Track: Undergraduate Student Paper Competition

Abstract

The type of currency market or capital market investment that we know is commonly called forex or foreign exchange. Forex is one of the investments that gets a lot of attention today. The methods used in this research are Neural Network Backpropagation and Exponential Moving Average. Backpropagation Neural Network (BPNN) is an Artificial Neural Network that moves forward and has no repetition where the signal is from the input neuron to the output neuron with a 3-9-1 architecture where 3 inputs, 9 hidden layers and 1 output, while Exponential Moving Average (EMA) is a type of Moving Average that adds weighting in close price movements with the EMA range used is 50. Dataset collection through the website www.tradermade.com to retrieve API and then to retrieve offline data from the site www.eatradingacademy.com with the period January 1, 2007 to June 30, 2022 for the 1-day timeframe and January 3, 2021 to June 30, 2022 for the 1-hour timeframe. Then the data processing is carried out. The data will be divided into 80% training data and 20% testing data which will be used in the data processing process. The purpose of this research is to predict foreign exchange rates in the foreign exchange business using the BPNN and EMA methods to predict future prices. The predicted currencies use 5 currencies with the highest volume transactions such as EUR/USD, USD/JPY, GBP/USD, USD/CAD and AUD/USD. Accuracy used for testing using MAPE with an accuracy value of 1.3338 for EUR/USD, 1.6160 for USD/JPY, 1.3118 for GBP/USD, 1.2997 for USD/CAD and 3.2058 for AUD/USD. The average of the MAPE values is 1.75342.