Forecast evaluation
Four selection folds, frozen baselines, 12-level WRMSSE, and one aggregate-only official evaluation.
Evidence map
Skills, demonstrated—not listed.
Every capability below points to a frozen artifact, a receipt, or a testable control. Status is explicit where publication still depends on later release points.
01 · Capability evidence
What the work proves.
v1.0 Core · reviewer-approved
Decision analytics
Six predeclared asymmetric-cost ratios and leakage-safe quantile recalibration at bottom level before aggregation.
Inventory policy
Twenty-seven simulated policies with lost sales, empirical buffers, denominator rules, and aggregate-only exposure outputs.
FP&A and financial modeling
Forecast-to-actual PVM, linked three-statement Excel model, revolver logic, scenarios, sensitivity, and visible checks.
Power BI
Native four-page semantic model with governed measures, hierarchy drill, refresh evidence, PDF export, and contract tests.
SEC / XBRL
Canonical-period selection, concept mapping, filing lineage, derived free cash flow, ratios, and reconciliation checks.
SQL and data governance
SCD Type 2 price history, checksum gates, hierarchy controls, immutable receipts, and fail-closed lifecycles.
Macroeconomic context
Four frozen FRED observations used only as separate descriptive context, with no regime or causal claim.
Reproducibility
Pinned Python graph, portable web build, content-addressed evidence, focused tests, and explicit rerun boundaries.
02 · Release boundary
What is finished—and what is not.
- core Evidence
- Reviewer-approved
- curated Home
- Reviewer-approved
- presentation Pages
- Reviewer-approved
- last Updated
- 9 September 2026
- full Verification
- Point [56] reviewer-approved; full raw reproduction incomplete
- career Signal
- Not executed; owner-accepted omission; not PASS
- release Review
- External review pending
- github Release
- Published v1.0.0
- vercel Production
- Portfolio site deployed