geometrybase/hk-ipo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Refresh IPO analysis model calibration

Browse Source 
Request:
- Re-analyze the IPO model using the updated historical archive after T1 demand backfill.

Changes:
- Regenerate the v0 analysis dataset from the current SQLite archive.
- Refresh the v0 calibration report with expanded T1 coverage and new empirical bucket rates.
- Update the report template to show pending T1 rows and field-level blanks.
- Clarify v0 limitations and record why the score formula stays unchanged for this refresh.

Verification:
- Ran scripts/build_analysis_dataset.py against data/hk_ipo.sqlite.
- Ran py_compile for scripts/build_analysis_dataset.py.
- Checked dataset row count, T1 demand coverage, source-only T1 gaps, and repo-relative paths.
- Ran git diff --check.

Next useful context:
- T1 structured coverage is now 291 rows, with 06106 and 06675 still pending_not_due.
- The high-conviction T1 bucket remains differentiated, but middle and low buckets are still not monotonic enough for a v1 rule change.
This commit is contained in:
geometrybase
2026-06-15 14:05:34 +00:00
parent 6d05056609
commit 58ad869f84
5 changed files with 347 additions and 312 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rules/ipo_score_v0.yaml
+2 -2
View File
@@ -143,8 +143,8 @@ calibration_policy:
dataset_path: data/snapshots/analysis_model_v0_dataset.csv
known_limitations:
- T0 has weaker discrimination than T1 because demand data is not available yet.
- T1 structured demand facts are incomplete for older HTML-only allotment notices.
- T0 has weaker discrimination than T1 because subscription demand is not available at prospectus stage.
- T1 is structurally complete for listed rows, but some individual demand fields remain null when the archived source does not explicitly state them.
- Industry labels and issuer fundamentals are not yet sufficiently structured for v0.
- T2 grey-market signal is intentionally disabled until a reliable source exists.
- Current D1 outcomes are from archived market data and should be reviewed for extreme corporate-action or data-vendor anomalies.
rules/rule_change_log.md
+25
View File
@@ -1,5 +1,30 @@
# Rule Change Log
## 2026-06-15 - Refresh `ipo_score_v0` after T1 demand backfill
Request:
- Re-analyze the model using the known historical archive after T1 demand text backfill.
Change:
- Regenerated `data/snapshots/analysis_model_v0_dataset.csv` from the current SQLite archive.
- Refreshed `reports/2026-06-15_analysis_model_v0.md` with the expanded T1 demand coverage and new empirical calibration.
- Kept the `ipo_score_v0` score formula unchanged because the expanded sample still shows non-monotonic middle and low score buckets.
- Updated model limitations to reflect that T1 is structurally complete for listed rows, while field-level NULLs remain when source documents do not explicitly state a field.
Rationale:
- T1 structured coverage increased from 154 to 291 rows after archivist backfilled demand facts from extracted PDF text.
- The high-conviction bucket remains clearly differentiated, but the rest of the calibration is not strong enough to justify a v1 rule change.
- Avoiding a threshold rewrite here preserves the feedback loop: future rule changes should be tied to reviewed predictions and named error cases.
Verification:
- Rebuilt the analysis dataset and model report from `data/hk_ipo.sqlite`.
- Confirmed post-listing returns remain labels only and are not score inputs.
- Confirmed durable source paths remain repo-relative.
## 2026-06-15 - Introduce `ipo_score_v0`
Request: