Skip to main content
Back to Prompt Library
Implementation planning and evidence-linked development

Build the complete application

Inspect the original prompt language first, then copy or adapt it once you know how it fits your workflow.

Linked challenge: Build a Filing-to-Valuation Financial Statement Normalizer

Format
Text-first
Lines
51
Sections
16
Linked challenge
Build a Filing-to-Valuation Financial Statement Normalizer

Prompt source

Original prompt text with formatting preserved for inspection.

51 lines
16 sections
No variables
28 checklist items
## User
A valuation analyst preparing comparable financial histories.

## Build
Convert dated company filings into an analyst-reviewed valuation input workbook. Preserve restatements, fiscal periods, units, and the distinction between reported and adjusted measures.

## Input Contract
- Key: accession + entity + concept + context + unit.
- Fields: period_start, period_end, filed_at, value, taxonomy, adjustment_id.
- Keep instant balances separate from duration flows. Freeze a valuation-date information cutoff.

## Workflow
1. Extract reported facts with filing context.
2. Resolve fiscal periods and duplicate disclosures.
3. Record analyst-approved normalization adjustments separately.
4. Export linked statements and a valuation input workbook.

## Interface
A source-document viewer beside editable mappings, reconciliation differences, and an adjustment approval queue.

## Deliverables
- A filing-context extraction and mapping pipeline.
- A workbook linking each input to its accession and adjustment history.
- Runnable local application, setup commands, and a dependency lockfile.
- Versioned input data, data dictionary, acquisition instructions, and rights notes.
- Calculation traces, source references, assumption records, and exported review artifacts.
- A deterministic baseline and tests for normal, incomplete, contradictory, and failure-recovery cases.
- A recorded complete workflow with a reproducible manifest and explicit limitations.

## Acceptance Criteria
- Do not add overlapping year-to-date revenue periods.
- A later restatement is unavailable before its filing date.
- Adjusted measures retain a bridge to reported figures.
- Balance-sheet and cash-flow residuals remain visible.

## Contrast Case
The same restated value appears before and after the valuation cutoff. Only the later analysis may use it.

## Dataset Scope
The attached starter files are prototype inputs. Expected starter answers are visible. They are not independent holdouts.

Full target: Select six public companies with three years of filings. Add synthetic conflicting tags and restatement cases. Use filing footnotes when entity-wide facts are insufficient.

Complete target contracts and expert review remain pending. Public observations are not historical release vintages.

## Tools
Arelle, DuckDB, openpyxl

Use an available model appropriate to the task. Record provider, model identifier, version or retrieval date, settings, cost, and latency. Use deterministic code for calculations. Models may extract, retrieve, or explain evidence. They must not invent missing records, override approval rules, or claim professional authority.

## Constraints
Do not invent missing custom-tag mappings or management adjustments.

- No fabricated records, source citations, professional endorsements, or expert-adjudication claims.
- No future-data leakage into dated decisions, model fitting, or historical valuation inputs.
- No unauthorized payments, account restrictions, regulatory filings, or production changes.
- No automated appraisal certification, tax-compliance determination, or unsupported fraud accusation.
- No concealed uncertainty or silently invented inputs required for a numerical conclusion.

## Scoring
- Correct domain calculations and justified methods with documented assumptions. Weight: 30.
- Source lineage, temporal integrity, rights checks, and explicit synthetic-data labels. Weight: 25.
- A working application with review, export, error handling, and reproducible state. Weight: 20.
- Adversarial cases, uncertainty, missing-data behavior, and sensitivity analysis. Weight: 15.
- Evidence-linked conclusions that distinguish facts, assumptions, and unresolved judgments. Weight: 10.
Score each dimension from zero to two. Weighted score = sum(weight * score / 2). A failed safety gate blocks acceptance.

Adaptation plan

Keep the source stable, then change the prompt in a predictable order so the next run is easier to evaluate.

Keep stable

Preserve the source structure until you know which part of the prompt is actually driving the result quality.

Tune next

Change domain facts, examples, and tool context first before you rewrite the instruction scaffold.

Verify after

Validate one failure mode at a time so prompt changes stay attributable instead of getting noisy.