Fabian Corrêa Cardoso
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Fabian Corrêa Cardoso has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Eduardo Nunes Borges
Federal University of Rio Grande (FURG)
11 shared publications - Giancarlo Lucca
Universidade Federal do Rio Grande (FURG)
11 shared publications - Viviane Leite Dias de Mattos
Federal University of Rio Grande (FURG)
10 shared publications - Bruno Lopes Dalmazo
Universidade Federal do Rio Grande
9 shared publications - Rafael Alceste Berri
Federal University of Rio Grande (FURG)
9 shared publications - Richard F. Pinto
Federal University of Rio Grande (FURG)
6 shared publications
Latest Publications
- 2 July 2026
- Published by Springer Nature
- p. 71-87
LEAP: A Leakage-Free Evolutionary Alpha Pipeline for NLP-Driven DJIA Prediction
- 2 July 2026
- Published by Springer Nature
- p. 71-87
LEAP: A Leakage-Free Evolutionary Alpha Pipeline for NLP-Driven DJIA Prediction
- 1 July 2026
- Published by Springer Nature
- p. 173-184
Adversarial Detection in EEG-Based BCIs: A Comparative Study of Classical and Neuro-Fuzzy Approaches
- 1 July 2026
- Published by Springer Nature
- p. 173-184
Adversarial Detection in EEG-Based BCIs: A Comparative Study of Classical and Neuro-Fuzzy Approaches
- 1 July 2026
- Published by Springer Nature
- p. 315-332
Horizon-Aware Feature Selection and Evolutionary Optimization for Intraday Prediction in Futures Markets
- 1 July 2026
- Published by Springer Nature
- p. 315-332
Horizon-Aware Feature Selection and Evolutionary Optimization for Intraday Prediction in Futures Markets
- Open Access
- 24 March 2026
- Published by University Nove de Julho in Exacta
- Vol. 24, e28045
Redes Neurais de Estado de Eco com otimização por algoritmo genético e correção de eventos para previsão do preço de ações
Apesar das Redes Neurais de Estados de Eco serem bastante utilizadas por exigirem pouco treinamento, possuem a desvantagem da grande quantidade de hiperparâmetros a serem ajustados. Além disso, existe a possibilidade da alteração da métrica dos dados de entrada afetar seu desempenho. Com a intenção de avaliar o efeito desta alteração, objetivou-se comparar as previsões de uma ação da B3 (Brasil, Bolsa, Balcão), denominada VALE3, determinadas com a utilização de dados brutos e de dados transformados por meio de um fator de correção. Para tal, ajustou-se esses hiperparâmetros com algoritmo genético e testou-se uma forma de correção de eventos de mercado que, de certa forma, estabiliza os dados de entrada. Utilizaram-se parâmetros do algoritmo genético com cruzamento de 80%, elitismo de 10%, mutação de 10% e seleção de tour. Os resultados demostraram que as Redes Neurais de Estados de Eco podem fazer previsões de preços de ações com precisão bastante razoável, com um erro médio absoluto inferior a 1 centavo no período de validação e inferior a 2 centavos no período de teste, com dados brutos ou transformados. Também demostraram que o fator de correção, neste caso, melhorou a capacidade do modelo de previsão.
- 24 March 2026
- Published by University Nove de Julho in Exacta
- Vol. 24, e28045
Open AccessRedes Neurais de Estado de Eco com otimização por algoritmo genético e correção de eventos para previsão do preço de ações
Apesar das Redes Neurais de Estados de Eco serem bastante utilizadas por exigirem pouco treinamento, possuem a desvantagem da grande quantidade de hiperparâmetros a serem ajustados. Além disso, existe a possibilidade da alteração da métrica dos dados de entrada afetar seu desempenho. Com a intenção de avaliar o efeito desta alteração, objetivou-se comparar as previsões de uma ação da B3 (Brasil, Bolsa, Balcão), denominada VALE3, determinadas com a utilização de dados brutos e de dados transformados por meio de um fator de correção. Para tal, ajustou-se esses hiperparâmetros com algoritmo genético e testou-se uma forma de correção de eventos de mercado que, de certa forma, estabiliza os dados de entrada. Utilizaram-se parâmetros do algoritmo genético com cruzamento de 80%, elitismo de 10%, mutação de 10% e seleção de tour. Os resultados demostraram que as Redes Neurais de Estados de Eco podem fazer previsões de preços de ações com precisão bastante razoável, com um erro médio absoluto inferior a 1 centavo no período de validação e inferior a 2 centavos no período de teste, com dados brutos ou transformados. Também demostraram que o fator de correção, neste caso, melhorou a capacidade do modelo de previsão.
