in Asian Jou nal o Economics and Banking. Published by Eme ald Publishing Limi ed. This
a icle is published unde he C ea i e Commons A ibu ion (CC BY 4.0) licence. Anyone may
ep oduce, dis ibu e, ansla e and c ea e de i a i e wo ks o his a icle ( o bo h comme cial and non-
comme cial pu poses), subjec o ull a ibu ion o he o iginal publica ion and au ho s. The ull e ms o
his licence may be seen a h p://c ea i ecommons.o g/licences/by/4.0/legalcode
Con lic s o in e es : The au ho s decla e ha hey ha e no con lic o in e es .
The cu en issue and ull ex a chi e o his jou nal is a ailable on Eme ald Insigh a :
h ps://www.eme ald.com/insigh /2615-9821.h m
Recei ed 31 Oc obe 2023
Re ised 7 Decembe 2023
Accep ed 19 Decembe 2023
Asian Jou nal o Economics and
Banking
Vol. 8 No. 1, 2024
pp. 2-18
Eme ald Publishing Limi ed
e-ISSN: 2633-7991
p-ISSN: 2615-9821
DOI 10.1108/AJEB-10-2023-0106
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
p ice exceeds $68,000 pe coin. And a he end o 2022, he BTC p ice alls below $20,000
again. I is no ed ha , he BTC p ice is ex emely ola ile. This is why many specula o s and
inancial ins i u ions a e keen on in es ing in BTC. Specula o s a e e y in e es ed in making
good in es men decisions o a be e p edic ion o he BTC p ice.
A di e en s ages, BTC ading decisions should be di e en . Figu e 1 shows he daily
closing p ices o BTC om 2014 o 2022. A e he ou b eak o COVID-19 pandemic, BTC
e lec in o bull ma ke om 2019 o ea ly 2021, while he BTC ma ke en e s a comple e bea
ma ke om he second hal o 2021. BTC p ices all back o 2020 le els. This demons a es
ha BTC p o ed o be a highly aluable in es men un il 2021. Undoub edly, BTC in es o s
should conside implemen ing dis inc in es men ac ics p io o and ollowing he ou b eak
o he COVID-19 pandemic. I is obse ed ha BTC in es o s and inancial ins i u ions a e
keen o make mo e p o i s in bull ma ke s and educe in es men isk in bea ma ke s.
The e o e, how should in es o s make BTC in es men decisions in di e en pe iods? Is i
possible o achie e his expec a ion h ough a o ecas ing me hodology? These a e he
ques ions ha need o be add essed in his s udy. Some schola s belie e ha he dec ease in
BTC’s alue is in luenced by he p ice o he US dolla , gold and c ude oil, (Dyh be g, 2016;
Bou i e al., 2018;G obys, 2021;Bani-Khala and Taspina , 2023) e c. The US dolla , gold and
c ude oil also a e ega ded as adi ional hedging asse s (Wang e al., 2021). No ably, as he
BTC ma ke has boomed, some esea che s ha e ound ha BTC has also become an
essen ial hedging asse (Bou i e al., 2018;Wang e al., 2019a). In addi ion, BTC and he
adi ional inancial asse s ha e become mo e closely ela ed, pa icula ly since he ou b eak
o he COVID-19 pandemic. BTC and he global inancial asse s ha e become mo e closely
ela ed, especially since he ou b eak o he new c own epidemic. I is well known ha he
iche he in o ma ion known, he mo e bene icial i is o ou in es men decisions. Based on
his ac , an in e es ing ques ion is whe he he co ela ion be ween adi ional inancial
asse s and BTC can be bene icial o BTC in es men ansac ions?
This pape aims o o ecas BTC in es men decisions by in eg a ing he dynamic
condi ional co ela ion gene alized au o eg essi e condi ional he e oscedas ici y (DCC-
GARCH) model and a i icial neu al ne wo k (ANN) me hods. B ie ly, we add he co ela ion
and co a iance be ween BTC and he global inancial asse s (gold, US dolla and c ude oil)
and ola ili y o BTC o p edic ing i s ading decisions. We compa e and analyze whe he
he inclusion o he co ela ion and ola ili y can enhance he p ecision o BTC ading
p edic ion. On he o he hand, we compa e he di e ences o ANN-DCC-GARCH models o
BTC ading p edic ion be o e and a e he ou b eak o COVID-19 pandemic.
0
10,000
20,000
30,000
4
0,000
50,000
60,000
70,000
14 15 1
6
17 18 19 2
0
21 22
Bi coin p ice
Sou ce(s): Figu e by au ho s
Figu e 1.
