Cardano cryptocurrency price from Twitter. A prediction algorithm from machine learning

• Autores: Piccarreta Acosta, Riccardo, Zavala Arana, Alejandra
• Formato: tesis de grado
• Fecha de Publicación: 2024
• Institución: Universidad de Lima
• Repositorio: ULIMA-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorio.ulima.edu.pe:20.500.12724/21708
• Enlace del recurso: https://hdl.handle.net/20.500.12724/21708
• Nivel de acceso: acceso abierto
• Materia: Algoritmos; Aprendizaje automático; Criptomonedas; Twitter; https://purl.org/pe-repo/ocde/ford#2.11.04

Cryptocurrencies are a growing market that has attracted the attention of many investors in recent years. While cryptocurrencies offer a secure and decentralized form of payment, this market is highly volatile. Factors influencing price changes include the balance of supply and demand, its utility, trading indicators, and market confidence. The present research aims to predict the price of the Cardano cryptocurrency by using machine learning techniques, specifically SVM, LSTM and BiLSTM models. In addition to accounting for financial indices, Twitter activity was used as a data source to measure market sentiment. The study analyzes various predictive horizons, including time ranges of 1 day, seven days, 14 days, 21 days and 30 days. The results obtained were validated with different performance indicators, and it was determined that the model predicts Cardano prices one month ahead with a MAPE of less than 22%, providing valuable information for investors interested in the volatile Cardano cryptocurrency market.