Available via license: CC BY
Content may be subject to copyright.
Copyright © 2018 Authors. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted
use, distribution, and reproduction in any medium, provided the original work is properly cited.
International Journal of Engineering & Technology, 7 (4.15) (2018) 400-404
International Journal of Engineering & Technology
Website: www.sciencepubco.com/index.php/IJET
Research paper
FoRex Trading Using Supervised Machine Learning
Thuy Nguyen Thi Thu1*, Vuong Dang Xuan2
1ThuongMai University, Mai Dich, Cau Giay, Hanoi, Vietnam
2Sonla Gifted HighSchool, Sonla, Vietnam
*Corresponding author E-mail: Thuynguyenthithu@tmu.edu.vn
Abstract
The exchange rate of each money pair can be predicted by using machine learning algorithm during classification process. With the help
of supervised machine learning model, the predicted uptrend or downtrend of FoRex rate might help traders to have right decision on
FoRex transactions. The installation of machine learning algorithms in the FoRex trading online market can automatically make the
transactions of buying/selling. All the transactions in the experiment are performed by using scripts added-on in transaction application.
The capital, profits results of use support vector machine (SVM) models are higher than the normal one (without use of SVM).
Keywords: Classification; Foreign Exchange rate; Supervised Machine Learning; Transaction.
1. Introduction
Supervised machine learning can apply into many areas in com-
puter science such as decision making, forecasting, and specially
is to predict stock price or money exchange rate, and so on. Su-
pervised machine learning means the teaching and monitoring the
computer to classify or cluster data from observed data set. With
the explosion of data nowadays, particular with business big data,
supervised machine learning techniques can support people to
look depth and analyze their data in order to extract useful infor-
mation for their company’s purposes. In stock market, the transac-
tions of money exchange can be predicted rise or fall with helping
of machine learning and the observed data set collected in the past.
The artificial intelligent systems are used with supervised machine
learning methods like Logic, Perceptron, Statistics methods
(Bayes Networks, Instance based) [2]. The goals are the classifica-
tion or regression dataset into alternative classes so-called as out-
puts. These target variable (y) is usually the nominal (discrete) or
continuous values [1]. For simplicity, the output of y in the classi-
fication problem can be seen as binary values such as 0 or 1.
Support Vector Machines (SVMs), which is one of Supervised
Machine Learning techniques, are the techniques that separate
data in hyperplane space into two data classes [3]. These tech-
niques create maximized margins and the distance between them
to separate data into alternative sides of hyperplane [2]. So, SVMs
can be a possible candidate of use to predict Foreign Exchange
(Forex) trend (Up or Down) in money exchange rate problems.
The predicting FoRex price is performed by using historical data
such as Open", "Close", "Low" and "High" in different time rang-
es of FoRex transactions.
Alternative experiments with different support vector machine
models are performed in this paper. These are to show the ad-
vantage of using them in FoRex rate prediction. The detail re-
search on SVM in FoRex market can be seen in [27]. The analysis
of FoRex rate prediction can be seen in section 3. This shows
some specific concepts which contribute to the predicted trend of
FoRex rate in the trading market. These specific terms are the
basic explanations for the classification output results.
To be more generality, a framework of supervised machine learn-
ing (SML) techniques in predicting FoRex rate has been shown in
section 4. Based on this framework, a representative SVM model,
which is proposed from research experiment in [27], is chosen to
install into expert advisor (robotics) in FoRex transaction software
to make the actual FoRex transactions with demo account.
2. Related Works
The Foreign Exchange problems can be seen as the ones of pre-
dicting FoRex rates (up or down) of each currency pairs. There-
fore, they can be seen as classification problems where their out-
puts (FoRex rates) are the binary values. There are some previous
techniques such as Auto Regressive Integrated Moving Average
(ARIMA) used for predicting time series data. According to [6],
ARIMA, however, is an unvaried model in general. Moreover, the
techniques are performed with the assumption of the linear and
stationary time series.