Bi coin p ices om
Sep . 2014 o Dec. 2022
Bi coin: Oil,
gold and USD
3
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
The main con ibu ions o his pape a e: i s , his pape p oposes he ANN-DCC-GARCH
model o he i s ime and uses i o in es men decision p edic ion in BTC. Second, om
he pe spec i e o empi ical analysis, we alida e he bene i s o adi ional inancial asse s on
BTC in es men ading decisions. Thi d, we compa e he di e ences in BTC in es men
ading decisions p io o and ollowing he occu ence o he COVID-19 pandemic. Fou h,
we also ind ha o di e en isk p e e ences, he ANN-DCC-GARCH model p edic s e y
la ge di e ences in BTC ading decisions.
The eminde o his s udy is o ganized as ollows: Sec ion 2 e iews and summa izes he
p e ious s udies ela ed o BTC and he global inancial asse s, and Sec ion 3 gi es he da a
sou ces and makes he da a desc ip i e. Sec ion 4 in oduces he DCC-GARCH models, ANN
and he p ocess o he ANN-DCC-GARCH app oach. Sec ion 5 exhibi s he p ac ical indings
and he conclusion is gi en a he end.
2. Li e a u e e iew
Some schola s ha e ied o in e he p ice end o BTC by s udying he associa ion be ween
BTC and o he inancial commodi ies o gi e sugges ions o BTC in es men (E das and
Cagla , 2018;Al Mamun e al., 2020;Oko ie and Lin, 2020). Fo example, Bau and Dimp l
(2021) es ima ed he connec ion be ween gold and BTC and concluded ha he e is a
subs i u ion o ca ch-up e ec be ween BTC and gold. Acco ding o Selmi e al. (2018), BTC
was ega ded as a hedge and a di e si ie o oil p ice luc ua ions. Addi ionally, in es o s
can exchange gold o BTC o ice e sa, in o de o align hei ma ke exposu e wi h he
weigh o gold. Kwon (2020) ound ha BTC and USD we e nega i e co ela ed and igu ed
ou ha BTC can be a hedge o he dolla .
Kang e al. (2019) s udied on co-mo emen s be ween BTC and Gold o easy di e si ica ion
and hedge. By analyzing ex eme ela ions, Das e al. (2020) disco e ed ha BTC su passes
gold and commodi y in e ms o hedging c ude oil implied ola ili y. Nguyen (2022), h ough
co ela ion analysis, demons a ed he absence o a ola ili y spillo e e ec om BTC o he
S&P500. Zhang and Mani (2021) ound ha dynamic co ela ion be ween gold and BTC is
posi i e om 2020 o 2021. In addi ion o co ela ion analysis be ween BTC and o he asse s,
he e a e also o he kinds o ela ionship a e ocused, such as isk spillo e e ec o con agion
(Khal aoui e al., 2023;Rehman e al., 2023), causali y ela ionship (Palazzi e al., 2021;Rehman
and Kang, 2021) and coin eg a ion (Hu e al., 2020;Tan e al., 2021).
The e ha e been a ious s udies conduc ed by Jana and Das (2020),A ouxe e al. (2022),
Sa kodie e al. (2022) and Liu e al. (2023), which demons a e he no able in luence o he
COVID-19 pandemic on BTC’s p ice luc ua ions and ola ili y. Ob iously, he associa ion
be ween BTC and o he inancial asse s should also change signi ican ly du ing he epidemic.
Ma iana e al. (2021) ound a nega i e co ela ion be ween BTC and S&P500 du ing he
epidemic. Ja e~
no e al. (2021) showed a s ong co ela ion be ween c yp ocu encies and c ude
oil du ing he COVID-19 pandemic. Bhuiyan e al. (2021) demons a ed a causal ela ionship
be ween BTC and gold. Tiwa i e al. (2024) measu ed he co ela ion be ween BTC and he
clean enewable ene gy s ock index, and hei s udy shows ha BTC o e s a obus hedging
and hedging mechanism compa ing wi h in es men s in enewable ene gy. Du a e al. (2020)
con i med ha BTC can di e si y he in es men isk o c ude oil du ing he epidemic. I is
e iden ha BTC is mo e a ec ed by he epidemic, and hen he gap in BTC in es men s
should be la ge be o e and a e he epidemic.
I is obse ed om abo e discussion ha he e is an inc easing amoun o esea ch on he
associa ion be ween BTC and he global inancial asse s ollowing he occu ence o he
COVID-19 epidemic. Some models, such as quan ile-on-quan ile eg ession, Wa ele quan ile,
quan ile causali y, ime- a ying pa ame e ec o au o eg ession (TVP-VAR) and DCC-
GARCH models a e popula applied in o associa ion analysis be ween BTC and o he
AJEB
8,1
4
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
inancial asse s. Kuma and Padakandla (2022) es ed he applicabili y o gold and BTC as
hedges based on he Wa ele quan ile co ela ion me hod in he con ex o he COVID-19-
ela ed s ock ma ke c ash. The esul s show ha in he long and sho e m, gold
consis en ly demons a es i s epu a ion as a sa e ha en asse ac oss all ma ke s, while BTC
p esen s a combina ion o ou comes. Fu he mo e, gold can also se e as a eliable ool o
hedging isks and di e si ying in es men s. Nguyen (2022) examined he impac o COVID-
19 and o he pe iods o unce ain y in he s ock ma ke on BTC.
