Agricultural markets exhibit significant price volatility, creating financial uncertainty for farmers who lack real-time market intelligence. Traditional systems provide historical price data without predictive capabilities or location- specific market recommendations. This paper presents a Machine learning-based decision support system that forecasts crop prices across multiple mandis while recommending optimal selling venues based on predicted revenue, distance, and market conditions. The system integrates Convolutional Neural Networks with Long Short-Term Memory networks to capture spatial and temporal price patterns. A hybrid CNN-LSTM architecture processes historical price data, weather parameters, and demand indicators from government databases. Experimental validation demonstrates improved prediction accuracy over conventional ARIMA and single-model LSTM approaches. The web-based interface enables farmers to access forecasts, compare markets, and receive automated alerts regarding favorable selling opportunities. This system addresses critical gaps in agricultural market transparency by combining predictive analytics with actionable market recommendations