Optimal Strategy: A Comprehensive Model for Predicting Price Trend and Algorithm Optimization

Abstract

In this paper, we study the price changes of bitcoin and gold over time by building an LSTM-based price prediction model and a genetic algorithm-based yield optimization model. Specifically, we build a two-layer LSTM-based time series forecasting model. In order to study how to make decisions on capital allocation and profit maximization, a single-objective optimization model based on a genetic algorithm is established. Furthermore, to prove that our decision model is optimal, we use the error evaluation metrics MSE, MAE, and R2 to perform statistical analysis on the predicted asset value and the actual asset value, where R2 reaches 0.8857. Finally, we perform a sensitivity analysis on transaction commissions.