According to [7], the noisy date in the FoRex prediction problems
made them to be challenged issues in time series forecasting field.
Some publications in [7], [8], [6], [9] used RBF, and MLP, and
gene to predict the FoRex rates.
The works, for instance, in [6] used Multilayer Perceptron Net-
work (MLP) with the training set, testing set of 1600 and 225 over
1825 instances. The disadvantage in the work is a use of small
data set. This might cause the quick regression or mis-predicting
output result. Another work of Rumani stock with MLP models
can be seen in [10]. The publication works shown the advantages
in use supervised learning models in forecast a value of rise or fall
for stock price.
There are some ANN and decision tree models which have been
used with forecast the uptrend or downtrend in stock market [11-
14]. Although these works reported about their abilities to work
International Journal of Engineering & Technology
401
with real time data, however, their experiments have reported of
lacking of using high frequency data [15].
The nonlinear regression problems, especially in predicting things,
can be solved with SVMs models. There are some publications
which have been successful in use nonlinear regression predictions
such as in 16], [17] [23], [24], [25]. However, there are limited
publications used SVMs models for financial time-series forecast-
ing [19], [20] although the SVMs models can perform with non-
linear problems and they have been reported as the advantage
techniques. Therefore, continue to the research in [27], the repre-
sentative supervised machine learning method of SVM is selected
to estimate the uptrend or downtrend of each currency rates. It is
because of SVMs can avoid some issues such as over-fitting prob-
lems (like ANN in [1]), or no parameters to be harmony except the
upper bound C [21].
3. Research Background
3.1. Supervised Support Vector Machine
The Support Vector Machines (SVMs) can be divided in to linear
or nonlinear SVMs. These are based on linear or nonlinear hyper-
plane lines which separate the training data set in hyperplane
space. The linear form in linear hyperplane in explained as follow-
ing:
Where and
then
Assume that ,
. is labelled as output
class. Therefore, outputs can be formed as:
The hyperplane H need separate the data onto alternative classes.
So, SVMs have to build two hyperplane as: ;
and with no existing in and , &
margin of and is max.
In some cases, the SVM need to use one special margin so-called
as soft margin to separate the data set into two classes including
some mis-classified training instances [4]. This is because SVM
can not find any separating hyperplane which can separate exactly
the data set.
It seems to be difficult to solve with nonlinear training dataset.
The solution is that the data is mapped onto a higher dimensional
space so-called as feature space in where the linear hyperplane can
separate the data set in to two classes.
By using kernel function, any new points in feature space are rep-
resented for the data in the given data set [5].
Assume that is predicting error rate for each pair of ,
selection of will be produced in two heuristic loops so that
will max.
3.2. Foreign Exchange Rate Prediction
The FoRex market has its own characteristics of high benefit. In
there, the transaction can be performed online via the internet
services. This means that the investors can request their transac-
tions at any time and from anywhere in the world. The buying and
selling can be done with any currencies pairs such as EUR/USD,
EUR/JPY, or GBP/EUR etc. in order to make profit.
A term of PIP (Price Interest Point) is defined in order to measure
the profit or loss in the FoRex. A PIP is the smallest price change
that a given exchange rate can make. Every money pair has a cor-
responding value (a quote). For example, a quote of 1.2776 is the
exchange rate of every EUR is USD 1.2776 (it costs the trader of
1.2776 to buy a single EUR). If the rate is lightly moved from
1.2776 to 1.2777, it shows 1 PIP increase.
The PIP is calculated as:
For example, if a trader purchase 10,000 units with the rate of
EUR/USD of 1.2776. Therefore, PIPvalue= 0.0001/1.2776 x 10,000
= 0.7827 EUR.
The concept of the transaction trend describes the direction of the
market moving as uptrend or downtrend. In the uptrend, a re-
sistance level means there is a change in supply and demand
(more sellers than buyers) whereas in a downtrend, there is a con-
cept of support level, in which there are more buyers than sellers.
