Integration between technical indicators and artificial neural networks for the prediction of the exchange rate: Evidence from emerging economies

Figures & data

Figure 1. Simple Moving Average

Figure 2. Momentum chart

Figure 4. RSI chart

Figure 5. ANN Structure

Figure 6. Artificial neural network Program.

Figure 3. MACD chart

Figure 7. Artificial neural network Parameters.

Table 1. Study variables

Sources	Symbol	Indicator	Variable
Technical analysis indicators based on historical exchange rates	SMASj,N	The simple moving average (SMA) of the country’s currency (J) based on (N) period.	Independent
	MSj,N	The momentum of the country’s currency (J) based on (N) period.
	MACDSj,N	The Moving average convergence divergence (MACD) of the country’s currency (J) based on (N) period.
	RSISj,N	The relative strength index (RSI)of the country’s currency (J) based on (N) period.
ANN	SjT	The exchange rate of the country’s currency (J) at the time (T)	Dependent

Table 2. Components of the study sample

No.	Abbreviations	Emerging economy currency	Emerging economy
1	USD/AED	UAE Dirham	UNITED ARAB EMIRATES (THE)
2	USD/ARS	Argentine Peso	ARGENTINA
3	USD/BRL	Brazilian Real	BRAZIL
4	USD/CLP	Chilean Peso	CHILE
5	USD/CNY	Yuan Renminbi	CHINA
6	USD/EGP	Egyptian Pound	EGYPT
7	USD/ETB	Ethiopian Birr	ETHIOPIA
8	USD/INR	Indian Rupee	INDIA
9	USD/IQD	Iraqi Dinar	IRAQ
10	USD/JOD	Jordanian Dinar	JORDAN
11	USD/KES	Kenyan Shilling	KENYA
12	USD/MAD	Moroccan Dirham	MOROCCO
13	USD/NGN	Naira	NIGERIA
14	USD/PEN	Nuevo Sol	PERU
15	USD/PKR	Pakistan Rupee	PAKISTAN
16	USD/PLN	Zloty	POLAND
17	USD/RUB	Russian Ruble	RUSSIAN FEDERATION (THE)
18	USD/SDG	Sudanese Pound	SUDAN (THE)
19	USD/TND	Tunisian Dinar	TUNISIA
20	USD/TRY	Turkish Lira	TURKEY
21	USD/TWD	New Taiwan Dollar	TAIWAN (PROVINCE OF CHINA)
22	USD/VND	Dong	VIET NAM
23	USD/XAF	CFA Franc BEAC	GABON
24	USD/ZAR	Rand	SOUTH AFRICA

Figure 8. Distribution of the hold rate for USD/AED.

Figure 9. Distribution of the hold rate for USD/ARS.

Figure 10. Distribution of the hold rate for USD/BRL.

Figure 11. Distribution of the hold rate for USD/CLP.

Figure 12. Distribution of the hold rate for USD/CNY.

Figure 13. Distribution of the hold rate for USD/EGP.

Figure 14. Distribution of the hold rate for USD/ETB.

Figure 15. Distribution of the hold rate for USD/INR.

Figure 16. Distribution of the hold rate for USD/IQD.

Figure 17. Distribution of the hold rate for USD/JOD.

Figure 18. Distribution of the hold rate for USD/KES.

Figure 19. Distribution of the hold rate for USD/MAD.

Figure 20. Distribution of the hold rate for USD/NGN.

Figure 21. Distribution of the hold rate for USD/PEN.

Figure 22. Distribution of the hold rate for USD/PKR.

Figure 23. Distribution of the hold rate for USD/PLN.

Figure 24. Distribution of the hold rate for USD/RUB.

Figure 25. Distribution of the hold rate for USD/SDG.

Figure 26. Distribution of the hold rate for USD/TND.

Figure 27. Distribution of the hold rate for USD/TRY.

Figure 28. Distribution of the hold rate for USD/TWD.

Figure 29. Distribution of the hold rate for USD/VND.

Figure 30. Distribution of the hold rate for USD/XAF.

Figure 31. Distribution of the hold rate for USD/ZAR.

