Application of the LSTM-Xgboost Combined Model and Bootstrap Method in Gold Price Interval Forecasting

With the increasing uncertainty in financial markets, gold price forecasting has become an important topic of research. Traditional time series models face limitations when dealing with nonlinear relationships and large-scale data, often unable to effectively capture market volatility and uncertainty. To address this issue, this paper proposes a weighted ensemble prediction model based on Long ShortTerm Memory (LSTM) and Extreme Gradient Boosting (XGBoost), combined with the Bootstrap method for interval forecasting. Experimental results show that the proposed hybrid model achieves a Mean Absolute Error (MAE) of 18.57 and an R2 of 0.92 on the test set, significantly outperforming traditional standalone models. Additionally, the Bootstrap method provides dynamic prediction intervals for the LSTMXGBoost model, with high coverage and narrow widths at high confidence levels, offering a new approach to effectively manage gold price forecasting and uncertainty analysis.