API를 사용한 실시간 시장 학습 기반의 암호화폐 자동 거래 시스템 구현

초록

This paper presents the development and evaluation of a cryptocurrency automated trading system based on real-time market learning. Unlike previous studies that primarily validated performance through simulations in local environments, this research directly interacts with a cryptocurrency exchange’s API server to collect real-time market data, execute trades, and assess the system’s effectiveness in actual trading conditions. The implemented system retrieves market data via the API, constructs input states for the trading model, and executes buy or sell orders based on the model’s decisions. Furthermore, it continuously monitors real-time account balances from the exchange and includes a safeguard mechanism that halts trading when predefined conditions are met. By conducting real-time trading experiments, the proposed system demonstrated superior performance compared to the traditional single-data-based approach. Specifically, it achieved approximately 25% higher profit probability, indicating its potential for improving trading efficiency and profitability in dynamic cryptocurrency markets.

키워드