Science AI
arXiv preprint presents live hybrid LLM trading agent that led FinMMEval 2026 TSLA leaderboard
An arXiv preprint describes Fin-Analyst, a hybrid trading agent that ranked first on the final FinMMEval 2026 Task 3 leaderboard for Tesla (TSLA) with a +13.51% return and a Sharpe ratio of 4.10 over a live evaluation window accessed on 2026-07-05. The agent combines an eight-specialist LLM pipeline, covering news, SEC filings, fundamentals, analyst forecasts, technical indicators, and social sentiment, aggregated by a meta-agent for TSLA, while a lightweight rule-based three-signal vote handles Bitcoin (BTC). The BTC vote finished flat but outperformed a sharply falling baseline. Ablation identified event-driven 8-K disclosures as the most influential TSLA signal. Error analysis found that memoryless LLM specialists repeated incorrect calls for days, and that fixed-threshold BTC rules lost money trading noise in a sideways market while the LLM pipeline gained under similar conditions, motivating a memory-aware, LLM-based successor for both assets.
Sources
In this story
Published by Tech & Business, a media brand covering technology and business.
This story was sourced from cs.AI updates on arXiv.org and reviewed by the T&B editorial agent team.

