The same line - "Net Income: $2 billion" - means something completely different at a tech company, at a bank, at an oil major, at a restaurant chain. Beginners make one recurring mistake: reading the number without reading the sector context. This piece starts from the fundamentals - the 3 core statements, how to find filings on the SEC - then moves to the key metrics for each sector: Tech, Banking, Energy, Consumer Staples, Consumer Discretionary, F&B, Healthcare. Including a special case: companies like McDonald's that have negative shareholders' equity and are still perfectly healthy. It closes with 12 common mistakes and how to avoid them.
Scope: This article focuses on US-listed companies (SEC filings). Illustrative figures are taken from public 10-K reports. Formulas and sector benchmarks are for reference only - always cross-check against primary sources.
Note: This is not investment advice. It's an educational resource to help beginners understand the structure of financial statements and how to read them. Any investment decision needs further, deeper analysis.
Part I - The Three Core Financial Statements
Every US-listed company must file reports with the SEC (Securities and Exchange Commission). The two most important filing types:
Every filing is on SEC EDGAR (edgar.sec.gov). Type the company name or ticker, pick the filing type. Completely free, no account needed.
Inside every 10-K/10-Q, the three core financial statements tell you the same story from three different angles:
1. Income Statement
This is the statement most beginners read first - and also the one most prone to causing confusion. It shows how much revenue the company earned, how much it spent on operations, and how much profit is left over. But the "profit" here is accounting profit, not cash in the vault.
The key lines, top to bottom:
- Revenue - total money from selling goods/services. Sometimes split into "product revenue" vs "service revenue"
- Cost of Revenue / COGS - the direct cost of producing the goods/services
- Gross Profit = Revenue - COGS - gross profit, reflecting pricing power
- Operating Expenses (OpEx) - R&D, SG&A (sales, general & administrative), depreciation
- Operating Income (EBIT) - profit from the core business
- Net Income - after-tax profit, after everything - the "bottom line"
Revenue ≠ cash received. Under accrual accounting (US GAAP), revenue is recognized when earned, not when cash hits the account. A company can report $100M of revenue while not having collected a single dollar of it - the uncollected portion sits in Accounts Receivable on the Balance Sheet.
2. Balance Sheet
If the Income Statement is the "movie", the Balance Sheet is the "photograph" at one point in time. The invariant formula:
It splits into two main sides:
- Assets: Current assets (cash, accounts receivable, inventory - convertible to cash within 1 year) vs Non-current assets (PP&E, goodwill, intangibles)
- Liabilities: Current liabilities (accounts payable, short-term debt - due within 1 year) vs Long-term liabilities (bonds, long-term debt, lease obligations)
- Shareholders' Equity: the remainder belonging to shareholders = common stock + retained earnings + accumulated other comprehensive income (AOCI)
Goodwill is the premium paid when acquiring another company above the fair value of its net assets. It sits on the Balance Sheet as an "asset" but can't be sold and can't be used to pay down debt. When a company writes down goodwill, that's usually an admission it "overpaid".
3. Cash Flow Statement
This is the statement professional investors respect most - because it's harder to "dress up" than the Income Statement. Cash either exists or it doesn't - there's no room for "accounting estimates".
Three sections:
- Operating Activities (CFO): Cash from the core business. Starts from Net Income, then adds back non-cash items (depreciation, SBC) and adjusts for working-capital changes. This is the most important line.
- Investing Activities (CFI): Cash for CapEx (buying machinery, building plants), acquiring other companies, or selling assets. Usually negative at growing companies.
- Financing Activities (CFF): Cash from issuing stock, borrowing, repaying debt, buying back stock, paying dividends.
FCF tells you how much real cash the company has left after sustaining operations and necessary investment. This is the money available to pay down debt, buy back stock, pay dividends, or M&A.
Part II - Structure Of A 10-K: What To Read First?
A 10-K usually runs 100-300 pages. Nobody reads all of it. The priority order for beginners:
Item 1A: Risk Factors - Read This First
Counterintuitively, this is the section you should read before the financial section. Risk Factors is where the company itself admits what risks it carries: legal, competitive, dependence on large customers, regulatory risk, concentration risk.
