Optimize technology risk limit allocation
Allocate scarce aggregate technology risk limits across discrete locally executable operating envelopes, preserving option relations and common loss once; maximize expected net value after a capital charge subject to nominal, expected-loss, CVaR, economic-capital and RAROC appetite, then reconcile selected unit capital with exact or seeded Shapley allocation.
What it's for
Moves technology risk appetite from one enterprise ceiling to an investable operating system: which units receive scarce risk capacity, what value it enables, and how diversification changes the true capital consumed.
What you give it
Every tool is a pure function: it computes from what you pass it and reads nothing on its own. Inputs split into evidence you supply from your connected systems, calibration your team owns, and numerical controls that affect precision but never the result's meaning.
| Field | Type | Role | Required |
|---|---|---|---|
| beam_width | integer ≥ 10, ≤ 100000 | Numerical control | Optional |
| capital_charge_rate | number ≥ 0, ≤ 10 | Your calibration | Optional |
| confidence_level | number ≥ 0.5, ≤ 0.999 | Your calibration | Optional |
| cvar_loss_limit | number ≥ 0 | Your calibration | Yes |
| detail_limit | integer ≥ 1, ≤ 500 | Your calibration | Optional |
| economic_capital_limit | number ≥ 0 | Your calibration | Yes |
| exact_state_limit | integer ≥ 1, ≤ 2000000 | Your calibration | Optional |
| expected_loss_limit | number ≥ 0 | Your calibration | Yes |
| minimum_portfolio_raroc | number ≥ -10, ≤ 100 | Your calibration | Optional |
| permutation_draws | integer ≥ 10, ≤ 100000 | Numerical control | Optional |
| scenarios | array of objects (2 fields) | Evidence | Yes |
| seed | integer ≥ 0 | Numerical control | Optional |
| shared_loss_groups | array of objects (4 fields) ≥ 0 items | Evidence | Optional |
| total_nominal_limit | number ≥ 0 | Your calibration | Yes |
| unit_limit_options | array of objects (8 fields) | Evidence | Yes |
Each unit_limit_options
record
| Field | Type | Required |
|---|---|---|
| dependency_option_ids | array of string | Yes |
| evidence_verified | boolean | Yes |
| exclusion_option_ids | array of string | Yes |
| gross_value_by_scenario | object | Yes |
| id | string (non-empty) | Yes |
| loss_by_scenario | object | Yes |
| nominal_risk_limit | number (≥ 0) | Yes |
| unit_id | string (non-empty) | Yes |
{
"cvar_loss_limit": 140,
"economic_capital_limit": 110,
"expected_loss_limit": 40,
"scenarios": [
{
"id": "base",
"probability": 0.8
},
{
"id": "stress",
"probability": 0.15
},
{
"id": "severe",
"probability": 0.05
}
],
"shared_loss_groups": [
{
"evidence_verified": true,
"id": "cloud-common",
"loss_by_scenario": {
"base": 0,
"severe": 20,
"stress": 10
},
"member_option_ids": [
"platform-grow",
"data-grow"
]
}
],
"total_nominal_limit": 100,
"unit_limit_options": [
{
"dependency_option_ids": [],
"evidence_verified": true,
"exclusion_option_ids": [],
"gross_value_by_scenario": {
"base": 0,
"severe": 0,
"stress": 0
},Truncated for display — the full payload is 109 lines.
What you get back
This is the actual output of running the example above — computed by the same function the platform calls, not an illustration.
{
"capital_allocation": {
"effort": 4,
"method": "exact_shapley",
"reconciliation_error": 0
},
"configuration": {
"capital_charge_rate": 0.1,
"confidence_level": 0.95,
"cvar_loss_limit": 140,
"economic_capital_limit": 110,
"expected_loss_limit": 40,
"minimum_portfolio_raroc": 0,
"option_count": 4,
"scenario_count": 3,
"total_nominal_limit": 100,
"unit_count": 2
},
"decision": "risk_limits_allocated",
"failed_gates": [],
"finding": "risk_adjusted_limit_allocation_identified",
"governance": [
"Limits apply to aggregate technology risk units; they are not budgets, targets or scores for named people or teams.",
"Finance and risk own loss/value scenarios, appetite, capital charge and RAROC; engineering owners supply executable operating envelopes and dependency evidence.",
"Shapley allocation preserves negative diversification contributions and reconciles to portfolio economic capital; it does not imply causality or accounting ownership.",
"A beam result is explicitly incomplete and requires challenger allocations before approval."
