Effective forecasting of agricultural market prices is crucial for farmers, traders, and policymakers to make educated decisions and effectively manage risks. Variations in market prices considerably influence agricultural planning, trading strategies, and food security. This research investigates both statistical and deep learning methodologies for forecasting soybean prices in Rajasthan, India, using market data collected over nearly two decades. Well-established models like ARIMA and SARIMA are used to detect historical patterns and seasonal variations, whereas Long Short-Term Memory (LSTM) networks are structured to interpret complex time-dependent relationships. Additionally, hybrid approaches that combine statistical methods with deep learning techniques are developed to improve the accuracy of predictions. The suggested methods seek to increase forecasting reliability and provide valuable insights into market behaviors.