LSTM network based agent for cryptocurrency trading

Abstract

With the growth of the internet and increasing interest in cryptocurrencies, multiple online trading platforms have been created, giving investors a possibility to trade cryptocurrencies from the comfort of their homes. The aim of this study is to test, whether delegating trading to artificial intelligence can be profitable. As a result of this research script consisting of two parts was written. The first part is responsible for data preprocessing and model training. The second one is the automation algorithm, which output is based on model prediction from the previous part. The output of the automation algorithm is a decision when to buy and sell Bitcoin to make a profit. All algorithm decisions are logged to analyze profit in the given cycles.