Huang e al. (2023) employed a TVP-VAR model wi h s ochas ic ola ili y o examine he
ma ke ela ionship be ween BTC and g een asse s, bo h be o e and du ing he COVID-19
pandemic. Sha ma e al. (2023) used quan ile- o-quan ile eg ession (QQR) and quan ile G ange
causali y me hods o in es iga e he associa ion be ween BTC and he g een economy.
Meanwhile, Zhang and Mani (2021) employed a ious GARCH models, including Exponen ial
Gene alized Au o eg essi e Condi ional He e oskedas ici y (EGARCH), Glos en-Jaganna han-
Runkle (GJR)-GARCH and Mul i a ia e GARCH (MGARCH) models, o explo e he ola ili y,
asymme y and co ela ion among well-known c yp ocu encies and gold. The s udy also
assessed asse co ela ions using he DCC app oach and e ealed ha he linkage be ween gold
and BTC in ensi ied du ing he COVID-19 pandemic, eaching i s highes poin a he peak o he
pandemic. Wang e al. (2021) also used he DCC-GARCH models o measu e he dynamic
co ela ion be ween hedge asse s and global s ock ma ke s. A la ge numbe o s udies ha e
applied he DCC-GARCH models o analyze dynamic co ela ions and se e as a basis o o he
s udies, such as alue-a - isk, po olio and sys emic inancial isk. We can ind ha he DCC-
GARCH models a e well used o measu e ime- a ying co ela ions.
In ecen yea s, many schola s di ec ly s udied on p edic ion o BTC (Adcock and G adoje ic,
2019;A salakis e al., 2019;Pabuccu e al., 2020;Hau e al., 2021;T ipa hi and Sha ma, 2023;Wang
and Hausken, 2022). Huang and Gao (2022) used leas absolu e sh inkage and selec ion ope a o
(LASSO) app oach o p edic BTC e u ns om 2018 o 2019. Wang e al. (2019b) used he
au o eg essi e jump in ensi y (ARJI) model o p edic he ola ili y o BTC in 2018. Nakano e al.
(2018) conduc ed expe imen s on high equency BTC p ice da a using a i icial ne wo ks o
e alua e a ious ading s a egies. Adcock and G adoje ic (2019) u ilized neu al ne wo ks o
make p edic ions on BTC p ices. Pabuccu e al. (2020) employed se e al machine lea ning and
Logi models as ools o o ecas ing BTC p ices. Mos o s udies showed ha machine lea ning
me hods ha e a be e pe o mance han ime se ies models o BTC p edic ion.
F om he a o emen ioned li e a u e, i is clea ha BTC and he global inancial asse s ha e
some ela ionship. Nume ous esea che s ha e gi en special ocus o he dependence o BTC
and he inancial asse s as well as he isk p emiums be ween hem. Many schola s ha e poin ed
ou ha he es ima ion o inancial p oduc co ela ions and isk spillo e is bene icial o
po olio and in es men decisions. Howe e , none o hem adop a quan i a i e app oach o
elucida e in es men decisions ega ding inancial p oduc s. Since BTC and he global inancial
asse s ha e ce ain ela ionships, is i bene icial o make in es men decisions in BTC based on
cla i ying hese ela ionships? I his asse ion is co ec , hen he inding has impo an p ac ical
implica ions o bo h BTC in es o s and inancial ins i u ions.
3. The da a
This s udy collec s he da a se om he Wind da abase. The da a is daily and he sample
pe iod co e s om Sep embe 17, 2014 o Decembe 23, 2022. Inpu a iables include daily
high, low and open p ices o BTC and bina y a iables o gold, US dolla and c ude oil. A
0 ep esen s a alling p ice and a 1 indica es a ising p ice. All inpu a iables a e one-pe iod
lag. All a iables, excep bina y a iables, a e no malized using he en opy weigh me hod.
The 2019 da a a e ega ded as ou -o -sample da a be o e he COVID-19 ou b eak, and he
2022 da a a e ou -o -sample a e he ou b eak o COVID-19.
Bi coin: Oil,
gold and USD
5
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
Table 1 epo s he da a desc ip i e o he ull sample and he sub-samples o 2019 and
2022. I is clea ha he sample mean alue in 2019 is 7383.25 and 28355.82 in 2022, which
implies he p ice o BTC changes g ea ly p io o and ollowing he occu ence o he COVID-
19 pandemic. BTC p ice ola ili y is also high, as can be seen om he minimum and
maximum alues in bo h 2019 and 2022.
4. Me hodology
We i s ly in oduce DCC-GARCH models and hen demons a e ANN me hod. Finally, s eps
o using he ANN-based DCC-GARCH models a e summa ized.
4.1 DCC-GARCH model
The GJR-GARCH (p, q) model can be de ined as below:
σ
2
¼
ω
þX
p
i¼1ð
α
iþγiI -iÞ
ε
2
-iþXq
j¼1βj
σ
2
-j;(1)
whe e:
I -1¼0i
ε
-1≥0;
1i
ε
-1<0;(2)
In addi ion, γ ep esen s he le e age e ec Also, he e o e m
ε
¼
σ
z and z is
independen ly and iden i y s uden - dis ibu ion.
The DCC s uc u e is exp essed as:
Q ¼ 1- X
M
m¼1
θ1m-X
N
n¼1
θ2n!QþXM
m¼1θ1m
ε
-m
ε
，
-mþX
N
n¼1
θ2nQ -n;(3)
R ¼Q*-1
Q Q*-1
;(4)
whe e Q ep esen s he uncondi ional co a iance o he s anda dized esiduals ha a ise om
he i s s ep es ima ion and
Q*
¼2
6
4ffiffiffiffiffiffi
q11
p 0
.
.
.
1.
.
.
0 ffiffiffiffiffiffiffi
qkk
p3
7
5;(5)
Bi coin To al sample Subsample in 2019 Subsample in 2022
Mean 12860.64 7383.25 28355.82
Median 7106.54 7853.04 23222.24
Maximum 67566.83 13016.23 47465.73
Minimum 178.10 3399.47 15787.28
S d. De . 16197.13 2656.19 10183.79
Skewness 1.54 0.057687 0.44
Ku osis 4.28 1.80 1.55
Obse a ions 2,134 258 254
Sou ce(s): Au ho s’compu a ion
Table 1.
The da a desc ip i e
and s a is ics
AJEB
8,1
6
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
So ha Q*
o ma ix Q*
, i will be a diagonal ma ix whe e each elemen is he squa e oo o
he co esponding diagonal elemen o ma ix Q
. As o ma ix R
, i s ypical elemen will be
in he o m o
ρ
ij ¼qij
ffiffiffiffiffiffiffi
qiiqjj
p.
4.2 A i icial neu al ne wo k
ANNs a e employed o model he non-linea connec ions be ween inpu and ou pu a iables
(Mahdiani and Khamehchi, 2016;Rod iguez e al., 2022). The main unc ion o ANN is
p edic ion. The ANN includes h ee laye s o uni s, namely, he inpu , hidden and ou pu laye s.
The ANN has also many nodes ha a eused olink he h ee laye s anda e called neu ons. The
basic p inciple o an ANN is ha each neu on akes he ini ial inpu alues and mul iplies hem
by a ce ain weigh , adds he alues o he o he inpu s in o his neu on and inally wo ks ou a
sum, which is hen adjus ed by he bias o he neu on and inally no malizes he ou pu alues
wi h an exci a ion unc ion (Kulka ni and Haida , 2009;Gup a and Nigam, 2020).
The mo e alid in o ma ion is a ailable as inpu a iables, he mo e accu a e he
p edic ions o he ANN. The DCC-GARCH model can ob ain dynamic co ela ions and
ola ili ies, which a e aluable in o ma ion in inancial ma ke analysis (Yıldı ıme al., 2022).
The e o e, we a e well placed o use dynamic co ela ion and ola ili y as inpu a iables o
he ANN app oach. In iew o his, we me ge he ANN and DCC-GARCH models and e e o
hem as he ANN-based DCC-GARCH model. Following Bah ambeygi and Moeinzadeh
(2017), we cons uc he a chi ec u e o ANN-based DCC-GARCH model in Figu e 2. The
closing, opening, high and low p ices o BTC a one-lag pe iod a e used as he mos basic
inpu a iables. Bina yi; 1 ep esen s he bina y a iables o BTC and he global inancial
asse s a ime 1. The bina y a iable is equal o 1 when he p ice ises and 0 o he wise.
ρ
ij; 1is he co ela ion be ween BTC and he global inancial asse s, while
σ
ij; 1 ep esen s
hei co a iance and he ola ili y o BTC.
The model is pe o med in he ollowing s eps.
1. The in o ma ion is sepa a ed in o a g ouping o da a o aining pu poses and ano he
o p edic ion.
2. The DCC-GARCH model is employed o app oxima e dynamic co ela ions, ola ili ies
and co a iances o he ull sample. One-lag pe iod o all a iables, including he dynamic
co ela ion, ola ili y, co a iance as well as o he indica o s a e used as inpu a iables in
he aining da ase , while he dummy a iable o log e u ns was he ou pu a iable.
