Research on Operation Strategy Based on Grey Model, Neural Network and Dynamic Programming

In recent years, virtual currency has been presented to market traders with a new attitude and has become an indispensable part of the financial market. In the emerging international market, market traders are eager to maximize the benefits. However, due to the complexity of financial market, they urgently need to establish a model about cash, gold, and bitcoins (CGB Model) to guide them how to buy, hold or sell their portfolio. First and foremost, we predict the profit and loss of the next trading day through grey prediction and neural network, with standardized errors concentrated at 0.0. Furthermore, we build an investment decision model based on dynamic programming, and predict that the initial $1000 investment on September 10, 2021 will be worth $23,225. Finally, we use the Morris method for sensitivity test, and conclude that change of the commission rate of Bitcoin has little change in the choice of Xn and Yn, while the change in the commission rate of gold affects the choice of Xn and Yn to a certain extent. Besides, and have a greater impact on Xn, but less on Yn. Our model has an appropriate sensitivity to transaction costs, and can make a strategy change even when the commission suddenly changes under the national macro adjustment, so as to maximize the profit.