Result Tables Reference
This page documents the data tables returned by the get_job_result MCP tool after a computation completes. The tool always returns a lightweight summary and can optionally include specific data tables on request.
Summary (always returned)
The summary is included in every response and contains:
| Field | Type | Description |
|---|---|---|
n_sims |
integer | Number of Monte Carlo simulations run |
n_orgs |
integer | Number of organizations in the analysis |
n_multipliers |
integer | Number of impact multipliers (including formula-derived) |
n_distinct_portfolios |
integer | Total number of distinct optimal portfolios found |
funding |
object | Budget parameters: BudgetTotal, BudgetYearly, Years, VirtualOrgsPerYear, VirtualOrgsSamplingStrategy |
validation |
object | Validation results with warnings, errors, oks, infos arrays |
metadata |
object | Job metadata including data_source_title |
portfolios_evs_summary |
object | Key output statistics (see below) |
tables_available |
array | List of requestable table names with descriptions |
portfolios_evs_summary
This is the core output — summary statistics across all portfolios:
| Sub-field | Type | Description |
|---|---|---|
col_means |
named numeric | Mean expected value per portfolio across all simulations |
col_medians |
named numeric | Median expected value per portfolio |
col_quantiles |
matrix (7 x N) | Quantiles at p05, p10, p25, p50, p75, p90, p95 per portfolio |
risk_averse_cara_ev |
named numeric | CARA risk-averse certainty equivalent per portfolio |
cara_a |
numeric | CARA risk aversion coefficient (default 0.02) |
best_portfolio_per_sim |
named character | Which portfolio was optimal in each simulation |
frequency |
named numeric | How often each portfolio appears as optimal (proportion) |
virtual_org_pct |
named numeric | Percentage of funding allocated to virtual organizations per portfolio |
n_sims |
integer | Number of simulations |
n_portfolios |
integer | Number of portfolios |
Requestable Tables
These tables can be requested via the tables parameter. All are truncated to max_rows (default 100).
multiplier_range
Input multiplier definitions used for the analysis.
| Column | Type | Description |
|---|---|---|
| ImpactDifferentiator | string | Unique name for the multiplier |
| Min | numeric | Minimum value (5th percentile for normal/lognormal, literal bound for uniform) |
| Max | numeric | Maximum value (95th percentile for normal/lognormal, literal bound for uniform) |
| Distribution | string | One of: normal, lognormal, uniform, custom, formula, unity |
| Type | string | Weight type controlling how multiplier combines with org value |
| Category | string | Optional grouping category |
Typical size: 150-200 rows.
portfolios
Top portfolios ranked by mean expected value, with per-organization funding allocations. Each portfolio appears as multiple rows (one per organization per year).
| Column | Type | Description |
|---|---|---|
| portfolio | string | Portfolio identifier |
| organization | string | Organization name |
| funding | numeric | Funding allocated to this organization in this portfolio |
| year | integer | Funding year |
| n_is_optimal | integer | Number of simulations where this portfolio was optimal |
| expected_value | numeric | Mean expected value (from summary metrics) |
| risk_averse_ev | numeric | CARA risk-averse EV (from summary metrics) |
| frequency | numeric | Proportion of simulations where optimal |
| share_rff | numeric | Funding as share of room for funding |
The max_rows parameter controls how many top portfolios to return (not row count). All rows for the selected portfolios are included.
expected_values
Per-organization expected values, filtered to year 1.
| Column | Type | Description |
|---|---|---|
| organization | string | Organization name |
| value | numeric | Expected value for this organization |
| year | integer | Year (filtered to 1) |
Typical size: 30-60 rows (one per organization).
org_funding
Organization funding parameters per bracket.
| Column | Type | Description |
|---|---|---|
| organization | string | Organization name |
| variable | string | Parameter type (e.g., Room for funding, Discount, Utility multiplier, Model Coverage) |
| value | numeric | Parameter value |
| bracket | integer | Funding bracket number |
| year | integer | Year |
Typical size: 400-1000 rows.
org_characteristics
Organization values for each impact multiplier.
| Column | Type | Description |
|---|---|---|
| ImpactDifferentiator | string | Multiplier name |
| organization | string | Organization name |
| value | numeric | Mean characteristic value |
| VariableType | string | Variable type (e.g., Indicator restricts to 0/1) |
| SD | numeric | Standard deviation |
Typical size: 4,000-10,000 rows (n_orgs x n_multipliers).
multipliers_info_value
Information value analysis identifying which multipliers drive the most uncertainty. Returns up to 4 sub-tables:
- ev_regret: Expected value of regret per multiplier — how much expected value would improve if uncertainty about this multiplier were resolved
- correlation: Correlation between each multiplier and portfolio expected values
- mutual_information: Mutual information between multiplier values and optimal portfolio choice
- js_divergence: Jensen-Shannon divergence measuring how much the portfolio distribution shifts with each multiplier
Each sub-table typically has one row per multiplier.
multipliers
Raw simulated multiplier draws — the states of the world matrix.
| Column | Type | Description |
|---|---|---|
| sim | integer | Simulation index |
| (multiplier columns) | numeric | One column per multiplier, named by ImpactDifferentiator |
Typical size: 10,000 sims x 150-285 multipliers. Truncated to max_rows simulations.
org_specs
Per-organization specification matrices combined into a single long-format table. Shows how each multiplier value translates into organization-specific values after applying weight types.
| Column | Type | Description |
|---|---|---|
| organization | string | Organization name |
| sim | integer | Simulation index |
| (multiplier columns) | numeric | Organization-specific multiplier values |
Typical total size: n_orgs x n_sims x n_multipliers. Truncated to max_rows rows.
exp_values
Expected values for every simulation, organization, and year in long format.
| Column | Type | Description |
|---|---|---|
| sim | integer/string | Simulation index |
| organization | string | Organization name |
| year | integer/string | Year |
| value | numeric | Expected value |
Typical total size: n_sims x n_orgs x n_years. Truncated to max_rows rows.
Tables NOT exposed via MCP
- portfolios_evs: Full simulation x portfolio expected values matrix. Usually excluded at save time to reduce storage. Can be regenerated on demand but is too large for JSON transport.