3. All inpu a iables in addi ion o he dummy a iables a e no malized (Kulka ni and
Haida , 2009). The no malized p ocess is as ollows:
Xi ¼xi -minðxiÞ
maxðxiÞ-minðxiÞ;(6)
whe e xi ep esen s he i- h inpu a iable a ime and Xi s anda ds o no malized alues.
4. A me hod is used o de e mine he numbe o hidden neu ons. Acco ding o Rod iguez
e al. (2022), he e is cu en ly no di ec app oach o asce ain he quan i y o neu ons.
Following Zeng e al. (2023), he numbe o possible neu ons is calcula ed as ollows:
L¼ffiffiffiffiffiffiffiffiffiffiffiffiffi
wþ
ω
pþ
α
;(7)
whe e wand
ω
deno e he numbe o nodes in he ou pu laye and inpu laye , espec i ely,
and
α
is an in ege numbe om 1 o 10. L ep esen s he numbe o nodes in he hidden laye .
Bi coin: Oil,
gold and USD
7
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
5. The aining o ANN models wi h a ying numbe s o hidden laye s is conduc ed using
he esilien back p opaga ion algo i hm, inco po a ing weigh back acking. The bes
model is de e mined based on i s highes accu acy achie ed on he aining se . The
accu acy o p edic ion is compu ed using he ollowing o mula:
AP ¼1
T1X
T1
¼1IbP >0:5*I
ð >0Þ;(8)
whe e T
1
ep esen s he numbe o obse a ions om 2014 o 2018, and om 2014 o 2021 in
his s udy, espec i ely. b
P deno es he es ima ed p obabili y alue, and I is an indica o
unc ion, is he log e u n o BTC. The model is imp o ed as he alue o AP inc eases. F om
his we can de e mine he op imal numbe o neu ons, i.e. he bes model.
6. Finally, we b ing he p edic ion da ase in o he bes ANN model o p edic he ou pu .
5. Empi ical esul s
The empi ical esul s in his s udy a e calcula ed using he RS udio so wa e. We mainly
u ilized he “ uga ch”and “neu alne ”packages in he Rso wa e. In he ANN models,
Opening
p ice
Closing
p ice
High
p ice
Low
p ice
Bina y
i -1
ρ
ij -1
σ
ij -1
Inpu laye
H
H
H
...
Hidden
laye
Ou pu
Ou pu
laye
Sou ce(s): Figu e by au ho s
Figu e 2.
The a chi ec u e o
ANN-based DCC-
GARCH model
AJEB
8,1
8
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
we used he algo i hm o esilien backp opaga ion wi h weigh back acking and he c oss-
en opy me hod o calcula e he con e gence e o . In he ANN-DCC-GARCH model,
co ela ions be ween BTC and he h ee asse s (c ude oil, USD and gold), co a iance o BTC
wi h he h ee asse s and ola ili y o BTC a e no malized ollowing Equa ion (8). The one-lag
pe iod da a a e ega ded as he inpu a iables.
5.1 Es ima e esul s o he DCC-GARCH models
Figu e 3 shows he DCC be ween BTC and he global inancial asse s. Ob iously, he
co ela ions be ween BTC and he global inancial asse s ha e a clea dynamic na u e.
Posi i e co ela ions domina e BTC and gold, as well as BTC and c ude oil. The co ela ion
be ween BTC and gold is sligh ly highe han ha be ween BTC and c ude oil o mos o he
ime pe iod, which was e idenced by Gkillas e al. (2022), BTC and he US dolla ha e a
nega i e co ela ion, and he nega i e co ela ion becomes signi ican ly s onge in 2020 and
2022. These indings acili a e he analysis o BTC and adi ional inancial asse po olios.
Howe e , he esul s om he co ela ion es ima es only do no allow o a subs an i e
analysis o BTC in es men decisions.
Figu e 4 shows he ola ili y o BTC and he co a iance o BTC wi h he global inancial
asse s. We can clea ly see ha BTC has a high ola ili y, pa icula ly in 2017 and 2020.
Smales (2019) and Long e al. (2021) also e idence ha BTC has a g ea e ola ili y han gold.
In 2020, he co a iance be ween BTC and gold and BTC and c ude oil a e qui e s onge ,
which may a ibu e o he ou b eak o he COVID-19 pandemic. In pa icula , he co a iance
be ween BTC and c ude oil changes e y much in 2020. The co a iance be ween BTC and he
US dolla exhibi s a yea - o-yea s eng hening end, wi h he mos p onounced
s eng hening end in 2022.