When the trend is ended, there are two possible outcomes of oppo-
site starting trend or in range market (stop at certain resistance or
support levels). The task for the Expert Advisor (in the experiment
in later section) is to determine the trend, calculate the time of
lasting and decide if the timing is right to place a good transaction
of Bid/Ask.
Fig.1: Illustration of Uptrend or Downtrend
To predict target of exchange rate as uptrend or downtrend, the
following parameters are defined.
Target Profit (TP) shows the target profit (positive num-
ber).
Stop Loss (SL) indicates the acceptable loss (positive
number).
Time Trend Detection (T) indicates the time starting
from present stage (positive number).
Hence, current rate (P), the trend of rate can be as:
up if the rate of (P+TP) is increased, and there is a de-
creasing of an amount of (P-SL). All these are calculated
in the time T.
Low when the (P-TP) is decreased, and an amount of
P+SL is increased. All these are calculated in the time T.
4. Supervised Machine Learning Prediction
Framework
The supervised machine learning prediction frame work can be
seen in Figure 2, which is based on data mining process.
Fig.2: Supervised machine learning prediction framework
The detail as follows:
402
International Journal of Engineering & Technology
Step 1: Feature Selection. Collected data set might include un-
necessary features for predicting process. Therefore, these features
are removed.
Step 2: Select dataset of currency pair. In this step, a defined pair
of money eg. EUR vs USD is chosen. Also in this step, the data is
taken from alternative candles, time frames etc. in order to be
easily used in installation with Robotics.
Step 3: Processing data with supervised machine learning (SML)
models (here is support vector machine models). Dataset is divid-
ed into training, testing and validation sets. All data in mapped on
the range of [0,1].
Step 4: Evaluation models with assessment metrics of mean
square errors (MSE).
Step 3,4 is repeated many time in order to choose a model with
minimum MSE.
The output is a certain definition of trend (up or down).
5. Experiment
5.1. Data analysis and Model configuration
The experimental data is collected from 01/01/2013 to 30/09/2016
by using MetaTrader 4 [22]. For general use, the pair of
EUR/USD has been chosen. Two data sets are created with the
one of training of D (including the instances in a time of
01/01/2013 to 31/12/2015), and one of testing D’ (including data
in a period 1/01/2016 to 30/09/2016).
The features are chosen basing on the profits. In this paper, the
profit is taken of bout 10 – 15 Pips in each transaction. These se-
lections might effect to the experimental results. The input fea-
tures for samples exchange rates "Open", "Close", "Low" and
"High". These rates are taken in alternative candles in the window
time (time frame). There are 1 minute, 5 minutes, 15 minutes,1
hour, 1 day according to M1, M5, M15, H1, D1 respectively. Oth-
er features are taken from Custom Indicators, Bollinger Bands,
RSI and MA.
To continue with the experiment in [27], 196 features are chosen
(in Table 1). The addition more features (increase from 137 to 196)
for the classification is for expecting to have better results in the
classification process and having better transactions. This means
data is taken from an extension of candlesticks and timeframe.
This is to concentrate more detail in specific historical transaction
data.
Table 1: Feature selections
Fea-
ture#
Data
N
o
Explanations
#1-20
O , H , L , C on M1
20
04 value X 5 nearest
candles
#21-40
O , H , L , C on M5
20
04 value X 5 nearest
candles
#41-56
O , H , L , C on M15
16
04 value X 4 nearest
candles
#57-72
O , H , L , C on H1
16
04 value X 4 nearest
candles
#73-92
O, H, L, C on D1
20
4 values x 5 nearest
candles
#93-97
Time data
5
Time attributes
#98-
105
RSI(7) on M5, M15
8
2 values x 4
Timeframe
#106-
113
RSI(14) on M5, M15
8
2 values x 4
Timeframe
#114-
145
MA(9), MA(12), MA(100),
MA(200)
32
8 values x 4
Timeframe
#146-
165
Custom Indicator
20
8 PAX + 12 MKC
#166-
174
Bollinger Bands
9
3 Lines x 3 Timeframe
#175-
180
Average True Range
6
2 values x 3
Timeframe
#181-
196
Highest, Lowest M1, M5
16
4 values x 4
Timeframe
Model configurations
The SVMs models are chosen in the experiments. The models
parameters are used the same as using in [27]. Therefore, the
model parameters are:
RBF:
with
Polynomial:
with and .
