Financial Statement Analysis

A structured methodology for fundamental analysis of a company's financial statements. Given a 10-K/10-Q, an income statement, a balance sheet, a cash-flow statement, or a handful of key numbers, this skill produces a disciplined read of liquidity, solvency, profitability, efficiency, an ROE decomposition (DuPont), distress and earnings-manipulation scores (Altman Z, Beneish M), a red-flag scan, and a one-paragraph health verdict — all grounded in standard, real methodology with sector-aware benchmarks.

Not investment advice. This is an educational analytical framework. It does not account for forward guidance, qualitative moats, management quality, macro conditions, or your risk tolerance. Always verify numbers against the audited source and consult a licensed professional before acting.

When to Activate

Activate this skill when the user:

• Shares a company's financials (10-K, 10-Q, annual report, income statement, balance sheet, cash-flow statement) and asks whether the company is healthy, improving, risky, or worth investing in.
• Asks for ratio analysis, fundamental analysis, due diligence, credit review, or investment screening on a named company or pasted figures.
• Wants to compare a company against peers or sector norms.
• Asks "are these numbers a red flag?", "is the earnings quality good?", "what is the Altman Z-score / Beneish M-score?", or "is this company likely to go bankrupt / manipulating earnings?".

If only a single number is given (e.g. just revenue), ask for the minimum line items needed (see Step 2) before computing ratios. Do not fabricate inputs.

Step 1: Scope & Context

Before any math, establish and restate the context so the analysis is anchored:

State each explicitly in the output header. If the sector is unknown, ask — benchmarks are meaningless without it.

Step 2: Data Extraction

Pull the specific line items the ratios need. Flag anything missing or estimated, and note any restatements.

Income statement: Revenue (net sales), COGS, Gross profit, Operating expenses (SG&A, R&D), Operating income (EBIT), Interest expense, Pre-tax income, Income tax, Net income; plus Depreciation & amortization (for EBITDA — often only on the cash-flow statement).

Balance sheet: Cash & equivalents, Short-term investments, Accounts receivable, Inventory, Total current assets; Total assets; PP&E (net), Goodwill & intangibles; Accounts payable, Short-term debt / current portion of long-term debt, Total current liabilities; Long-term debt, Total liabilities; Total shareholders' equity; retained earnings; shares outstanding.

Cash-flow statement: Cash flow from operations (CFO), Capital expenditures (capex), Cash flow from investing, Cash flow from financing, dividends paid.

Rules:

• Use averages for balance-sheet items in ratios that mix flow and stock. ROA, ROE, asset turnover, inventory days, DSO, DPO all divide a flow (a full period of income/sales/COGS) by a balance — use the average of beginning and ending balance, e.g. (opening + closing) / 2, when both are available. Note when you only had a closing balance.
• EBITDA = EBIT + D&A. EBIT = Operating income. If "operating income" bundles unusual items, normalize and say so.
• Flag estimates. If a number is derived or assumed, mark it (est.).
• Note restatements / one-offs. Prior-period restatements, impairments, litigation charges, and discontinued operations distort trend; call them out.
• Watch sign conventions. Capex and dividends are usually shown negative on the cash-flow statement; FCF = CFO − capex (capex as a positive magnitude).

Step 3: Ratio Analysis

Compute the four groups below. For each ratio: the formula, what it means, and a benchmark band. Benchmark bands are general — they shift by sector, so the "sector note" column flags where the default lies.

Convention: Healthy = typically comfortable; Watch = monitor / trend matters; Concern = often a problem absent a sector-specific reason. These are heuristics, not hard rules — context overrides the table.

3a. Liquidity — can it pay near-term bills?

3b. Solvency / Leverage — can it survive its debt?

3c. Profitability — does it make money, and how well?

3d. Efficiency — how hard do assets and working capital work?

When a ratio mixes an income-statement flow with a balance-sheet stock, use the average balance over the period (Step 2).

Step 4: DuPont + Distress / Manipulation Scores

4a. DuPont decomposition of ROE

3-step: ROE = Net margin × Asset turnover × Equity multiplier

ROE = (Net income / Revenue) × (Revenue / Avg assets) × (Avg assets / Avg equity)
        └ profitability ┘      └ efficiency ┘            └ leverage ┘

5-step (separates tax and interest drag):

ROE = Tax burden × Interest burden × Operating margin × Asset turnover × Equity multiplier
    = (NI/EBT)  × (EBT/EBIT)      × (EBIT/Revenue)    × (Revenue/Avg assets) × (Avg assets/Avg equity)

Why: DuPont shows where ROE comes from. A 20% ROE built on a 5× equity multiplier (heavy leverage) is far riskier than the same 20% built on strong margins. Always check whether ROE is "earned" (margins/efficiency) or "borrowed" (leverage).