- Open Access
Machine Learning vs. Randomness: Challenges in Predicting Binary Options Movements
Binary options trading is often marketed as a field where predictive models can generate consistent profits. However, the inherent randomness and stochastic nature of binary options make price movements highly unpredictable, posing significant challenges for any forecasting approach. This study demonstrates that machine learning algorithms struggle to outperform a simple baseline in predicting binary options movements. Using a dataset of EUR/USD currency pairs from 2021 to 2023, we tested multiple models, including Random Forest, Logistic Regression, Gradient Boosting, and k-Nearest Neighbors (kNN), both before and after hyperparameter optimization. Furthermore, several neural network architectures, including Multi-Layer Perceptrons (MLP) and a Long Short-Term Memory (LSTM) network, were evaluated under different training conditions. Despite these exhaustive efforts, none of the models surpassed the ZeroR baseline accuracy, highlighting the inherent randomness of binary options. These findings reinforce the notion that binary options lack predictable patterns, making them unsuitable for machine learning-based forecasting.
Open AccessMachine Learning vs. Randomness: Challenges in Predicting Binary Options Movements
Binary options trading is often marketed as a field where predictive models can generate consistent profits. However, the inherent randomness and stochastic nature of binary options make price movements highly unpredictable, posing significant challenges for any forecasting approach. This study demonstrates that machine learning algorithms struggle to outperform a simple baseline in predicting binary options movements. Using a dataset of EUR/USD currency pairs from 2021 to 2023, we tested multiple models, including Random Forest, Logistic Regression, Gradient Boosting, and k-Nearest Neighbors (kNN), both before and after hyperparameter optimization. Furthermore, several neural network architectures, including Multi-Layer Perceptrons (MLP) and a Long Short-Term Memory (LSTM) network, were evaluated under different training conditions. Despite these exhaustive efforts, none of the models surpassed the ZeroR baseline accuracy, highlighting the inherent randomness of binary options. These findings reinforce the notion that binary options lack predictable patterns, making them unsuitable for machine learning-based forecasting.
- 6 November 2025
- Published by Springer Nature
- p. 471-481
Machine Learning vs. Randomness: Challenges in Predicting Binary Options Movements
- 6 November 2025
- Published by Springer Nature
- p. 471-481
Machine Learning vs. Randomness: Challenges in Predicting Binary Options Movements
- 26 March 2024
- Published by Elsevier in Knowledge-Based Systems
- Vol. 293
Echo state network and classical statistical techniques for time series forecasting: A review
- 26 March 2024
- Published by Elsevier in Knowledge-Based Systems
- Vol. 293
Echo state network and classical statistical techniques for time series forecasting: A review
- Open Access
- 1 April 2023
- Published by University Nove de Julho in Exacta
- Vol. 23 (1) , 134
Normality tests: a study of residuals obtained on time series tendency modeling
The normality analysis in the distribution of residuals is a determining criterion to verify and validate a model. For example, in modeling financial time series by linear regression, the residuals should be independent of each other, identically distributed, have a normal distribution, and be homoscedastic. Thus, it was aimed to study the performance of some normality tests applied on the residuals obtained from linear regression modeling of time series tendency using polynomials of different degrees. The Jarque-Bera, Anderson-Darling, Kolmogorov-Smirnov-Lilliefors, Doornik-Hansen, and Shapiro-Wilk tests were used, there was agreement in almost all test results, with the exception of the Doornik-Hansen test.
- 1 April 2023
- Published by University Nove de Julho in Exacta
- Vol. 23 (1) , 134
Open AccessNormality tests: a study of residuals obtained on time series tendency modeling
The normality analysis in the distribution of residuals is a determining criterion to verify and validate a model. For example, in modeling financial time series by linear regression, the residuals should be independent of each other, identically distributed, have a normal distribution, and be homoscedastic. Thus, it was aimed to study the performance of some normality tests applied on the residuals obtained from linear regression modeling of time series tendency using polynomials of different degrees. The Jarque-Bera, Anderson-Darling, Kolmogorov-Smirnov-Lilliefors, Doornik-Hansen, and Shapiro-Wilk tests were used, there was agreement in almost all test results, with the exception of the Doornik-Hansen test.