The parameter of has been used with .
Table 2: Experimental Model Configurations
Models
Poly
1
Poly
2
Poly
3
GsRBF
1
GsRBF
2
GSRBF
3
C
1.0
1.0
1.0
2.0
1.0
2.0
Kernel
Poly
Poly
Poly
RBF
RBF
RBF
Parame-
ters
Pow-
er 2
Pow-
er 3
Pow-
er 3
=5.0
=2.0
=2.0
No of
vectors
2500
3000
3972
3000
4500
5000
5.2. Classification results
Data set D data can be separated to two parts and posi-
tive and negative output respectively. Which can be further split to
and
sub-sets with each
sub-sets has
samples. D_Test set is taken with each
sub-set. D_Train includes k-1 remaining samples in the sub-sets.
In here, k is 5.
Following Table 3 & 4 show experimental results with 6 alterna-
tive models. The performance between alternative models.is eval-
uated and compared by using the Accuracy Exactness Rate, Preci-
sion of Positive, Negative, Micro-Average and Macro-Average
individually.
Table 3: Results of RBF Configuration Model
Kernel: Gaussian RBF
Features: 196
GsRBF1
GsRBF2
GSRBF3
D_Trai
n
D_Tes
t
D_Trai
n
D_Tes
t
D_Trai
n
D_Tes
t
Accurate
84.22%
55.80
%
83.14%
58.14
%
83.81%
58.17
%
Precisio-
n
Positive
84.05%
55.59
%
82.59%
57.55
%
83.12%
57.56
%
Negative
84.39%
55.98
%
83.71%
58.70
%
84.52%
58.76
%
Micro-
Average
84.22%
55.80
%
83.14%
58.14
%
83.81%
58.17
%
Macro-
Average
84.22%
55.79
%
83.15%
58.13
%
83.82%
58.16
%
Table 4: Poly Configuration Models Results
Kernel: Polynomial
Features:196
Poly1
Poly2
Poly3
D_Train
D_Test
D_Train
D_Test
D_Train
D_Test
Accurate
86.15%
74.20%
86.15%
74.35%
86.25%
74.32%
Precisio-n
Positive
86.73%
80.29%
86.58%
80.18%
86.80%
80.21%
Negative
85.59%
70.36%
85.73%
70.62%
85.72%
70.56%
Micro-Average
86.15%
74.20%
86.15%
74.35%
86.25%
74.32%
Macro-Average
86.16%
75.32%
86.16%
75.40%
86.26%
75.38%
5.3. Expert Advisor experimental results
The Poly2 SVM model is chosen for installation into Robotics
(Expert Advisor- EA) in MetaTrader 4 [30]. The script (written in
International Journal of Engineering & Technology
403
EA) has performed with a collection of observed data set. The
information incorporates some specialized markers and exchanges
of purchase or offer FoRex match of USD and EUR.
For correlation with the utilization directed machine learning
model in Robotics, the typical exchange (test of Robotics without
utilizing backing of SVM) additionally has been performed.
The results can be seen in Table 5. The bold numbers in the table
show the comparison between the transactions use or not use
SVM.
Table 5: Results Indicators’.
In-
dex
Indicators
No_SVM
SVM
1
Net Profit
328.73
462.5
2
Gross Profits
1238.6
713.93
3
Gross Loss
-
909.87
-
251.44
4
Profit rate
1.36
2.84
5
Growth
32.9%
46.3%
6
Average Growth (in a
month)
3.21%
4.31%
7
Strict Drawdown
155.07
17.49
8
Fair Drawdown
436.2
3
34.05
%
315.6
2
24.31
%
9
Volume
188
94
10
Profit Transactions
139
73.9%
76
80.9%
11
Loss Transactions
49
26.1%
18
19.2%
12
Ask Profit Transactions
86
95.4%
49
100.0
%
13
Bid Profit Transactions
102
55.9%
45
60.0%
5.4. Discussion
Alternative models results with different kernel functions (Gaussi-
an RBF, Polynomial) can be seen in Table 3 and Table 4. There
are big gap results of test set compared to the results of training set
(see in table 3, about 57% and 83.8 % respectively). The ad-
vantage use of SVM in the transaction Robotics can be seen in
Table 5. The indicators’ results show that the net profit of using
SVM in add-on robot is higher than the one without using SVM.
For instances, the number of samples for the script using support
vector machine is half of number of SVM (94 & 188) and its prof-
it rate can achieve of 2.84 compared to 1.36.
Fig 3: Comparison between 196 features-models and 137 features-models
in [27].
The accuracy rates of models using with 137 features in [27] are
used to compare to the ones using with 196 features in this re-
search (see in Figure 3). In general, there are not big gap of rates
between these models. This means the increase of features does
not have better improvement in the accuracy rate.
For more comparison, the fair drawdown values, which shows the
rate of capital reduction, taken from [27] and this research are
represented on the chart (Figure 4).
It is clear from Figure 4 that the Robotics using SVM with 196
features can prevent the capital reduction compared to the normal
Robotics’ transactions. In other words, more features are used in
SVM, the profit loss can be better control.
Fig. 4: Drawdown rate comparison between 137 features model [27] and
196 features model.
6. Conclusion
The support vector machine can help to forecast FoRex trend of up
or down. This is because the outputs of FoRex rate are received as
binary values (uptrend or downtrend). The FoRex problems now
can be seen as classification ones which can be solved by supervised
learning particular with SVMs models. The experiment has shown
the power of SVMs in performing a huge, complex and time series
data in order to help the investors with their transactions decisions.
The experiment has been performed with alternative kernel func-
tions (Gaussian and Polynomial). Elective models with elective
parameters additionally have been per-framed to win the higher
outcomes. The consequences of utilizing Polynomial Kernel
demonstrated the best outcomes contrasted with models utilized
Gaussian when all is said in done.
The picked model to introduce in Robotic exchange has a place with
the best Poly show in above analyses. Its exchange results can ac-
quire a larger number of benefits than the conventional one (without
use of SVM). Therefore, an increase number of features (from 137
to 196) compared to [27] has shown the better profit lost models can
get (reduce the drawdown rate). This means that although there has
been no different in classification accuracy rates (Figure 3), but the
more concentrating on specific data the better controlling of loss we
can get.
Acknowledgement
Many thanks to ThuongMai University and research group for
supporting the research.
References
[1] Witten IH, Frank E, Hall MA (2011), Data mining: practical
machine learning tools and techniques, Morgan Kaufmann
Publishers.
[2] Kotsiantis SB (2007), Supervised Machine Learning: A Review of
Classification Techniques, Informatica: 31, 249-268.
[3] Vapnik V (1995), The Nature of Statistical Learning Theory,
Springer Verlag.
[4] Veropoulos K, Campbell C, Cristianini N (1999), Controlling the
Sensitivity of Support Vector Machines, In Proceedings of the
International Joint Conference on Artificial Intelligence,
(IJCAI99).
[5] Scholkopf C, Burges JC, Smola AJ (1999), Advances in Kernel
Methods, MIT Press.
[6] Baasher Aa, Mohamed WF (2010), FOREX Trend Classification
using Machine Learning Techniques, Arab Academy for Science
and Technology, 41-47.
[7] Deboeck GJ (1994), Trading on the Edge: Neural, Genetic and
Fuzzy Systems for Chaotic Financial Markets, New York Wiley.
[8] Box GEP, Jenkins GM (1970), Time Series Analysis: Forecasting
and Control, Holden- Day, San Francosco, CA.
