JZUS - Journal of Zhejiang University SCIENCE

Frontiers of Information Technology & Electronic Engineering

Accepted manuscript available online (unedited version)

FinSphere: a real-time stock analysis agent with instruction-tuned LLMs and domain-specific tool integration

Author(s): Shijie HAN^1, ³, Jingshu ZHANG^1, ³, Yiqing SHEN^3, ⁴, Kaiyuan YAN^3, , Hongguang LI^3,
Affiliation(s): ¹Department of Industrial Engineering and Operations Research, Columbia University, New York 10027, USA; more
Corresponding email(s): sh4460@columbia.edu, zhangjingshu@mail.shufe.edu.cn, yshen92@jhu.edu, yankaiyuani@163.com, harvey2@mail.ustc.edu.cn
Key Words: Instruction-tuned financial LLM, Real-time stock analysis, Evaluation framework & dataset

Share this article to： More <<< Previous Paper \|Next Paper >>>

Shijie HAN^1,³, Jingshu ZHANG^1,³, Yiqing SHEN^3,⁴, Kaiyuan YAN^3,, Hongguang LI^3,. FinSphere: a real-time stock analysis agent with instruction-tuned LLMs and domain-specific tool integration[J]. Frontiers of Information Technology & Electronic Engineering,in press.https://doi.org/10.1631/FITEE.2500414

@article{title="FinSphere: a real-time stock analysis agent with instruction-tuned LLMs and domain-specific tool integration",
author="Shijie HAN^1,³, Jingshu ZHANG^1,³, Yiqing SHEN^3,⁴, Kaiyuan YAN^3,, Hongguang LI^3,",
journal="Frontiers of Information Technology & Electronic Engineering",
year="in press",
publisher="Zhejiang University Press & Springer",
doi="https://doi.org/10.1631/FITEE.2500414"
}

%0 Journal Article
%T FinSphere: a real-time stock analysis agent with instruction-tuned LLMs and domain-specific tool integration
%A Shijie HAN^1
%A³
%A Jingshu ZHANG^1
%A³
%A Yiqing SHEN^3
%A⁴
%A Kaiyuan YAN^3
%A
%A Hongguang LI^3
%A
%J Frontiers of Information Technology & Electronic Engineering
%P
%@ 2095-9184
%D in press
%I Zhejiang University Press & Springer
doi="https://doi.org/10.1631/FITEE.2500414"

TY - JOUR
T1 - FinSphere: a real-time stock analysis agent with instruction-tuned LLMs and domain-specific tool integration
A1 - Shijie HAN^{1
A1 -}³
A1 - Jingshu ZHANG^{1
A1 -}³
A1 - Yiqing SHEN^{3
A1 -}⁴
A1 - Kaiyuan YAN^{3
A1 -}
A1 - Hongguang LI^{3
A1 -}
J0 - Frontiers of Information Technology & Electronic Engineering
SP -
EP -
%@ 2095-9184
Y1 - in press
PB - Zhejiang University Press & Springer
ER -
doi="https://doi.org/10.1631/FITEE.2500414"

Abstract
Chinese Summary
Academic Network
Reviewer Comment

Abstract: Current financial large language models (FinLLMs) exhibit two major limitations: the absence of standardized evaluation metrics for stock analysis quality and insufficient analytical depth. We address these limitations with two contributions. First, we introduce AnalyScore, a systematic framework for evaluating the quality of stock analysis. Second, we construct Stocksis, an expert-curated dataset designed to enhance LLMs'financial analysis capabilities. Building on Stocksis, together with a novel integration framework and quantitative tools, we develop FinSphere, an AI agent that generates professional-grade stock analysis reports. Evaluations with AnalyScore show that FinSphere consistently surpasses general-purpose LLMs, domain-specific FinLLMs, and existing agent-based systems, even when the latter are enhanced with real-time data access and few-shot guidance. The findings highlight FinSphere's significant advantages in analytical quality and real-world applicability.

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference

Open peer comments: Debate/Discuss/Question/Opinion

<1>

- Go to

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference