stockvaluation-io/stockvaluation_io
Local-first automated DCF valuation platform with deterministic valuation math and optional LLM-powered research/narratives using Aswath Damodaran dataset and methodology.
What's novel
Local-first automated DCF valuation platform with deterministic valuation math and optional LLM-powered research/narratives using Aswath Damodaran dataset and methodology.
Code Analysis
10 files read · 3 roundsAn agent-native valuation workflow that splits AI research/explanation from deterministic DCF math, enforcing evidence-governed assumptions, anchored scenario values, and auditable guided-question trails before any valuation number is produced.
Strengths
The anchored-value system (driver_anchors.py) is a genuinely novel mechanism that prevents LLM agents from injecting arbitrary numbers into valuation models — every numeric scenario value must match a server-computed anchor or be explicitly flagged as user input. The evidence governance layer (evidence_packet.py, segment_economics.py) enforces strict provenance, URL validation, confidence gating, and conflict detection before any evidence can affect recalculation.
Weaknesses
The mcp_tools.py file at 203KB is a monolith that likely contains significant duplication and would benefit from further decomposition. I could not verify the Java DCF service implementation or the test suite depth, leaving the numerical core and test coverage partially unverified.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
34
Contributors
1
Files
657
Active weeks
2
Repository
Language
TypeScript
Stars
8
Forks
2
License
Apache-2.0