Solve distributionally robust product portfolio
Solve a budgeted product portfolio under ambiguity in scenario probabilities, acyclic dependencies, mutually exclusive choices, and governed pairwise cannibalization or synergy using a total-variation uncertainty set.
What it's for
Finds the product portfolio that still makes sense when forecast probabilities are wrong—and shows exactly which downside scenarios the adversary exploits.
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 |
|---|---|---|---|
| budget | number ≥ 0 | Your calibration | Yes |
| max_detail_rows | integer ≥ 1, ≤ 500 | Numerical control | Optional |
| maximum_exact_states | integer ≥ 2, ≤ 16777216 | Numerical control | Optional |
| nominal_scenario_probabilities | array of number ≥ 2 items | Evidence | Yes |
| pairwise_effects | array of objects (4 fields) | Evidence | Yes |
| products | array of objects (5 fields) ≥ 1 item | Evidence | Yes |
| total_variation_radius | number ≥ 0, ≤ 1 | Your calibration | Optional |
Each products
record
| Field | Type | Required |
|---|---|---|
| contribution_scenarios | array of number (≥ 2 items) | Yes |
| dependency_ids | array of string | Yes |
| exclusive_group | any | Optional |
| id | string (non-empty) | Yes |
| investment_cost | number (≥ 0) | Yes |
{
"budget": 500000,
"nominal_scenario_probabilities": [
0.2,
0.5,
0.3
],
"pairwise_effects": [
{
"contribution_effect_scenarios": [
0,
100000,
250000
],
"id": "platform-enterprise-synergy",
"product_a": "platform",
"product_b": "enterprise"
}
],
"products": [
{
"contribution_scenarios": [
250000,
350000,
450000
],
"dependency_ids": [],
"exclusive_group": null,
"id": "platform",
"investment_cost": 200000
},
{
"contribution_scenarios": [
-100000,
700000,
1500000
],
"dependency_ids": [
"platform"
],
"exclusive_group": "growth-bet",
"id": "enterprise",
"investment_cost": 300000
},Truncated for display — the full payload is 60 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.
{
"adversarial_probability_shifts": [
{
"adversarial_probability": 0.15,
"nominal_probability": 0.3,
"probability_change": -0.15,
"scenario_index": 2
},
{
"adversarial_probability": 0.35,
"nominal_probability": 0.2,
"probability_change": 0.15,
"scenario_index": 0
}
],
"assumptions": [
"Contribution scenarios are incremental gross contribution before submitted investment cost; investment is subtracted once from every scenario and also consumes the governed budget.",
"Dependencies are complete and acyclic, mutually exclusive groups permit at most one selected product, and each unordered pairwise effect is a governed symmetric cannibalization or synergy estimate without hidden double counting.",
"The total-variation radius is calibrated from out-of-time probability error and reallocates probability only among represented joint scenarios; it cannot protect against omitted regimes, structural model error, or wrong contribution estimates.",
"Exact mode certifies only the submitted discrete model; heuristic mode has no global optimality certificate, and neither output is investment advice or evidence about individual employee performance."
],
"configuration": {
"budget": 500000,
"maximum_exact_states": 262144,
"pairwise_effect_count": 1,
"product_count": 3,
"represented_state_count": 8,
"scenario_count": 3,
"solver_mode": "exact_enumeration",
"state_count_is_capped": false,
"total_variation_radius": 0.15
},
"decision": "distributionally_robust_product_portfolio_optimized_exact",
"method": "dependency_interaction_total_variation_robust_product_portfolio_v1",
"product_diagnostics": [
{
"dependency_ids": [
"platform"
],
"exclusive_group": "growth-bet",
"investment_cost": 300000,
"nominal_expected_standalone_contribution_before_investment": 780000,
"nominal_optimum_selected": true,
"product_id": "enterprise",Truncated for display — the full payload is 98 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 Freeze gross contribution paths, investment costs, a dependency DAG, exclusivity groups, symmetric pair effects, nominal joint-scenario probabilities, and an out-of-time calibrated total-variation radius.
- 2 For each feasible portfolio, subtract investment once, apply every selected pair effect, and move probability adversarially from higher- to lower-value represented scenarios within the ambiguity budget.
- 3 Select maximum worst-case expected net value, compare it with the nominal optimum, expose adversarial probability shifts, and label dependency-closure/synergy fallback results as uncertified screening.
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.
- Scenario contributions are gross before submitted investment, dependencies/exclusivities and pair effects are complete without double counting, nominal columns are joint futures, and the ambiguity radius reflects later probability error.
- The recommendation is optimal only for its stated objective, feasible set, evidence, and solver guarantee; it is not a universal management optimum.
- Total-variation robustness cannot invent omitted regimes or repair wrong payoff models; heuristic mode has no global certificate, and the result is neither investment advice nor a person-level assessment.
Minimum evidence
- products: at least 1 rows/items
- pairwise_effects: required and organization-defined
- nominal_scenario_probabilities: at least 2 rows/items
- budget: 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
- aligned gross contribution scenarios, nominal joint-scenario probabilities, and out-of-time probability-error calibration for the ambiguity radius
- contribution/investment perimeter, horizon/currency, dependency and exclusivity completeness, symmetric interaction attribution, nominal probability vintage, ambiguity-radius backtest, budget, and solver promotion
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": "solve a budgeted product portfolio under" }
→ finds "solve_distributionally_robust_product_portfolio"
gitrevio_capability_describe
{ "capability_id": "solve_distributionally_robust_product_portfolio" }
→ returns the input schema and agent guidance shown on this page
gitrevio_capability_run
{ "capability_id": "solve_distributionally_robust_product_portfolio", "arguments": { ... } }
→ returns the result shown aboveWorks in Claude Desktop, Claude Code, Cursor, Cline, Continue.dev, Goose and Aider. See the MCP server.