4b. Altman Z-score — distress / bankruptcy risk

Original (manufacturing / public):

Z = 1.2·X1 + 1.4·X2 + 3.3·X3 + 0.6·X4 + 1.0·X5

X1 = Working capital / Total assets
X2 = Retained earnings / Total assets
X3 = EBIT / Total assets
X4 = Market value of equity / Total liabilities
X5 = Revenue / Total assets

Zones: Safe > 2.99 · Grey 1.81 – 2.99 · Distress < 1.81

Z″ variant — non-manufacturers, service firms, private, and emerging markets (drops X5 sales-turnover term; uses book equity in X4):

Z'' = 6.56·X1 + 3.26·X2 + 6.72·X3 + 1.05·X4

X4 = Book value of equity / Total liabilities

Z″ zones: Safe > 2.60 · Grey 1.10 – 2.60 · Distress < 1.10

Use Z″ for SaaS, retail, services, banks-adjacent, private companies, or anywhere a market cap isn't available. State which variant you used and why.

4c. Beneish M-score — earnings-manipulation likelihood

Probabilistic model flagging the likelihood that earnings have been manipulated. Needs two consecutive years (t and t−1).

M = -4.84 + 0.920·DSRI + 0.528·GMI + 0.404·AQI + 0.892·SGI
        + 0.115·DEPI − 0.172·SGAI + 4.679·TATA − 0.327·LVGI

8 components:

Threshold: M > −1.78 ⇒ likely manipulator (more negative = cleaner). TATA (accruals) and DSRI usually drive the score; a high M almost always coincides with net income running well ahead of operating cash flow.

The M-score is a screen, not a verdict — it flags statistical similarity to known manipulators and produces false positives (especially for fast-growing firms). Treat a trip as "look harder", not "fraud".

Step 5: Red-Flags Quick Scan

Scan for these classic deterioration / quality signals. Each is a prompt to investigate, not an automatic condemnation:

• Revenue up but CFO down / flat — earnings not converting to cash.
• Rising DSO or inventory days vs sales — channel stuffing, demand softening, collection problems, or obsolescence.
• Widening gap between net income and operating cash flow — accruals-heavy earnings; low earnings quality (see accruals ratio in Step 6).
• Ballooning goodwill / intangibles — acquisition-driven growth; impairment risk if acquired businesses underperform.
• Debt maturities > cash + expected CFO — refinancing/liquidity wall; check the maturity schedule.
• Frequent "one-time" / restructuring charges — recurring "non-recurring" items inflate adjusted earnings vs GAAP.
• Negative & worsening working-capital trend — unless the business model is structurally cash-upfront (verify).
• Aggressive revenue recognition — bill-and-hold, long-dated contracts pulled forward, bundled deals, rising "unbilled receivables".
• Margin expansion with no operational explanation, gross margin diverging from peers, or capitalizing costs that peers expense.
• Rising leverage while profitability falls, interest coverage trending toward < 2×, or dividends/buybacks funded by debt.

Step 6: Output Format

Produce the report in this order:

1. Header

Company · sector · period(s) · currency/units · audited? · statements provided · variant choices (which Altman Z used and why).

2. Scorecard table

One row per key metric; columns: Metric · Value · vs Benchmark (Healthy/Watch/Concern) · Trend (use ↑ ↓ → arrows across periods).

3. DuPont readout

3-step (and 5-step if data allows). State whether ROE is margin-driven, efficiency-driven, or leverage-driven.

4. Distress & manipulation scores

• Altman Z (variant) = value → zone with a plain-English read.
• Beneish M = value → likely / unlikely manipulator, naming the top 1–2 components driving it.

5. Cash-flow quality

• CFO vs Net income (CFO/NI ratio; healthy ≈ ≥ 1.0).
• Accruals ratio = (Net income − CFO) ÷ Avg total assets. High positive = lower earnings quality.
• Free cash flow = CFO − Capex; note FCF margin and whether it's positive and growing.

6. Prioritized red-flags list

Most material first, each with the evidence (the numbers) behind it.

7. Health verdict (one paragraph)

Conclude with exactly one rating and a one-paragraph justification:

• Strong — healthy across liquidity, solvency, profitability; Z in Safe zone; clean M; CFO ≥ NI; no material red flags.
• Adequate — generally sound with one or two watch items; Z in Safe/upper Grey; manageable leverage.
• Weak — multiple watch/concern ratios, Z in Grey, thinning margins or rising leverage, or earnings-quality concerns.
• Distressed — Z in Distress zone, interest coverage < 1.5×, liquidity crunch, and/or M-score flag with corroborating red flags.

Disclaimer

This skill provides educational financial analysis, not investment, legal, or accounting advice. Ratios and scores are heuristics that require sector and qualitative context; the Altman Z and Beneish M models are screens with known false positives and were calibrated on historical datasets that may not fit a given firm. Always verify every figure against the audited primary source (the filed 10-K/10-Q), consider forward-looking and qualitative factors the statements don't capture, and consult a licensed financial professional before making any investment or credit decision. This skill is not affiliated with or endorsed by Anthropic or any data provider.