An Attention-based LSTM price prediction model based on a single time series information

With the continuous improvement of people's living standards, people's income from work has gradually increased, and their remaining savings have also gradually increased. In order to avoid the problem of currency devaluation and shrinking savings caused by inflation, investment is regarded as an indispensable asset allocation method. Among investment products, gold has always been favored for its store of value; in the past decade, the cryptocurrency Bitcoin has been limited in number due to its unique technology, causing a new wave of investment in Bitcoin. At present, most of the existing forecasting models are for forecasting stock prices, and there are multiple forecasting models. We used grayscale prediction, time series analysis method and LSTM prediction model based on Attention mechanism to predict and analyze the results of two investment products. The experiment found that the LSTM prediction model based on Attention mechanism had the best prediction performance of the two investment products.