Table 3. Cross-sectional analysis between actual values of exchange rates and four indicators of technical analysis

Model 1: Fixed-effects, using 1920 observations Included 24 cross-sectional units Time-series length = 80 Dependent variable: AS
	Coefficient	Std. Error			t-ratio		p-value
const	1.19672	0.00251418			476.0		<0.0001	***
SMA	2.65370e-05	0.000170104			0.1560		0.8760
M	−9.09544e-06	5.34970e-07			−17.00		<0.0001	***
MACD	−5.59764e-05	2.37798e-05			−2.354		0.0187	**
RSI	0.00155946	0.000326602			4.775		<0.0001	***
Mean dependent var	1.192728			S.D. dependent var		2.202823
Sum squared resid	0.021908			S.E. of regression		0.003403
LSDV R-squared	0.999998			Within R-squared		0.171205
LSDV F(27, 1892)	29783994			P-value(F)		0.000000
Log-likelihood	8201.372			Akaike criterion		−16346.74
Schwarz criterion	−16191.06			Hannan-Quinn		−16289.46
rho	0.972808			Durbin-Watson		0.028434

Joint test on named regressors -

Test statistic: F(4, 1892) = 97.7083

with p-value = P(F(4, 1892) > 97.7083) = 1.15584e-075

Test for differing group intercepts -

Null hypothesis: The groups have a common intercept

Test statistic: F(23, 1892) = 1.83428e+006

with p-value = P(F(23, 1892) > 1.83428e+006) = 0

Table 4. Enter technical indicators for the study model

Variables Entered/Removed
Model	Variables Entered	Variables Removed	Method
1	MACD^a	.	Enter
2	M^a	.	Enter
3	RSI^a	.	Enter
4	MA^a	.	Enter

a. All requested variables entered.

b. Dependent Variable: S

Table 5. Significant of model layers

ANOVA
Model		Sum of Squares	df	Mean Square	F	Sig.
1	Regression	0.015625	1	0.015625	0.003218	0.954766^a
	Residual	9311.794	1918	4.85495
	Total	9311.81	1919
2	Regression	57.47017	2	28.73508	5.952359	0.002648^b
	Residual	9254.34	1917	4.827512
	Total	9311.81	1919
3	Regression	65.30612	3	21.76871	4.510769	0.003691^c
	Residual	9246.504	1916	4.825942
	Total	9311.81	1919
4	Regression	8823.271	4	2205.818	8646.474	0.000^d
	Residual	488.5392	1915	0.255112
	Total	9311.81	1919

a. Predictors: (Constant), MACD

b. Predictors: (Constant), MACD, M

c. Predictors: (Constant), MACD, M, RSI

d. Predictors: (Constant), MACD, M, RSI, MA

e. Dependent Variable: S

Table 6. Cross-sectional analysis between expected and actual values of exchange rates

Model 2: WLS, using 1920 observations Included 24 cross-sectional units Dependent variable: AS Weights based on per-unit error variances
	Coefficient	Std. Error		t-ratio	p-value
Const.	0.0589883	0.000889447		66.32	<0.0001	***
PS	0.897423	0.00152651		587.9	<0.0001	***
Statistics based on the weighted data:
Sum squared resid	1583.125		S.E. of regression		0.908518
R-squared	0.994481		Adjusted R-squared		0.994478
F (1, 1918)	345616.2		P-value(F)		0.000000
Log-likelihood	−2539.155		Akaike criterion		5082.309
Schwarz criterion	5093.429		Hannan-Quinn		5086.401
Statistics based on the original data:
Mean dependent var	1.117683		S.D. dependent var		1.993524
Sum squared resid	403.3134		S.E. of regression		0.458561

Table 7. Wilcoxon signed ranks test outputs

Test Statistics
AS - PS
Z	−12.8821^a
Asymp. Sig. (2-tailed)	0.000

a. Based on positive ranks.

b. Wilcoxon Signed Ranks Test

Table 8. Kruskal-Wallis test outputs

Test Statistics
	D
Chi-Square	1910.124
df	23
Asymp. Sig.	0.000

a. Kruskal Wallis Test

b. Grouping Variable: CURR