# 🤖 VQuant: An Agentic Large Language Model Architecture for Automated Quantitative Financial Analysis and Document Generation

UQO Working Paper No. 13 — Reasoning-Augmented Tool Orchestration, In-Loop Numerical Computation, and Multi-Format Reporting

TL;DR — VQuant is a production web platform that turns a natural-language request into a complete, reproducible quantitative financial report. A reasoning-augmented LLM acts as planner and synthesizer; all numerical work is delegated to a deterministic execution substrate: an agentic orchestration loop over 200+ tools (265+ market-data endpoints, four web-research services, a 17-library Python sandbox), a two-stage context-management algorithm for unbounded conversations, a single-call computation paradigm (data → estimation → figures in one sandboxed execution), and a unified document pipeline emitting PDF, DOCX, Beamer and spreadsheets with live formulas from one Markdown artifact. The paper documents each mechanism as implemented so the design can be reproduced.

# 📌 Status

# 🎯 What this paper does

Large language models are fluent in financial discourse, but fluency does not guarantee numerically rigorous, reproducible or citable analysis. This paper describes the architecture of VQuant, a platform that couples a reasoning-augmented LLM with a deterministic execution substrate, and argues that the resulting design pattern — the model plans and synthesizes, auditable code computes — is reusable beyond finance.

Four mechanisms are documented at implementation level:

1. an agentic orchestration loop exposing 200+ financial and analytical tools through a constrained function-calling protocol (recursive tool-calling loop, tool-use/tool-result invariant, batched-execution throttle);
2. a context-management algorithm preserving coherence over effectively unbounded histories via adaptive summarization under a token budget (two-stage machinery, token-accounting correction);
3. a single-call computation paradigm that consolidates data acquisition, estimation and figure rendering into one sandboxed Python execution (inter-process sandbox that exec's model-authored code and captures its figures);
4. a unified document pipeline emitting PDF, DOCX, LaTeX Beamer decks and spreadsheet workbooks with live formulas from a single Markdown artifact, streamed through a Server-Sent-Event protocol with eight-way event dispatch.

The classical quantitative methods the platform automates are conventional and relegated to the appendix.

# 🏆 Headline results

# 🗂 Repository layout

vquant-llm-quant-platform/
└── paper/                    # main.tex + sections/ + figures/ (TikZ diagrams) + tables/ + appendix/ → main.pdf
    ├── figures/              # architecture, loop, context, sandbox, streaming, docpipeline (TikZ)
    └── tables/               # tools, models, events, robustness

No src/, scripts/ or data/: the paper has no empirical pipeline.

# 🚀 Quick start

cd paper && make      # latexmk → main.pdf (pdflatex + bibtex, TikZ diagrams compiled inline)

# 🗃 Data

Not applicable. The paper describes a software system; the platform itself (TypeScript/Python) is maintained in the spbgit repository vquant.git and deployed on the MacLustr cluster.

# 🔬 Methodology

1. Architecture description — components, invariants and protocols are specified as implemented (§3–§4).
2. Mechanism walkthroughs — orchestration loop, context management, single-call computation, document pipeline (§4).
3. Design evaluation — qualitative robustness discussion of failure modes and mitigations (§6).
4. Appendix — the conventional quantitative methods automated by the platform.

# 🖼 Figures & tables inventory

Tables (paper/tables/*.tex): tool catalogue, models, streaming events, robustness.

# ⚠️ Limitations

• Descriptive system paper: no benchmark against alternative agent frameworks is reported.
• Tool counts and endpoint inventories reflect the platform as of June 2026.
• Reproducibility of the design is documented; reproducing the platform requires the private vquant.git code base and API keys.

# 📝 Citation

@techreport{boucher2026vquant,
  author      = {Boucher, Simon-Pierre},
  title       = {VQuant: An Agentic Large Language Model Architecture for Automated Quantitative Financial Analysis and Document Generation},
  institution = {Universit\'e du Qu\'ebec en Outaouais, D\'epartement des sciences administratives},
  type        = {Working Paper},
  number      = {13},
  year        = {2026},
  month       = {June}
}

# 👤 Author & contact

Simon-Pierre Boucher — Département des sciences administratives, Université du Québec en Outaouais (UQO), Gatineau, QC. 📧 simon-pierre.boucher@uqo.ca

# 🗒 Notes

• Numbering: first circulated as “Working Paper No. 10” (June 2026); renumbered 13 on 2026-10-17 because No. 10 is The Assessment Gap in Quebec (see Recherche-UQO/HARMONISATION.md).
• Cover date normalised from “June 13, 2026” to “June 2026”.
• The VQuant platform code is not in this repository (spbgit vquant.git).