50.00%
60.00%
70.00%
80.00%
90.00% Accuracy
D_train_137
D_Test_137
D_train_196
D_Test_196
50.67%
34.69% 34.50%
24.31%
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
137_NoSVM 137_SVM 196_NoSVM 196_SVM
Drawdown Comparison
404
International Journal of Engineering & Technology
[9] Punniyamoorthy M, Thoppan JJ (2013), ANN-GA based model for
stock market surveillance, Journal of Financial Crime, vol. 20, no.
1, 52-66.
[10] Nemeş Dm, Butoi A (2013), Data Mining on Romanian Stock
Market Using Neural Networks for Price Prediction, Informatica
Economică, vol. 17, no. 3,125-136.
[11] Huang W, Nakamori Y, Wang SY (2004), Forecasting foreign
exchange rates with artificial neural networks, a review.
International Journal of Information Technology & Decision
Making, Vol 3, No 1, 145-165.
[12] Huang CL, Tsai CY (2010), A hybrid SOFM-SVR with a filter
based feature selection for stock market forecasting, Expert Systems
with Applications, 36 (2, Part 1), 1529 – 1539.
[13] Tsai CF, Hsiao YC (2010), Combining multiple feature selection
methods for stock prediction: Union, intersection, and multi-
intersection approaches, Decision Support Systems, 50(1), 258 –
269.
[14] Shen W, Guo X, Wu C, Wu D (2011), Forecasting stock indices
using radial basis function neural networks optimized by artificial
fish swarm algorithm, Knowledge-Based Systems, 24(3), 378–385.
[15] Reboredo M, Rubio JJ, Nonlinearity in forecasting of high-
frequency stock returns, Computational Economics, 40(3), 245–
264.
[16] Müller KR, Smola A, Rätsch G, Schölkopf B, Kohlmorgen J, and
Vapnik V, (1997), Predicting time series with support vector
machines, In proceedings of the IEEE workshop on neural networks
for signal processing 7, 511 – 519.
[17] Mukherjee S, Osuna E, Giroso F (1997), Nonlinear prediction of
chaotic time series using SVM. In proceedings of the IEEE
workshop on neural networks for signal processing 7, 511 – 519.
[18] Kim K (2003), Financial time series forecasting using Support
Vector Machines, Neurocomputing 55, 307- 319.
[19] Tay EFH, Cao L (2001), Application of support vector machines in
Financial time series forecasting, Omega 29, 309–317.
[20] Liu L, Wang W (2008), Exchange Rates Forecastiing with Least
Squares SVM, International Conference on Computer Science and
Software Engineering, 2008.
[21] Drucker H, Wu D, Vapnik VN, Support vector machines for spam
categorization, IEEE Trans. Neural Networks 10 (5),1048–1054
[22] Metatrader 4, (2016). Website: http://www.metaquotes.net/
[23] Kamruzzaman J, Sarker RA, Ahmad I (2003), SVM Based Models
for Predicting Foreign Currency Exchange Rates, Third IEEE
International Conference on Data Mining, ICDM 2003, 557 – 560.
[24] Sidehabi SW, Indrabayu, Tandungan S (2016), Statistical and
Machine Learning approach in forex prediction based on empirical
data, IEEE International Conference on Computational Intelligence
and Cybernetics, CYBERNETICSCOM 2016, 63 – 68.
[25] Sharma DK, Hota HS, Handa R (2017), Prediction of foreign
exchange rate using regression techniques” Review of Business and
Technology Research, Vol. 14, No. 1, 29-33.
[26] Ramakrishnan SK, Butt S, Chohan MA, Ahmad H (2017),
Forecasting Malaysian exchange rate using machine learning
techniques based on commodities prices, IEEE International
Conference on Research and Innovation in Information Systems,
ICRIIS 2017.
[27] Thuy NTT, Vuong DX (2018), Using Support Vector Machine in
FoRex Predicting, IEEE International Conference on Innovative
research and development, ICIRD 2018.
