Deep Learning Algorithm-Based Financial Prediction Models

In this paper, a new FEPA portfolio forecasting model is based on the EMD decomposition method. The special empirical modal of financial time series, principal component analysis, and artificial neural network to forecast for nonlinear, nonstationary, multiscale complex series predict stock market indices foreign exchange rates empirically investigate hot area in research. combined proposed paper idea decomposition-reconstruction synthesis, which effectively improves model’s prediction internal series. we select CSI 300 Index rate as data establish seven models make predictions about short-term running trend closing price. interval algorithm introduced considering both high low prices be contained input output. By analyzing price, index at same time, volatility its can better captured.