Capstone — North Shore Outfitters: Close → Clean → Explain → Forecast → Decide
This capstone pulls together Track D into one realistic deliverable set. You will close, clean, explain, forecast, and recommend actions for North Shore Outfitters (NSO).
The capstone is designed to feel like a real analyst/accountant engagement:
You start with messy-but-usable operational and accounting exports.
You build a documented, analysis-ready dataset.
You explain what happened (with uncertainty and guardrails).
You produce forecasts that tie across P&L + Balance Sheet + Cash.
You end with a short decision memo that a CFO could act on.
You do not need to invent complex accounting mechanics. Track D is built so your work can be grounded in the NSO tables already produced by the simulator and the chapter scripts.
How to get the NSO case dataset
Track D uses a synthetic NSO dataset generated by the repo’s simulator. From the repo root:
make business-nso-sim
This writes the NSO v1 dataset under:
data/synthetic/nso_v1/(monthly tables, 24 months)
Optional (recommended) validation step:
make business-validate
Submission format
Submit a single folder (zip is fine) that contains your deliverables. Use filenames that make review easy, for example:
01_close_controls/02_analysis_ready_data/03_performance_analysis/04_forecast_pack/05_decision_memo/
Deliverables
1) Close & Controls Pack
Minimum contents:
Bank reconciliation summary (reconciling items categorized)
AR tie-out (AR aging or AR rollforward, with exceptions)
AP tie-out (AP aging or AP rollforward, with exceptions)
Exception report (duplicates, timing outliers, one-offs, negative balances)
Corrections log (what you changed, why, and impact)
High-value additions:
A short narrative: “What could go wrong next month if controls stay weak?”
A control owner list: who signs off on each check
2) Analysis-ready Dataset
Minimum contents:
A tidy GL table (one row per posting/transaction line)
Supporting tables you relied on (chart of accounts, customers/vendors, etc.)
A data dictionary (table name, key fields, units, definitions)
An assumptions log (what you inferred or standardized)
High-value additions:
A reproducible build note: the steps to go from raw tables to tidy tables
A data quality checklist with pass/fail and comments
3) Performance Analysis
Minimum contents:
Trend and variance analysis (month-over-month, year-over-year)
A driver narrative grounded in Track D methods (not vibes)
A short “uncertainty” section (what you can’t conclude)
Acceptable analysis styles:
Structured variance bridges (volume, price, mix, one-offs)
Regression-style driver models (with diagnostics and practical interpretation)
Guardrails:
Avoid causal overclaiming (“associated with” is safer than “caused”)
Use accountant-friendly interpretation (rates, baselines, ranges)
4) Forecast Pack
Minimum contents:
12-month rolling forecast (P&L + Balance Sheet + Cash tie-out)
13-week cash forecast (direct method)
Three scenarios (base, downside, upside) with explicit assumption changes
A simple backtest summary and error metrics (at least one forecast horizon)
Forecast hygiene expectations:
Forecasts reconcile across statements (cash tie-out is explicit)
Assumptions are written down, versioned, and owned
5) Decision Memo
Minimum contents (1–2 pages):
Recommended actions (specific, time-bounded)
Expected impact ranges (not point estimates only)
Risks and trade-offs
Monitoring plan: KPIs + thresholds + owners
Suggested build order
A practical sequence that stays close to the book:
Generate dataset:
make business-nso-simValidate and clean: Ch06–Ch07 patterns
Explain performance: Ch08, Ch16–Ch17, Ch22
Forecast:
Cash: Ch19
Integrated three statements: Ch20
Scenario + stress: Ch21
Communicate: Ch23 templates + dashboards + governance
Templates and rubric
Use the provided templates to keep your deliverables structured:
Your work is graded using the rubric: