A Prediction Model for Short Price Jump in Cryptocurrency Market

Abstract

In recent years, the global cryptocurrency market has attracted a diverse range of traders, from seasoned professionals to newcomers, resulting in a highly volatile environment. This volatility presents numerous opportunities for traders to capitalize on rapid price fluctuations. In this context, we introduce a Random Forest model designed to predict whether a coin will experience growth in the next trading candle, using several input features. We used Binance historical daily data from 1 Jan 2018 to 31 Dec 2021 to train our models and evaluated them using different time spans (varied between Jan 2022 to Oct 2023) as testing datasets. Moreover, we also used an over sampled training dataset to enhance the training process. Demonstrating notable precision, especially with a growth rate of 1%, the model has proven effective across various scenarios, consistently yielding profits. To be more specific, regarding the testing datasets of 1 to 31 Oct 2023, 1 Jul 2023 to 30 Sep 2023, and LSK/USDT from 1 Jan 2022 to 31 OCT 2023, using a growth rate of 1%, we achieved 18%, 30%, and 68% profits, respectively. This study underscores the potential for leveraging well-designed machine learning models to achieve significant profits, even in bearish market conditions.