# OpenProse Quant

Contracts for building quantitative models, forecasting, calibration, simulation, optimization and decisions under uncertainty. Use the selected [OpenProse kernel](../core/README.md); the execution environment supplies calculation tools and permitted effects. Reading an index does not adopt or invoke its entries.

| Work to perform | Contract |
|---|---|
| Assemble inputs as they were knowable at a decision time | [Point-in-time data](contracts/point-in-time-data.md) |
| Construct and fit a usable quantitative model | [Model construction](contracts/model-construction.md) |
| Fit dependent component losses and calculate joint tail risk | [Joint loss construction](contracts/loss-distribution.md) |
| Recompute acceptance evidence for frozen models and uses | [Model revalidation](contracts/model-revalidation.md) |
| Turn a monitoring episode into a model-use disposition | [Model renewal](contracts/model-renewal.md) |
| Issue a point, quantile or probability forecast | [Forecast production](contracts/forecast-production.md) |
| Compare forecasting strategies over historical origins | [Forecast experiment](contracts/forecast-experiment.md) |
| Fit and assess a probability recalibration mapping | [Probability calibration](contracts/probability-calibration.md) |
| Turn uncertainty and explicit losses into a decision rule | [Decision policy](contracts/decision-policy.md) |
| Estimate a quantity by executing a simulation | [Simulation study](contracts/simulation-study.md) |
| Solve a constrained quantitative problem | [Constrained optimization](contracts/constrained-optimization.md) |
| Reproduce an existing calculation | [Model reproduction](contracts/model-reproduction.md) |
| Document a model or assess supplied quantitative evidence | [Specialist contract index](contracts/README.md) |

Supply the selected work with its subject and data identities, decision or information date, units and horizon, permitted calculation scope, output location, and the contract-valued workers or criteria it requests. State which roles are shared across comparisons. A review of supplied results does not execute the model; an active calculation requires actual permitted capabilities. Neither implies authority to trade, transfer funds or change accepted operational state.

[Tests and invocable commissions](tests/README.md) include a temporal model comparison, probability calibration and constrained decisions, and simulation and optimization. They use original synthetic records, not customer data. Test definitions do not establish successful execution.

For a connected starting point, [dependent-loss renewal](tests/renewal-tail/program.md) constructs and validates a replacement under changed joint losses, with a permitted [reference-shrinkage](methods/shrunk-joint-law.md) substitution. [Horizon renewal](tests/renewal-horizons/program.md) fits separate probability mappings and handles successful short-horizon validation alongside an overdue long-horizon evidence gap and proposed retirement. These examples preserve pending approvals and prior forecasts; they do not install ongoing monitors or enact financial changes.

The specialist definitions originated in Open Quant; this package retains their [MIT license](LICENSE.md). The package contains their maintained definitions and new active-work compositions. Select only the components relevant to the requested commission.