Optimize forecast elicitation portfolio
Choose which independent human or model forecasts to obtain next by learning chronologically validated contextual directional skill, simulating conservative entropy reduction, pricing decision relevance and removing duplicated information under budget and source-capacity constraints.
What it's for
Gives CTOs and investors a rigorous answer to 'who or what should we ask before making this bet?'—buying diverse information rather than more opinions from the same reporting chain.
What you give it
Inputs split into evidence read from your connected systems, calibration your team owns, and numerical controls that affect precision but never the result's meaning.
| Field | Type | Role | Required |
|---|---|---|---|
| conservative_quantile | number > 0, < 0.5 | Your calibration | Optional |
| current_questions | array of objects (6 fields) | Evidence | Yes |
| elicitation_budget | number ≥ 0 | Your calibration | Yes |
| elicitation_options | array of objects (6 fields) | Evidence | Yes |
| exact_option_limit | integer ≥ 1, ≤ 22 | Your calibration | Optional |
| max_detail_rows | integer ≥ 1, ≤ 500 | Numerical control | Optional |
| minimum_source_context_support | integer ≥ 3, ≤ 100000 | Your calibration | Optional |
| minimum_training_questions | integer ≥ 20, ≤ 100000 | Your calibration | Optional |
| minimum_validation_forecasts | integer ≥ 10, ≤ 100000 | Your calibration | Optional |
| resolved_forecasts | array of objects (7 fields) | Evidence | Yes |
| resolved_questions | array of objects (5 fields) | Evidence | Yes |
| seed | integer ≥ 0, ≤ 4294967295 | Numerical control | Optional |
| simulation_draws | integer ≥ 200, ≤ 20000 | Numerical control | Optional |
| source_capacities | array of objects (3 fields) | Evidence | Yes |
| source_dependencies | array of objects (5 fields) ≥ 0 items | Evidence | Yes |
| validation_fraction | number ≥ 0.2, ≤ 0.5 | Your calibration | Optional |
Each resolved_forecasts
record
| Field | Type | Required |
|---|---|---|
| evidence_verified | boolean | Yes |
| forecast_at_ms | number (≥ 0) | Yes |
| id | string (non-empty) | Yes |
| probability | number (≥ 0, ≤ 1) | Yes |
| question_id | string (non-empty) | Yes |
| sealed_independent | boolean | Yes |
| source_ref | string (non-empty) | Yes |
{
"current_questions": [
{
"context_id": "roadmap",
"current_probability": 0.5,
"decision_value": 100000,
"evidence_verified": true,
"id": "decision-a",
"maximum_sources": 2
}
],
"elicitation_budget": 2,
"elicitation_options": [
{
"elicitation_cost": 1,
"id": "elicitation-expert-a",
"independence_group_id": "shared-a",
"question_id": "decision-a",
"response_probability": 1,
"source_ref": "expert-a"
},
{
"elicitation_cost": 1,
"id": "elicitation-copier-a",
"independence_group_id": "shared-a",
"question_id": "decision-a",
"response_probability": 1,
"source_ref": "copier-a"
},
{
"elicitation_cost": 1,
"id": "elicitation-expert-b",
"independence_group_id": "expert-b",
"question_id": "decision-a",
"response_probability": 1,
"source_ref": "expert-b"
},
{
"elicitation_cost": 1,
"id": "elicitation-noise",
"independence_group_id": "noise",
"question_id": "decision-a",
"response_probability": 1,
"source_ref": "noise"Truncated for display — the full payload is 4387 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.
{
"assumptions": [
"Historical sealed directional forecasts are exchangeable enough within the governed context to estimate a Beta skill distribution, and consistently anti-predictive sources may be direction-inverted prospectively.",
"Decision value is the maximum value of reducing uncertainty for the current question; declared dependence and shared independence groups conservatively reduce duplicated information."
],
"decision": "forecast_elicitation_portfolio_selected",
"limitations": [
"Entropy reduction is a model-based information-value proxy, not guaranteed cash value; response quality, strategic behavior, regime shifts and undisclosed common sources can reduce realized value.",
"Exact mode certifies only the supplied finite options and constraints. Greedy mode is deterministic but not globally optimal and neither mode authorizes coercive elicitation or personnel evaluation."
