<![CDATA[Advances in computing technology, artificial intelligence, and graphics accelerators have driven significant growth in the value of NVIDIA Corporation (NVDA) shares over the past decade. The high volatility and non-linear patterns in NVDA's stock price require the use of forecasting methods that can capture long-term dependencies in time-series data. This study proposes a Long Short-Term Memory (LSTM) model to predict NVDA's stock price using historical daily data up to 2025. The training process was carried out using a two-layer LSTM architecture, Adam optimizers, and a 60-day time window. The evaluation results show that the model is capable of achieving a Mean Absolute Percentage Error (MAPE) of 4%, which indicates a high level of accuracy in predicting price movement patterns. Visualization of the prediction results confirms that the model successfully follows actual trends consistently, with relatively small deviations at extreme fluctuation points. Additionally, the model was used to project NVDA stock prices ten steps ahead, showing that price movement trends remain consistent with historical patterns. These findings contribute methodologically to studies on high-risk stock forecasting through deep learning approaches and provide a basis for data-driven decision-making in the modern financial sector.]]>
Copyrights © 2026
