Researchers have developed ContestTrade, a novel multi-agent trading system designed to improve the performance of large language model (LLM)-based agents in financial markets. The system employs an internal contest mechanism with two specialized teams: a Data Team that processes market information into textual factors and a Research Team that generates trading decisions. This approach aims to mitigate the sensitivity of LLM agents to noisy market data and context window limitations, achieving superior backtested returns and risk-adjusted performance compared to baseline methods in a post-2024 A-share market simulation. AI
IMPACT This system could enhance the application of LLMs in quantitative finance by improving their ability to process market data and make trading decisions.
RANK_REASON The cluster contains a research paper detailing a novel system for LLM-based trading. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