Tip: Compare Risk Factors between two consecutive years. Newly added risks usually matter more than risks repeated verbatim.
Item 7: MD&A (Management Discussion & Analysis)
This is where management explains the financial numbers. Revenue up 15% - from price or from volume? Operating margin down - from R&D investment or rising input costs?
MD&A is the only section where management speaks in plain language about business results. Read the tone closely: optimistic or defensive? Where does hedging language ("may", "could", "potentially") cluster?
Item 8: Financial Statements & Footnotes
The three core statements (Income Statement, Balance Sheet, Cash Flow) live here, along with footnotes - the section most often skipped but holding the most important information.
Footnotes reveal: revenue recognition policy, how depreciation is calculated, debt detail (interest rate, maturity date), stock-based compensation, legal contingencies, operating lease obligations, segment breakdown.
Rule of thumb: If a number on the main statement makes you wonder, the answer almost certainly lives in the footnotes.
Item 1: Business Description & Item 2: Properties
Item 1 describes the business model, main products/services, markets, competition, and seasonal patterns. Especially useful when you don't yet understand what the company does.
Item 2 (Properties) tells you whether the company owns or leases its physical facilities - factories, data centers, offices. Less exciting but revealing of capital intensity.
Part III - Key Metrics By Sector
This is the core of the piece. The same metric - P/E, gross margin, debt/equity - means something completely different depending on the sector. A tech company with a P/E of 40 might be cheap; a bank with a P/E of 40 is almost certainly expensive. And there are companies like McDonald's, where ROE, P/B, D/E - the three most textbook-familiar ratios - are completely meaningless.
1. Technology / SaaS
Representative names: Microsoft, Salesforce, CrowdStrike, Snowflake, Palantir
Revenue Growth & ARR
Annual Recurring Revenue (ARR) - recurring annual revenue - is SaaS's most important yardstick. It tells you the "base revenue" a company can almost certainly count on next year, as long as customers don't churn.
Net Revenue Retention (NRR) above 120% means existing customers are spending 20% more than the year before - revenue still grows even without a single new customer.
Rule of 40 & Margins
Rule of 40: Revenue Growth (%) + FCF Margin (%) ≥ 40%. This is the benchmark balancing growth against profitability. A company growing 50%/year can lose 10% and still "pass".
Gross Margin: SaaS typically runs 70-85%. Below 60% raises the question: is this really software, or is there a lot of service/hardware mixed in?
Many tech companies report "non-GAAP" profits by excluding Stock-Based Compensation (SBC) - paying employees in stock. Average SBC at cloud SaaS companies is 21% of revenue (versus 2% for the S&P 500). When SBC equals 39% of FCF, the "Free Cash Flow" figure is effectively inflated because it hasn't accounted for personnel cost paid in stock. Always check GAAP earnings too for the real picture.
2. Banking / Finance
Representative names: JPMorgan Chase, Bank of America, Goldman Sachs, Wells Fargo
Banks don't operate like ordinary companies. A bank's revenue is mainly the spread on interest rates (lending at a higher rate than what it pays to borrow), not selling a product. The Balance Sheet matters more than the Income Statement. And a bank's "inventory" is... other people's money.
Net Interest Margin (NIM)
NIM = (Interest Income - Interest Expense) / Avg Earning Assets. Usually 2-5%. NIM above 3% is healthy for a commercial bank. NIM compresses when the Fed cuts rates or when the yield curve flattens/inverts.
Non-Interest Income matters too: transaction fees, asset management, trading revenue. Investment banks (Goldman) depend on this line much more heavily.
CET1 Ratio & Efficiency
CET1 = Common Equity Tier 1 / Risk-Weighted Assets. Above 11% is a "war chest", below 8% is the danger zone where regulators step in.
Efficiency Ratio = Non-interest Expense / Revenue. Lower is better (unlike most other ratios). Below 55% is excellent, above 65% is poor.
