Quant-Arena: Benchmarking LLM Agents for Quantitative Trading Research
None of the nine models beat the equal-weight market in the sealed 2024 test.
GPT-6 Astra scored highest on Quant-Arena, which tests the full research loop on real A-share market data. Models build features and stock-scoring code using feedback from 2022–2023, then freeze one strategy for evaluation on 2024. The evaluator includes trading constraints, costs and market impact estimated from a simulated order book. Scores emphasize realized excess return, with additional measures for signal quality and feature diversity.
ArXiv · AI/CL/LG's note
GPT-6 Astra scored highest on Quant-Arena, which tests the full research loop on real A-share market data. Models build features and stock-scoring code using feedback from 2022–2023, then freeze one strategy for evaluation on 2024. The evaluator includes trading constraints, costs and market impact estimated from a simulated order book. Scores emphasize realized excess return, with additional measures for signal quality and feature diversity.
ArXiv · AI/CL/LG's note
score 4