Globally, category asset allocation as the core investment method has been the consensus of the industry. Macroeconomic index data and market data of various asset indexes reflect different economic states of each economic cycle and can calculate the risk and return characteristics of a certain type of asset, and construct passive investment portfolio. However, for China’s investors and investment environment, we need a more suitable macroeconomic and broad asset category forecast model. In order to find high-frequency and effective macroeconomic indicators, we divided China’s macroeconomic operation status from 2001 to 2021 into different economic states. In this paper, macroeconomic environment such as economic growth, inflation and interest rate of China in the next five years is simulated through macroeconomic models or other mathematical models. According to the above calculated GDP growth rate, the consumer price index (CPI) is used to calculate the inflation rate of money and commodities, and the monthly interest rate of demand deposit is calculated to simulate and forecast the macroeconomic development situation of China in the next five years. LSTM algorithm is used for sliding window forecast of economic growth, inflation and interest rates in the next five years. The risk-return characteristics, including expected return, standard deviation of return and Sharpe ratio, of major asset indexes in the data set under various economic conditions were calculated by using the four economic conditions divided by the Merrill Lynch clock frame. The indexes listed in the data set were arranged and combined, and the risk-return characteristics of the recent five years were selected for prediction, so as to obtain the investment portfolio with relatively suitable recovery stage and high return in the next five years.