5.2 P edic ion analysis o he ANN-DCC-GARCH models
We b ing he dynamic co ela ion and co a iance ob ained in he DCC-GARCH model as
inpu a iables in o he ANN model, hus es ing he p edic i e e ec o he ANN-DCC-
GARCH model on BTC ansac ions. In o de o de e mine whe he he esul s o he DCC-
GARCH model con ibu e o he ANN p edic ions, ou models a e cons uc ed. Model 1 is
an ANN model ha does no conside he es ima ed esul s o he DCC-GARCH model.
Model 2 is an ANN model ha adds only co a iance as inpu a iables. Model 3 is an ANN
–0.4
–0.3
–0.2
–0.1
0.0
0.1
0.2
0.3
0.4
14 15 16 17 18 19 20 21 22
C ude oil Gold USD
Sou ce(s): Figu e by au ho s
Figu e 3.
Pai wise co ela ions
be ween BTC and gold,
c ude oil and USD
Bi coin: Oil,
gold and USD
9
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
G obys, K. (2021), “When Bi coin has he lu: on Bi coin’s pe o mance o hedge equi y isk in he
ea ly wake o he COVID-19 ou b eak”,Applied Economics Le e s, Vol. 28 No. 10, pp. 860-865,
doi: 10.1080/13504851.2020.1784380.
Gup a, N. and Nigam, S. (2020), “C ude oil p ice p edic ion using a i icial neu al ne wo k”,P ocedia
Compu e Science, Vol. 170, pp. 642-647, doi: 10.1016/j.p ocs.2020.03.136.
Hau, L., Zhu, H., Shahbaz, M. and Sun, W. (2021), “Does ansac ion ac i i y p edic Bi coin e u ns?
E idence om quan ile-on-quan ile analysis”,The No h Ame ican Jou nal o Economics and
Finance, Vol. 55, 101297, doi: 10.1016/j.naje .2020.101297.
Hu, Y., Hou, Y.G. and Oxley, L. (2020), “Wha ole do u u es ma ke s play in Bi coin p icing?
Causali y, coin eg a ion and p ice disco e y om a ime- a ying pe spec i e?”,In e na ional
Re iew o Financial Analysis, Vol. 72, 101569, doi: 10.1016/j.i a.2020.101569.
Huang, W. and Gao, X. (2022), “LASSO-based high- equency e u n p edic o s o p o i able Bi coin
in es men ”,Applied Economics Le e s, Vol. 29 No. 12, pp. 1079-1083, doi: 10.1080/13504851.
2021.1908512.
Huang, J.-Z., Huang, W. and Ni, J. (2019), “P edic ing bi coin e u ns using high-dimensional echnical
indica o s”,The Jou nal o Finance and Da a Science, Vol. 5 No. 3, pp. 140-155, doi: 10.1016/j.
j ds.2018.10.001.
Huang, Y., Duan, K. and U quha , A. (2023), “Time- a ying dependence be ween Bi coin and g een
inancial asse s: a compa ison be ween p e-and pos -COVID-19 pe iods”,Jou nal o
In e na ional Financial Ma ke s, Ins i u ions and Money, Vol. 82, 101687, doi: 10.1016/j.in in.
2022.101687.
Jana, R.K. and Das, D. (2020), “Did Bi coin ac as an an ido e o he Chinese equi y ma ke and boos e
o Al coins du ing he No el Co ona i us ou b eak?”, SSRN 3544794, a ailable a : h ps://
pape s.ss n.com/sol3/pape s.c m?abs ac _id53544794 (accessed 4 Decembe 2023).
Ja e~
no, F., Gonz
alez, M.O., L
opez, R. and Ramos, A.R. (2021), “C yp ocu encies and oil p ice shocks: a
NARDL analysis in he COVID-19 pandemic”,Resou ces Policy, Vol. 74, 102281, doi: 10.1016/j.
esou pol.2021.102281.
Kang, S.H., McI e , R.P. and He nandez, J.A. (2019), “Co-mo emen s be ween Bi coin and Gold: a
wa ele cohe ence analysis”,Physica A: S a is ical Mechanics and i s Applica ions, Vol. 536,
120888, doi: 10.1016/j.physa.2019.04.124.
Kayal, P. and Rohilla, P. (2021), “Bi coin in he economics and inance li e a u e: a su ey”,SN
Business and Economics, Vol. 1 No. 7, p. 88, doi: 10.1007/s43546-021-00090-5.
Khal aoui, R., Hammoudeh, S. and Rehman, M.Z. (2023), “Spillo e s and connec edness among BRICS
s ock ma ke s, c yp ocu encies, and unce ain y: e idence om he quan ile ec o
au o eg ession ne wo k”,Eme ging Ma ke s Re iew, Vol. 54, 101002, doi: 10.1016/j.emema .
2023.101002.
Kulka ni, S. and Haida , I. (2009), “Fo ecas ing model o c ude oil p ice using a i icial neu al
ne wo ks and commodi y u u es p ices”, a Xi , a ailable a : h p://a xi .o g/abs/0906.4838
(accessed 4 Decembe 2023).
