In Cambodia, paddy rice is a staple food and a key income source, with Battambang Province being one of the leading rice-producing regions. Accurate paddy rice price forecasts are crucial for farmers, traders, policymakers, and financial institutions to make informed decisions, mitigate risks, and stabilize the market. Several factors influence price fluctuations, including domestic production, global market trends, input costs, and macroeconomic variables like crude oil prices and exchange rates. Developing forecasting models that capture these dynamics is crucial for agricultural sustainability and economic resilience. This study compares traditional time series models (ARIMAX) with machine learning models (XGBoost) and deep learning models (LSTM) to evaluate forecasting accuracy. The results show that XGBoost outperforms both ARIMAX and LSTM, achieving the lowest RMSE (29.90 KHR/Kg), MAE (25.47 KHR/Kg), and MAPE (1.70%). In contrast, ARIMAX performs moderately, while LSTM exhibits higher error rates. This study provides stakeholders with a data-driven approach to agricultural decision-making, supports sustainable farming practices, and enhances market predictability, thereby contributing to economic stability in Cambodia's rice industry. The findings also inform policy recommendations to improve trade efficiency and food security.