Quantitative models with identities, histories and evidence.
The catalogue combines statistical, forecasting, risk, simulation, optimization, decision and learned-model families inside one governed numerical runtime.
LQ Forecast
Trend, autoregression, state-space, probabilistic forecasting and model competition.
ForecastTime seriesLQ Risk
VaR, Expected Shortfall, dynamic covariance, volatility and probabilistic risk surfaces.
RiskStressLQ Optimize
Constrained mean-variance and numerical allocation systems.
OptimizationLQ Regime
Structural-break and probabilistic regime intelligence for changing systems.
RegimesLQ Simulate
Seeded stochastic simulation and correlated scenario generation.
SimulationLQ Decide
Adaptive routing, quantitative memory and model-selection policies.
DecisioningTrain, validate and promote numerical models.
The Foundry extends the governed registry with learned models while preserving dataset lineage, training evidence, calibration, benchmark results and production identity.
Ridge Regression LQM
Deterministic learned regression with safe JSON artifacts and evidence-bound training lineage.
Elastic Net LQM
Regularized regression for correlated quantitative features with explicit validation and calibration.
Gradient Boosted LQM
Nonlinear boosted numerical modelling with signed search and resource evidence.
Neural Forecast LQM
Windowed neural forecasting with checkpointing, measured resources and portable inference artifacts.
Compete before production
Model versions compete on measurable benchmark evidence before governance can advance them.
Execution evidence
Production results retain model identity, implementation hashes and verifiable execution provenance.