RDBU| Repositório Digital da Biblioteca da Unisinos
Cryptocurrency trading bot assisted by artificial intelligence
JavaScript is disabled for your browser. Some features of this site may not work without it.
| Autor | Souza, Eduardo Remor de; |
| Orientador | Costa, Cristiano André da; |
| Lattes do orientador | http://lattes.cnpq.br/9637121030877187; |
| Instituição | Universidade do Vale do Rio dos Sinos; |
| Título | Cryptocurrency trading bot assisted by artificial intelligence; |
| Resumo | Cryptocurrencies are getting more popular day after day as an alternative investment option. With this hype, multiple exchanges are available and it is easy to set up an account and start trading in the crypto market. However, trading itself is not an easy task and depends on technical knowledge and emotional control if a person wants to be a profitable trader over time. Most people do not take time to study and get ready for the market and, therefore, end up with losses in a short period. An automated algorithm - a trading bot - following a good trading strategy could take the technical and emotional requirements aside and potentially be profitable, without needing expertise from the user. The market is dynamic and has its different moments, unlike a trading bot, which follows a fixed strategy and is tied to the market’s movements. Artificial intelligence using a convolutional neural network (CNN) architecture could assist the trading bot by analyzing the macro scenario in a set of cryptocurrencies and indicating which ones are in a suitable moment for the trading strategy. The project’s evaluation was done by comparing the financial returns of a trading bot set up with a strategy based on Keltner Channels and RSI with the trading bot assisted by a CNN-based model. The results were interesting as the CNN model did not perform well enough to improve the result from the trading bot, but the labeling method proved to be efficient. The trading bot without assistance had an average return of 14.43%. With AI’s assistance, the average returns dropped to 13.28%. Considering the benchmark AI labels, the average return increased to 24.58%.; |
| Palavras-chave | Cryptocurrencies; Trading; Artificial Intelligence; Convolutional Neural Networks; |
| Tipo | TCC; |
| Data de defesa | 2022-11-23; |
| URI | http://repositorio.jesuita.org.br/handle/UNISINOS/13275; |
| Nivel | Graduação; |
| Curso | Sistemas de Informação; |
Arquivos deste item
Este item aparece na(s) seguinte(s) coleção(s)
-
Sistemas de Informação [21]
Sistemas de Informação