],
"method": "chronological_beta_information_value_dependency_portfolio_v1",
"portfolio": {
"dependency_redundancy_penalty": 0,
"elicitation_budget": 2,
"elicitation_cost": 2,
"globally_optimal_for_supplied_finite_model": true,
"gross_conservative_decision_value": 143703.4948,
"net_conservative_decision_value": 143703.4948,
"selected_options": 2,
"solver": "exact_subset_enumeration"
},
"profile_validation_passed": true,
"rejected_options": [
{
"option_id": "elicitation-copier-a",
"question_id": "decision-a",
"reason": "not_in_value_maximizing_feasible_portfolio"
},
{
"option_id": "elicitation-noise",
"question_id": "decision-a",
"reason": "not_in_value_maximizing_feasible_portfolio"
}
],
"reproducibility": {
"conservative_quantile": 0.1,
"seed": 7,
"simulation_draws": 200
},
"selected_elicitations": [
{
"allocated_redundancy_penalty": 0,
"conservative_decision_value": 74649.1028,
"conservative_information_nats": 0.517428,Truncated for display — the full payload is 79 lines.
How it works
Forecasting & survival — Estimate when something completes or fails, with censoring and unresolved work handled honestly rather than dropped.
- 1 Split resolved questions by time, estimate source-context directional skill with Beta shrinkage on earlier questions, and require the frozen profiles to beat chance on later forecasts.
- 2 For each current question-source option, simulate the mutual information of a symmetric binary signal, use a conservative posterior quantile, scale it by response probability and the governed value of resolving uncertainty.
- 3 Select a budget/capacity-feasible set with at most one source per declared independence group, subtract pairwise dependency redundancy, and disclose exact finite optimization versus deterministic greedy search.
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
- Training examples precede their outcomes, censoring and unresolved work are represented, and deployment populations remain comparable to validation cohorts.
- Historical questions and source identities match the current context, sealed forecasts precede outcomes, decision values are governed upper bounds, and dependency/independence groups cover common upstream information.
- A predictive interval or risk estimate is not a deadline promise, causal explanation, or individual-performance judgment.
- Information value is not cash realization, anti-predictive inversion must be prospectively governed, and the portfolio never authorizes forced participation or individual performance evaluation.
Minimum evidence
- resolved_questions: required and organization-defined
- resolved_forecasts: required and organization-defined
- current_questions: required and organization-defined
- elicitation_options: required and organization-defined
- source_dependencies: at least 0 rows/items
- source_capacities: required and organization-defined
- elicitation_budget: required and organization-defined
How to validate it
Use chronological train/calibration/test windows, compare proper scores and decision value with a simple baseline, and recalibrate only from outcomes resolved after prediction time.
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
- source-context directional correctness histories split by question resolution time, Beta skill posteriors, current entropy-reduction simulations and one finite option/constraint matrix with pairwise redundancy
- question/context stability, decision-value upper bound, source consent/availability, cost basis, response probability, source capacity, independence/dependency taxonomy, minimum skill support, later chance benchmark, conservative information quantile, budget and exact-versus-greedy disclosure
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 18 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": "choose which independent human or model" }
→ finds "optimize_forecast_elicitation_portfolio"
gitrevio_capability_describe
{ "capability_id": "optimize_forecast_elicitation_portfolio" }
→ returns the input schema and agent guidance shown on this page
gitrevio_capability_run
{ "capability_id": "optimize_forecast_elicitation_portfolio", "arguments": { ... } }
→ returns the result shown aboveWorks in Claude Desktop, Claude Code, Cursor, Cline, Continue.dev, Goose and Aider. See the MCP server.
Related tools
Audit forecast ensemble lineage integrity
Audit whether a claimed forecast ensemble is a complete, independently sealed and point-in-time evidence set rather than duplicated consensus, then score only mature uncontaminated outcomes.
Forecast dependency adjusted consensus
Combine independently sealed human and model forecasts while learning context base rates and source reliability on earlier questions, discounting empirical and declared information dependence, and abstaining unless later questions beat both base rates and naive consensus.
Audit analytical specification multiverse
Audit whether an analytical conclusion survives a prespecified multiverse of admissible windows, cohorts, metrics, and models using aligned bootstrap draws, a weighted specification curve, practical-effect support gates, and descriptive choice-influence diagnostics.
Audit analytics challenger independence
Audit whether an analytical challenger supplies genuinely independent error information: use paired temporal moving-block bootstrap bounds on error correlation, incumbent-failure catch rate and common-mode joint failure, with simultaneous Bonferroni control across every screened challenger and explicit evidence gates.
Audit analytics function calibration readiness
Gate analytical functions on paired out-of-time decision loss against a frozen baseline using temporal moving-block bootstrap, autocorrelation- and weight-adjusted effective sample size, evidence coverage, lower confidence bounds, improvement probability and recent degradation rather than declaring a model calibrated from training fit.
Audit analytics transportability
Audit whether locally calibrated analytical functions retain decision-loss improvement across target-similar operating environments using similarity-weighted random-effects meta-analysis, between-environment variance, I-squared, sign consistency and a conservative target prediction interval.