In financial markets,gold futures prices are influenced by a variety of factors,and accurate prediction of them holds significant importance.To address this issue,a new model for predicting gold futures prices is proposed that integrates multiple data sources through a LightGBM (Light Gradient Boosting Machine) feature selection method with an LSTM (Long Short-Term Memory) model.First,the paper preprocesses the acquired macroeconomic and technical indicators,then annotates the sentiment tendencies of unstructured news headlines using various methods to construct a weighted sentiment index.Additionally,it aggregates Baidu search indices of multiple keywords into a composite Baidu search index.Second,the LightGBM method is used to rank feature importance for both macroeconomic and technical indicators to extract key features.Finally,the selected features,along with the weighted sentiment index and the composite Baidu search index,serve as input variables for the LSTM forecasting model.Empirical results show that the LightGBM-LSTM model with multi-source data exhibits excellent prediction performance with the smallest prediction errors.Compared with the benchmark model,it achieves more accurate forecasts of the closing prices of gold futures.