Skip to content

Pennant test ledger

Append-only record of every test executed under the Pennant research line. New tests get the next sequential PEN-TEST-NNN ID at the moment they are planned (before execution), and the row is filled in as the test completes. PEN-TEST-001 through PEN-TEST-005 are retroactive assignments from the 2026-05-11 Phase 11 session and the 2026-05-12 Phase 12 Stage 2a population study.

The "Strategy" column references entries in strategies/registry.md; the "Cohort" column references directories under cohorts/; the "Report" column points into reports/. Each test also has its own directory at tests/<date>_<test-id>/ with the script(s), logs, and test-specific outputs.

Test ID Date Strategy Cohort Headline result Report Status
PEN-TEST-001 2026-05-11 (detection-only) DET-BASELINE-2026-05-11, DET-V1-2026-05-11 V1 (pennant 10–20, flagpole 1–5) cuts event volume 71% (15,534 → 4,533) with hit-rate at +15% MFE essentially unchanged (27.0% baseline vs ~27% V1) reports/Pennant/pennant_criteria_ab_test_2026-05-11.md complete
PEN-TEST-002 2026-05-11 (detection-only) DET-V2-2026-05-11 V2 (pennant 7–17, flagpole 1–5) keeps 39% of baseline volume (6,108 events) with hit-rate +0.4pp over baseline at +15% MFE reports/Pennant/pennant_criteria_ab_test_v2_2026-05-11.md complete
PEN-TEST-003 2026-05-11 (detection-only) DET-V3-2026-05-11, DET-V4-2026-05-11 V3 (flagpole 1–3): 5,200 events. V4 (flagpole 1–2): 4,101 events — too aggressive; 30-day endpoint mean below baseline despite tighter selection reports/Pennant/pennant_criteria_ab_test_v3_v4_2026-05-11.md complete
PEN-TEST-004 2026-05-11 PEN-STOCK-001 DET-BASELINE-2026-05-11, DET-V1-2026-05-11, DET-V2-2026-05-11, DET-V3-2026-05-11, DET-V4-2026-05-11 Baseline maximum CAGR (7.50%, $40,398 final from $10K); V2 best Sharpe (0.617) with –21.4% max DD vs Baseline's –38.1% reports/Pennant/pennant_strategy_backtest_2026-05-11.md complete
PEN-TEST-005 2026-05-12 (population study) DET-BASELINE-2026-05-11 15,528 baseline pennants: near-symmetric outcome distribution (41% reach +10% MFE, 36% reach –10% MAE). Median winner gives back 68% of peak by day 30; median loser recovers 88% of trough reports/Pennant/pennant_population_behavior_2026-05-12.md complete

Conventions

  • Test ID is sequential; never reused. Reserved before the test runs.
  • Date = the date the test produced its final headline result, in ISO format.
  • Strategy = strategy registry ID, or (detection-only) / (population study) for tests that don't simulate a strategy.
  • Cohort = comma-separated list of cohort IDs consumed.
  • Headline result = one-sentence summary. Keep terse; details in the report.
  • Report = relative path to the canonical report markdown.
  • Status = planned / running / complete / failed / superseded.

Reserving an ID for a future test

Pick the next unused PEN-TEST-NNN. Add a row with Status: planned, fill in the rest as the test executes. Do not reorder rows.