TECHNICAL ANALYSIS VERSUS NEURAL NETWORKS TO BITCOIN PRICE PREDICTION

Authors

DOI:

https://doi.org/10.5212/Admpg.v.12.21111.006

Abstract

In 2007, Bitcoin (BTC) was introduced as the world's first decentralized digital currency. BTC has shown a growth in its market value, has been used in speculative investments, and therefore it would be possible to use the same techniques already applied in the stock market. Some strategies that obtained a positive result were identified, including in moments of decline such as the period between July 2019 and June 2020. In this scenario, this study aims to analyze the application of a neural network to predict the price movement of Bitcoin in counterpoint and use of technical analysis. The first model built consisted of an agent that verified the signal emitted by certain technical indicators, and simulated the purchase or sale of BTCs. These experiments obtained good results, and also provided inputs for the second part of the work: a neural network built to make decisions based on technical indicators tested in the previous stage. The results of the neural networks were more positive than the results of each of the agents of technical indicators, obtaining up to 80% increase in the amount in a period when BTC lost 15% of its market value.

Published

2022-11-03

How to Cite

IGARASHI, W. .; MISAEL MACHADO, H. .; CORRÊA IGARASHI, D. C. . TECHNICAL ANALYSIS VERSUS NEURAL NETWORKS TO BITCOIN PRICE PREDICTION. ADMPG Journal, [S. l.], v. 12, n. 1, 2022. DOI: 10.5212/Admpg.v.12.21111.006. Disponível em: https://revistas.uepg.br/index.php/admpg/article/view/21111. Acesso em: 22 nov. 2024.

Most read articles by the same author(s)