Cryptocurrency price forecasting based on shortterm trend KNN model

Abstract

Since the appearance of bitcoin, the cryptocurrency market has gradually prospered. The price of cryptocurrency often changes rapidly and fluctuates greatly, which makes many investors rush into the market, ignoring the risk behind it. The price prediction of cryptocurrency can not only provide a reference for investors' investment but also reveal its financial laws to a certain extent and improve the risk early warning mechanism as well as improve its stability and security. In this paper, the KNN algorithm is used to forecast the cryptocurrency price, and the KNN model is improved based on the short-term price trend before the trading day. The experimental results show that the improved KNN model has a more accurate prediction result than the logistic regression model and traditional KNN model.