A carregar...
Publicação
LSTM-Based Market Forecasting: Hyperparameter Optimization Using Genetic and Differential Evolution Algorithms
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 14.37 MB | Adobe PDF |
Orientador(es)
Resumo(s)
This dissertation presents a forecasting framework for financial time series based on Long Short-Term Memory (LSTM) neural networks. The model is trained to perform one-day-ahead predictions on the S&P 500 Index closing price, capturing temporal dependencies in asset price dynamics. To improve model calibration, a hybrid optimisation procedure combining Genetic and Differential Evolution Algorithms is applied to the LSTM hyperparameters. The model predictions are converted into proportional trading signals using activation functions, which adjust signal intensity according to the magnitude of predictive error. A statistical residual filter removes predictions with high deviations. The activation-based allocation strategies yield financial performance that exceeds the Buy & Hold benchmark in annualised return, drawdown control, and Sharpe ratio. The proposed framework offers an alternative to fixed-threshold trading systems and supports broader applicability across forecasting models and asset classes.
Descrição
Dissertation presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data Science
Palavras-chave
Long Short-Term Memory Genetic Algorithms Differential Evolution Algorithm Financial Market Prediction Proportional Capital Allocation SDG 8 - Decent work and economic growth SDG 9 - Industry, innovation and infrastructure SDG 12 - Responsible production and consumption