PEN-TEST-002 — Phase 11a-2 Pennant criteria A/B (V2)¶

Field	Value
Test ID	PEN-TEST-002
Date	2026-05-11
Strategy	(detection-only)
Cohort produced	DET-V2-2026-05-11
Cohort consumed (for baseline comparison)	DET-BASELINE-2026-05-11
Status	complete

Purpose¶

Re-run of the Phase 11a A/B test against a less-aggressive variant (V2: pennant 7–17, flagpole 1–5). Baseline events and outcomes from PEN-TEST-001 are reused unchanged; only V2 is freshly scanned.

Method¶

run_v2.py is a small driver over the same harness machinery as PEN-TEST-001 — mutate config, call detector, write parquet, compute outcomes inline. Override mechanism, outcome formula, and universe are identical to PEN-TEST-001.

Headline¶

V2 keeps 39 % of baseline volume (6,108 events) with the +15 % MFE hit-rate +0.4 pp over baseline. V2 is the sweet spot in the duration × flagpole-tightness sweep — neither the loosest (Baseline) nor the strictest (V4 in PEN-TEST-003) variant matches it on per-event quality.

Files in this directory¶

run_v2.py — harness driver
analyze_v2.py — statistics emitter
run_v2.log, run_v2.stdout.log — run logs
summary_v2.json — headline JSON
report.md → ../../reports/Pennant/pennant_criteria_ab_test_v2_2026-05-11.md

Cohort outputs¶

Pennant/cohorts/DET-V2-2026-05-11/{events,outcomes}.parquet