Method Based on the Support Vec

  • Feb 05 2023
  • by
  • Analyst AZA
Method Based on the Support Vec

Application of the Moving Average (MA) Method Based on the Support Vec

  1. Introduction

Foreign exchange (forex) exchange or forex (Foreign Exchange) is an investment that trades one country's currency with another country's currency, where the currency of one country is traded with other countries for 24 hours continuously starting from Monday at 04.00 WIB in the morning to Saturday at 04.00 WIB/GMT+7. The goal is to get a profit (profit) from differences in currency values [1]. Algorithms or methods that are often used to predict stocks or forex such as algorithmArtificial Neural Networks(ANN) [2] which is one of the most commonly used methods in processing non-linear forecasting, and has a strong parallel process and its ability to handle fault tolerance, however, the practicality of ANN is limited due to several weaknesses such as requiring a large number of training datasets, " overfitting”, the convergence speed is slow and weak in the

optimal local extreme [3]. The Relevance Vector Machine RVM [4] algorithm is a probabilistic model similar to support vector machines ( SVM ) but the training data takes place in a bayesian framework and the output is a predictive distribution of estimated points.AlgorithmSupport Vector Machine(SVM) performance is very good for time series prediction but is limited by the manual selection of basic function parameters [5].AlgorithmSupport Vector Machine (SVM) [5] [2] is a promising method for time series prediction because it uses a risk function consisting of empirical error and routine terms derived from the principle risk minimization structure Kernel type [5] This method is based on statistical learning theory that can solve The 'over-fitting' problem [6] can

also, be used for optimal global solutions and low convergence rates and has a very good ability to generalize small samples [3] and is very good at predicting because this method can minimize misclassification and data deviation in data training [6] but the practicality of SVM is difficult to choose the appropriate parameters. [4] After several years of development, the SVM algorithm has been

successfully applied in several fields, such as pattern recognition and function regression,[3]. Some researchers have started to apply the SVM algorithm to predict time series data or predict stock price indexes, including predicting the GBP/USD exchange rate.

MethodSimple Moving Averages (SMA) [7] is one type of prediction method based on time series or so-called time series data. The SMA method uses values in the past to be used

as a reference in making predictions in the future. Moving Average MethodIt has three different variants vizSimple Moving Averages, Weighted Moving Averages AndExponential Moving Averages.

Each is a method moving Averages, it's just that the way to average it is different from one another. The main difference lies in the weighting of the data which appears frequently. The Simple Moving Average uses the same weighting for each data meanwhileWeighted Moving AverageAndExponential Moving Averagesadding more weight to the data

that appears frequently [8]. The focus of the problem in this study is the movement of the curve in forex trading. The goal is to compare curve patterns online forex trading by using an algorithm support vector machine (SVM) to form the first trend curve, then the results of the SVM process are processed using the moving

average method to form the second curve [8]. The dataset used in this study is the pound sterling vs. us dollar (GBPUSD-1H) [3] on

2. Related Research

Related studies regarding predictions use the moving average method and algorithms support vector machine (SVM) as in the research of Ding-zhou cao et al [3] using a support vector

machine (SVM) algorithm. The results show that the trend of the predicted value curve is identical to the actual value curve and can be used to forecast time series data. Novian Anggis Suwastika [9] uses the moving average method to predict the stock curve. The results obtained are good enough for a small SMA period but are not accurate for determining definitively the value of the stock price.

Kyoung-jae Kim [10] uses the Support factor machine algorithm for the prediction of times series. The experimental results show that SVM provides a promising alternative for stock market predictions, and Nugroho Dwi S [6] uses a support vector machine (SVM) algorithm to predict gold prices. Test results by measuring method Support Vector Machine using RMSE, it is known that variable A produces an RMSE value of 4.695 and variable B has an RMSE value of 4.620. With these results, the RMSE that has been obtained, then variable B (open, high, low, close and factory news) can improve the prediction results.

3. Method

AlgorithmSupport Vector Machine (SVM) was developed by Boser, Guyon, Vapnik, [6] [3] first

presented in 1992 at the Annual Workshop on Computational Learning Theory [11]. Support vector machine (SVM) algorithm is a relatively new technique for making predictions, both in the case of classification and regression, which is very popular recently [12] by always achieving the same solution for each run and trying to find-separator function (classifier) is optimal and able to separate two data sets from two different classes, its performance is convincing in predicting the class of a new data. The SVM algorithm is in principle a linear classifier. But SVM excels in classification for nonlinear problems [12] for non-linear problems, first, the data is projected into a new vector space, (feature space) higher dimensional so that the data can be separated linearly. Next in the new space, SVMsearcheshyperplaneoptimal, to overcome the problem of nonlinearity (nonlinearity). The kernel method is used to provide an alternative approach by mapping data from input space to feature space through a functionϕ so thatϕ:x _→ ϕ(x), for that a point in the input space to beϕ(x) in the feature space. The kernel type used in the SVM algorithm for this study is the Anova kernel type [13]with the equation:

( . ) = ∑ exp (− (− )2) =1

4. Conclusion

The conclusion in this study from the results and discussion it can be concluded that the trend curve produced by the algorithm support Vector Machineexactly the same as the online trans forex curve pattern, but the curve produced by the Moving Average Method is not too close to the curve pattern produced by online forex trading. In Figure 8 the curve (red color) is generated by the method Moving Averagesnot too close to the curve generated by online forex trading.


Related studies regarding predictions use the moving average method and algorithms support vector machine (SVM)


imgaza youtube

Profitable trading recommendations, forex analytics for beginners traders

Subscribe our channel

top authors

Fundamental analyst

Stan Zabar
Fundamental analyst

Head of Analysis Department

Michael Wallenberg
Head of Analysis Department

Economic Observer

Alan Dofine
Economic Observer

Call US Feedback
en de nl fr pt es it uk zh ko ja ar ru pl tr