基于BPNN-GARCH模型的互联网货币基金收益率预测

Forecasting the Rate of Return of Internet Monetary Fund Based on BPNN-GARCH Model

摘要: 在传统时间序列方法的基础上，引入非线性的BP神经网络模型，建立广义自回归条件异方差模型与BP神经网络模型相结合的组合模型对互联货币基金的收益率进行预测。以平均绝对误差(MAE)、均方误差(MSE)、平均误差(ME)、定向精度(DA)四个指标为检验标准对三个模型进行预测精度的比较。经实证分析，BPNN-GARCH组合模型对于互联网货币基金收益率的预测具有更高的准确性。

Abstract: On the basis of the traditional time series method, a nonlinear BP neural network model is intro-duced, and a combined model of generalized autoregressive conditional heteroskedasticity model and BP neural network model is established to predict the yield of the Internet monetary fund. The prediction accuracy of three models was compared with four indexes of MAE, MSE, ME and DA. Based on the empirical analysis, the BPNN-GARCH combined model has higher accuracy for the forecast of the Internet monetary fund yield.