Phronesis Decision Quality Bench

Generated 2026-10-25T04:19:25Z

Modules — overview

ModuleFamilyStatusn_resolvedSample basis
P-RESbench-specificinsufficient_sample04 forecast_resolution_event row(s)
P-CALassurance gradeinsufficient_sample03 scored forecast(s) in the public ledger
P-EVDassurance gradescored1112 Decision Assets in the public ledger, 1 retired
P-ACTassurance gradeinsufficient_sample0—
P-SHPbench-specificinsufficient_sample0—
P-BBLbench-specificinsufficient_sample0—
P-ROBbench-specificinsufficient_sample0—
P-COSTbench-specificinsufficient_sample0—

Modules — details

P-RES [bench-specific measure]

Status
insufficient_sample
n_resolved
0
Sample basis
4 forecast_resolution_event row(s)
Notes
P-RES emits a score once at least 5 resolution events are recorded; current count = 4

P-CAL [assurance grade]

Status
insufficient_sample
n_resolved
0
Sample basis
3 scored forecast(s) in the public ledger
Notes
P-CAL emits a score once at least 30 forecasts are scored against ground truth; current scored count = 3

P-EVD [assurance grade]

Status
scored
n_resolved
11
Sample basis
12 Decision Assets in the public ledger, 1 retired
Score payload
total_graded11
E0_share0.0909
E1_share0.0909
E2_share0
E3_share0.8182
E4_share0
E5_share0
E0_count1
E1_count1
E2_count0
E3_count9
E4_count0
E5_count0

P-ACT [assurance grade]

Status
insufficient_sample
n_resolved
0
Sample basis
—
Notes
P-ACT lands in Bench Phase 2 — action_boundary scoring requires DA spec Phase 4 (live action_boundary computation) + resolved outcomes

P-SHP [bench-specific measure]

Status
insufficient_sample
n_resolved
0
Sample basis
—
Notes
P-SHP lands in Bench Phase 1 — sharpness measures the usefulness of calibrated forecasts

P-BBL [bench-specific measure]

Status
insufficient_sample
n_resolved
0
Sample basis
—
Notes
P-BBL lands in Bench Phase 1 — edge vs naive / market / expert / model baselines

P-ROB [bench-specific measure]

Status
insufficient_sample
n_resolved
0
Sample basis
—
Notes
P-ROB lands in Bench Phase 2 — robustness ties to the Bench + evidence_graph.contradiction_flags

P-COST [bench-specific measure]

Status
insufficient_sample
n_resolved
0
Sample basis
—
Notes
P-COST lands in Bench Phase 2 — needs the per-DA retrieval cost ledger (Decision Feed Phase 1 follow-on)