The Best Investment Strategy for Gold and Bitcoin Combinations Guochi Zhang, Zihao He School of Automation and Information Engineering, Xi'an University of Technology, Xi'an, Shaanxi, 710048, China Abstract In this paper, we model an optimal trading strategy to improve the return on investment. For the optimal trading strategy model based on supervised learning, two sets of officially provided data are adjusted and supplemented, and the data is visualized to verify that prices and yields have transaction stability. Select 8 eight factors to filter the data. Introduce 5 classic algorithms in supervised machine learning to determine the optimal LGB classification model. For the trading strategy backtesting model, by using a new evaluation indicator for backtesting to compare five machine learning models, the results show that the LGB quantitative trading strategy model has the highest annual rate of return at 32.80307, and a Sharpe ratio of 5.213687, which was finally determined. Finally, robustness and sensitivity analyses are performed on the considered models. http://www.sjisr.org/download/sjisr-4-6-174-181.pdf