Capstone rubric (100 points)

This rubric is meant to be transparent and practical. It rewards (1) correct reconciliations and (2) clear, decision-ready communication.

1) Controls & reconciliation quality (25 pts)

  • 25–22: Reconciliations complete, reconciling items correctly categorized, clear corrections log, strong exception checks, and a clean audit trail.

  • 21–16: Mostly correct; minor gaps, light documentation, or a few unclear tie-outs.

  • 15–0: Material errors, missing tie-outs, weak audit trail, or unclear corrections.

2) Data preparation & documentation (20 pts)

  • 20–18: Clean, tidy dataset; clear data dictionary; assumptions log; reproducible steps.

  • 17–12: Dataset usable but inconsistent naming/definitions or weak documentation.

  • 11–0: Messy data; unclear sources/definitions; not reproducible.

3) Statistical analysis & interpretation (20 pts)

  • 20–18: Correct methods; practical interpretation; avoids causal overreach; highlights uncertainty and diagnostics.

  • 17–12: Reasonable analysis but some interpretation issues, missing diagnostics, or limited linkage to business meaning.

  • 11–0: Incorrect/irrelevant methods; misleading claims; no uncertainty handling.

4) Forecast quality & forecast hygiene (25 pts)

  • 25–22: Forecast ties across statements; includes backtesting; error metrics; scenario logic; governed assumptions; clear cash tie-out.

  • 21–16: Forecast mostly sound; limited backtesting or weak scenario discipline.

  • 15–0: Forecast does not reconcile; no evaluation; unjustified assumptions.

5) Decision memo & communication (10 pts)

  • 10–9: Clear actions, quantified impact ranges, risks, monitoring KPIs, and accountable owners.

  • 8–6: Good narrative but vague actions or limited quantification.

  • 5–0: Unclear, non-actionable, or not decision-focused.

Optional appendices (high value)

These are optional, but they often improve clarity and reproducibility:

  • A: Excel templates (bank rec, KPI dictionary, forecast assumptions log, 13-week cash)

  • B: QuickBooks export guide (what to pull, how to structure)

  • C: Python quick labs (pandas cleaning, time series baselines, regression driver model)

  • D: Common accounting-data pitfalls checklist (timing, misclass, duplicates, reversals, one-offs)