Explaining the S&P 500: How does certain commodities affect the index

• Autor(a) principal: Varela, Mailson Manuel Teixeira
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10071/29669
• topic: Machine learning; Financial market; Explainability; SHAP; LIME; Inteligência artificial -- Artificial intelligence; Mercado financeiro; Explicabilidade

In recent years, the proliferation of Artificial Intelligence (A.I.) has revolutionized decision-making processes across various domains. Deep Learning algorithms, particularly LSTM and XGBoost models, have emerged as powerful tools for accurate predictions in complex contexts, such as financial markets. However, the inherent challenge of interpreting these models has led to a tradeoff between accuracy and transparency. The need for Explainable Artificial Intelligence (XAI) becomes paramount in critical domains like finance, where understanding the model's reasoning is crucial for informed decision-making. Our study comprises two fundamental phases: model development and explanation. The initial phase focuses on crafting LSTM and XGBoost models, fine-tuning their hyperparameters, and optimizing their predictive performance for S&P 500 index forecasting. Rigorous evaluation metrics, encompassing MAE, MSE, MAPE and RMSE, guide our pursuit of accurate predictions. Dow Jones emerges as one of the most influential variables in forecasting S&P 500 along with Bitcoin, which, interestingly, wields a consistently negative impact in both models, penalizing both performance with its effects and unveiling its unique role. Our findings inform decision-making in finance, advocating for transparency and advancing predictive models and interpretability.