Skip to main content
Open category definition

FinanceGPT Large Quantitative Model specification

A governed quantitative or generative financial model module that consumes typed numeric, categorical or time-series features and emits typed, versioned quantitative evidence. It is not a language model.

Machine-readable specification

Allowed inputs

Numeric
Categorical
Time Series

Excluded inputs

×Raw Document Text
×Freeform Language

Task families

ForecastingValuationRiskSimulationOptimizationScenario GenerationDerivatives Surface Generation

Evidence contract

financegpt.lqm.evidence.v1

Checkpoint, evaluation, provenance and execution evidence stay attached to quantitative outputs.

Required evaluation gates

Statistical Fidelity
Temporal Structure
Cross Series Fidelity
Tail Behaviour
Scenario Diversity
Adversarial Robustness
Data Poisoning Robustness

Non-negotiable authority boundary

A QLM language runtime may explain a verified LQM Evidence Packet, but it never receives LQM latent state and never becomes numeric authority.

Publication to a model hub and promotion into FinanceGPT are separate governed actions. Neither implies the other.