Kuma , A.S. and Padakandla, S.R. (2022), “Tes ing he sa e-ha en p ope ies o gold and bi coin in he
backd op o COVID-19: a wa ele quan ile co ela ion app oach”,Finance Resea ch Le e s,
Vol. 47, 102707, doi: 10.1016/j. l.2022.102707.
Kwon, J.H. (2020), “Tail beha io o Bi coin, he dolla , gold and he s ock ma ke index”,Jou nal o
In e na ional Financial Ma ke s, Ins i u ions and Money, Vol. 67, 101202, doi: 10.1016/j.in in.
2020.101202.
Liu, Y., Nak nasukanjn, N., Tamp asi , A. and Ra anadam ongakso n, T. (2023), “Compa ison o he
asymme ic ela ionship be ween bi coin and gold, c ude oil, and he US dolla be o e and a e he
COVID-19 ou b eak”,Jou nal o Risk and Financial Managemen , Vol. 16 No. 10, p. 455, doi: 10.
3390/j m16100455.
AJEB
8,1
16
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
Long, S., Pei, H., Tian, H. and Lang, K. (2021), “Can bo h Bi coin and gold se e as sa e-ha en
asse s?—a compa a i e analysis based on he NARDL model”,In e na ional Re iew o
Financial Analysis, Vol. 78, 101914, doi: 10.1016/j.i a.2021.101914.
Mahdiani, M.R. and Khamehchi, E. (2016), “A modi ied neu al ne wo k model o p edic ing he c ude
oil p ice”,In ellec ual Economics, Vol. 10 No. 2, pp. 71-77, doi: 10.1016/j.in ele.2017.02.001.
Ma iana, C.D., Ekapu a, I.A. and Husodo, Z.A. (2021), “A e Bi coin and E he eum sa e-ha ens o
s ocks du ing he COVID-19 pandemic?”,Finance Resea ch Le e s, Vol. 38, 101798, doi: 10.1016/
j. l.2020.101798.
Nakano, M., Takahashi, A. and Takahashi, S. (2018), “Bi coin echnical ading wi h a i icial neu al
ne wo k”,Physica A: S a is ical Mechanics and i s Applica ions, Vol. 510, pp. 587-609, doi: 10.
1016/j.physa.2018.07.017.
Nguyen, K.Q. (2022), “The co ela ion be ween he s ock ma ke and Bi coin du ing COVID-19 and
o he unce ain y pe iods”,Finance Resea ch Le e s, Vol. 46, 102284, doi: 10.1016/j. l.2021.
102284.
Oko ie, D.I. and Lin, B. (2020), “C ude oil p ice and c yp ocu encies: e idence o ola ili y
connec edness and hedging s a egy”,Ene gy Economics, Vol. 87, 104703, doi: 10.1016/j.eneco.
2020.104703.
Pabuccu, H., Ongan, S. and Ongan, A. (2020), “Fo ecas ing he mo emen s o Bi coin p ices: an
applica ion o machine lea ning algo i hms”,Quan i a i e Finance and Economics, Vol. 4 No. 4,
pp. 679-692, doi: 10.3934/QFE.2020031.
Palazzi, R.B., J
unio , G.D.S.R. and Klo zle, M.C. (2021), “The dynamic ela ionship be ween bi coin and
he o eign exchange ma ke : a nonlinea app oach o es causali y be ween bi coin and
cu encies”,Finance Resea ch Le e s, Vol. 42, 101893, doi: 10.1016/j. l.2020.101893.
Rehman, M.U. and Kang, S.H. (2021), “A ime– equency como emen and causali y ela ionship
be ween Bi coin hash a e and ene gy commodi y ma ke s”,Global Finance Jou nal, Vol. 49,
100576, doi: 10.1016/j.g j.2020.100576.
Rehman, M.U., Ka siampa, P., Zei un, R. and Vo, X.V. (2023), “Condi ional dependence s uc u e and
isk spillo e s be ween bi coin and ia cu encies”,Eme ging Ma ke s Re iew, Vol. 55, 100966,
doi: 10.1016/j.emema .2022.100966.
Rod iguez, M.R., Bes ei o, R., O ega, J.A., Fe nandez, M.D. and A ango, T. (2022), “E olu ion and
neu al ne wo k p edic ion o CO2 emissions in weaned pigle a ms”,Senso s, Vol. 22 No. 8,
p. 2910, doi: 10.3390/s22082910.
Sa kodie, S.A., Ahmed, M.Y. and Owusu, P.A. (2022), “COVID-19 pandemic imp o es ma ke signals
o c yp ocu encies–e idence om Bi coin, Bi coin Cash, E he eum, and Li ecoin”,Finance
Resea ch Le e s, Vol. 44, 102049, doi: 10.1016/j. l.2021.102049.
Selmi, R., Mensi, W., Hammoudeh, S. and Bouoiyou , J. (2018), “Is Bi coin a hedge, a sa e ha en o a
di e si ie o oil p ice mo emen s? A compa ison wi h gold”,Ene gy Economics, Vol. 74,
pp. 787-801, doi: 10.1016/j.eneco.2018.07.007.
Sha ma, G.D., Shahbaz, M., Singh, S., Chop a, R. and Ci uen es-Fau a, J. (2023), “In es iga ing he
nexus be ween g een economy, sus ainabili y, bi coin and oil p ices: con ex ual e idence om
he Uni ed S a es”,Resou ces Policy, Vol. 80, 103168, doi: 10.1016/j. esou pol.2022.103168.
Smales, L.A. (2019), “Bi coin as a sa e ha en: is i e en wo h conside ing?”,Finance Resea ch Le e s,
Vol. 30, pp. 385-393, doi: 10.1016/j. l.2018.11.002.
Tan, Z., Huang, Y. and Xiao, B. (2021), “Value a isk and e u ns o c yp ocu encies be o e and a e
he c ash: long- un ela ions and ac ional coin eg a ion”,Resea ch in In e na ional Business
and Finance, Vol. 56, 101347, doi: 10.1016/j. iba .2020.101347.
Tiwa i, A.K., Abakah, E.J.A., Rehman, M.Z. and Lee, C.C. (2024), “Quan ile dependence o Bi coin wi h
clean and enewable ene gy s ocks: new global e idence”,Applied Economics, Vol. 56 No. 3,
pp. 286-300, doi: 10.1080/00036846.2023.2167921.
Bi coin: Oil,
gold and USD
17
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
T ipa hi, B. and Sha ma, R.K. (2023), “Modeling bi coin p ices using signal p ocessing me hods,
Bayesian op imiza ion, and deep neu al ne wo ks”,Compu a ional Economics, Vol. 62 No. 4,
pp. 1919-1945, doi: 10.1007/s10614-022-10325-8.
Wang, G. and Hausken, K. (2022), “A Bi coin p ice p edic ion model assuming oscilla o y g ow h and
leng hening cycles”,Cogen Economics and Finance, Vol. 10 No. 1, 2087287, doi: 10.1080/
23322039.2022.2087287.
Wang, G., Tang, Y., Xie, C. and Chen, S. (2019a), “Is bi coin a sa e ha en o a hedging asse ? E idence
om China”,Jou nal o Managemen Science and Enginee ing, Vol. 4 No. 3, pp. 173-188, doi: 10.
1016/j.jmse.2019.09.001.
Wang, J.-N., Liu, H.C., Chiang, S.M. and Hsu, Y.T. (2019b), “On he p edic i e powe o ARJI ola ili y
o ecas s o Bi coin”,Applied Economics, Vol. 51 No. 44, pp. 4849-4855, doi: 10.1080/00036846.
2019.1602714.
Wang, P., Zhang, H., Yang, C. and Guo, Y. (2021), “Time and equency dynamics o connec edness
and hedging pe o mance in global s ock ma ke s: bi coin e sus con en ional hedges”,
Resea ch in In e na ional Business and Finance, Vol. 58, 101479, doi: 10.1016/j. iba .2021.101479.
Yıldı ım, D.Ç., Esen, €
O. and E u
g ul, H.M. (2022), “Impac o he COVID-19 pandemic on e u n and
isk ansmission be ween oil and p ecious me als: e idence om DCC-GARCH model”,
Resou ces Policy, Vol. 79, 102939, doi: 10.1016/j. esou pol.2022.102939.
Zeng, L., Li, L. and Jiang, Sh. (2023), “S ock p ice p edic ion based on BP neu al ne wo k”,Jou nal o
Yunnan Minzu Uni e si y, Vol. 5, pp. 658-665 (in Chinese) (Na u al Sciences Edi ion).
Zhang, S. and Mani, G. (2021), “Popula c yp oasse s (Bi coin, E he eum, and Dogecoin), Gold, and
hei ela ionships: ola ili y and co ela ion modeling”,Da a Science and Managemen , Vol. 4,
pp. 30-39, (in Chinese), doi: 10.1016/j.dsm.2021.11.001.
Co esponding au ho
Yadong Liu can be con ac ed a : [email p o ec ed]
Fo ins uc ions on how o o de ep in s o his a icle, please isi ou websi e:
www.eme aldg ouppublishing.com/licensing/ ep in s.h m
O con ac us o u he de ails: [email p o ec ed]
AJEB
8,1
18
Downloaded om h p://www.eme ald.com/ajeb/a icle-pd /8/1/2/9525948/ajeb-10-2023-0106.pd by ZBW Ge man Na ional Lib a y o Economics use on 16 Decembe 2025
