Abstract
This study investigates the effectiveness of a novel ConvLSTM network in predicting option prices within financial markets. Option pricing remains a pivotal aspect of financial decision-making, yet traditional models often struggle to capture the intricate dynamics inherent in market behavior. Leveraging the capabilities of deep learning, particularly the fusion of CNN and LSTM architectures, the proposed ConvLSTM model aims to address these limitations by effectively learning spatial and temporal dependencies within financial data. Through empirical evaluation utilizing data collected from the US stock markets spanning from April 2018 to June 2020, the ConvLSTM model demonstrates remarkable predictive accuracy, achieving a precision rate of approximately 92% over 200 epochs. Furthermore, visual representations corroborate the model's efficacy, showcasing its ability to closely align predicted option prices with observed values. This research contributes to the ongoing discourse on deep learning applications in financial forecasting, offering insights into the potential of ConvLSTM networks to enhance risk management strategies and inform investment decisions in dynamic and volatile market environments.