],
"method": "risk_budget_value_cvar_capital_exact_or_beam_v1",
"pareto_frontier": [
{
"economic_capital": 0,
"expected_gross_value": 0,
"expected_loss": 0,
"expected_net_value": 0,
"loss_conditional_value_at_risk": 0,
"loss_value_at_risk": 0,
"option_ids": [
"platform-low",
"data-low"
],
"raroc": null,
"risk_adjusted_value": 0,
"total_nominal_limit": 0
},Truncated for display — the full payload is 133 lines.
How it works
Constrained optimization — Pick the best feasible option under real limits — budget, headcount, dependencies, capacity — rather than ranking a list and hoping it fits.
- 1 For each aggregate technology unit, admit only evidence-verified operating envelopes with a nominal risk limit, coherent gross-value/direct-loss vectors and complete option dependencies/exclusions; activate each shared loss once for any selected member option.
- 2 Enumerate the Cartesian allocation when the governed state count permits, otherwise use a deterministic unit-wise beam; reject allocations that breach nominal, expected-loss, CVaR, economic-capital, RAROC or relation constraints and retain the nondominated value/risk frontier.
- 3 Select the highest value after the finance-owned capital charge, then allocate expected loss and economic capital to chosen units with exact Shapley through twelve units or seeded permutations beyond that, preserving diversification contributions and reconciliation.
Before you trust it
Every tool in the catalog ships with the conditions under which its answer is meaningful — and the conditions under which it should abstain instead of guessing.
Assumptions & guardrails
- Objectives use commensurable locally governed value units, constraints reflect real feasibility, and uncertainty covers plausible adverse inputs.
- Every unit has a complete executable option set including its conservative operating envelope; values/losses share one horizon and price basis; option relations and common losses are complete; board appetite and finance capital charges are current.
- The recommendation is optimal only for its stated objective, feasible set, evidence, and solver guarantee; it is not a universal management optimum.
- The result allocates limits to aggregate technology-risk units, not budgets, targets or scores for named people or teams. A beam frontier is incomplete and no operating envelope is authorized without risk, finance and technology approval.
Minimum evidence
- unit_limit_options: required and organization-defined
- scenarios: required and organization-defined
- total_nominal_limit: required and organization-defined
- expected_loss_limit: required and organization-defined
- cvar_loss_limit: required and organization-defined
- economic_capital_limit: required and organization-defined
How to validate it
Backtest the chosen action against simple feasible baselines on held-out scenarios, sweep costs/constraints/risk tolerance, and require constraint feasibility under adverse inputs.
Calibrating it to your org
Same for everyone
The mathematical kernel, validation rules, method version, and JSON output semantics are organization-independent; no tenant-trained coefficients or company benchmark is embedded in the function.
Specific to you
- versioned unit-option-scenario matrix on one finance basis, verified option relation graph, common-loss activation map, exact or bounded beam allocation candidates, nondominated value/risk frontier and reconciled exact/permutation Shapley expected-loss/economic-capital allocation
- aggregate unit boundary, conservative operating envelope, prospective evidence, scenario value/loss and dependence, common-loss identity, relations, nominal/expected/CVaR/capital limits, confidence, capital charge, RAROC, exact/beam boundary, Shapley draws, human approval and refresh cadence
Calibration workflow
- 1 Define the management decision, target outcome, aggregate unit, privacy boundary, cadence, and prediction/intervention horizon for this organization.
- 2 Build a tenant-scoped historical cohort using only information available before each prediction or decision; preserve zero periods, censoring, assignment probabilities, and unresolved outcomes when the method requires them.
- 3 Estimate statistical parameters on training history, but obtain costs, utilities, risk tolerance, practical-effect thresholds, capacity, and policy constraints from accountable decision owners.
- 4 Validate on later time windows or held-out aggregate units at the deployment grain, against a simple baseline and the function-specific validation strategy.
- 5 Deploy only if the returned decision clears evidence, overlap, calibration, robustness, and guardrail checks; warning, unsupported, schema-gap, and fallback decisions are abstentions.
- 6 Monitor realized outcomes, data drift, coverage, and decision regret; recalibrate at a governed cadence or after a detected regime/definition change, never merely because a stakeholder dislikes the result.
Call it from your AI
You don't wire up 388 tools in your MCP client. The GitRevio MCP server exposes 20 tools, three of which let an agent search the catalog, read a tool's schema, and run it — so the assistant finds this one on its own.
gitrevio_capabilities_search
{ "q": "allocate scarce aggregate technology risk limits" }
→ finds "optimize_technology_risk_limit_allocation"
gitrevio_capability_describe
{ "capability_id": "optimize_technology_risk_limit_allocation" }
→ returns the input schema and agent guidance shown on this page
gitrevio_capability_run
{ "capability_id": "optimize_technology_risk_limit_allocation", "arguments": { ... } }
→ returns the result shown aboveWorks in Claude Desktop, Claude Code, Cursor, Cline, Continue.dev, Goose and Aider. See the MCP server.