Provision for Credit Losses (PCL) - the reserve set aside for bad loans. This is the largest "estimated" expense line on a bank's Income Statement. When PCL spikes, it means the bank is forecasting that borrowers won't repay. A sharp drop in PCL is also suspicious - management may be "releasing" reserves to flatter earnings.
3. Energy (Oil & Gas)
Representative names: ExxonMobil, Chevron, ConocoPhillips, Pioneer (now part of Exxon)
The oil & gas sector has its own quirk: profit depends almost entirely on commodity prices the company doesn't control. The Income Statement swings hard with the cycle. So analysts use a distinct set of metrics.
Reserve Replacement & EBITDAX
Reserve Replacement Ratio (RRR) = Proved Reserves Added / Production. Above 100% means the company is finding new oil faster than it's producing - healthy for the long run. Consistently below 100% means "eating into natural capital".
EBITDAX = EBITDA + Exploration Expenses. The sector-standard metric because exploration cost is huge and volatile. A Finding & Development cost (F&D) under $10/BOE is top quartile.
DD&A, Breakeven & FCF Yield
DD&A (Depreciation, Depletion & Amortization) is huge in oil & gas because fixed assets (rigs, pipelines, refineries) are extremely capital-intensive. The Income Statement can be misleading because of DD&A - look at Cash Flow instead.
Breakeven Price - the minimum oil price for the company to break even. US shale: $40-60/bbl. Middle East: $10-20/bbl. When oil trades near breakeven, a stress test is essential.
Upstream (E&P) is the most sensitive to commodity prices - large profits when oil is high, heavy losses when oil is low. Midstream (pipelines, storage) behaves like a utility - steady revenue from transport fees, less dependent on oil price. Downstream (refining, distribution) depends on the crack spread (the gap between crude oil price and refined product price).
4. Consumer Staples
Representative names: Procter & Gamble, Coca-Cola, Walmart, Costco, Colgate-Palmolive
The "most boring" sector on the surface, but the best test of your financial-statement-reading skill. Low growth (2-6%/year), stable margins, low volatility. Reading this sector means hunting for changes that are small but meaningful.
Gross Margin & Organic Growth
Gross Margin - a measure of pricing power. P&G holds ~48-50%, Coca-Cola ~60%. When gross margin falls while revenue still rises = the company is selling more volume but getting squeezed on price or facing rising input costs.
Organic Revenue Growth - separates growth from price vs volume vs mix. Growth driven purely by price while volume falls = customers are starting to push back.
Inventory Turnover & FCF Conversion
Inventory Turnover - sector average ~7.7x. Walmart runs higher (due to volume, thin margins), luxury brands run lower. Falling turnover = inventory is piling up.
FCF Conversion = FCF / Net Income. Good staples companies usually exceed 90%. If Net Income looks great but FCF is weak, check working capital: are receivables growing abnormally? Is inventory bloating? Are payables being stretched out?
Large FMCG conglomerates (P&G, Unilever, Coca-Cola) earn 50-60% of revenue outside the US. A strong USD = translated revenue falls even though local business is unchanged. Always read the "constant currency" section in the MD&A to see the real picture before FX effects.
5. Consumer Discretionary
Representative names: Amazon (retail), Tesla, Nike, Starbucks, Home Depot, McDonald's
Unlike Staples (bought out of need), Discretionary sells things people buy when they have money to spare. So this sector is much more sensitive to the economic cycle, consumer confidence, and interest rates.
Same-Store Sales (Comp Sales)
Same-Store Sales (SSS) - revenue growth at stores open at least 12 months. This is the single most important metric, because it strips out the effect of opening new stores.
SSS = Traffic × Average Ticket. Split the two apart: if SSS is rising because of higher prices (ticket up) while traffic is falling, that's a sign demand is actually weakening.
Unit Economics & Store Margin
Restaurant/Store-Level Margin - profit at the store level before corporate overhead. Starbucks US: ~18-20%. McDonald's (franchise-heavy): 80%+ because the franchisor just collects fees.
New Store Productivity - new-store revenue compared to existing stores. If productivity falls with each new wave of openings = the market is saturating.
McDonald's owns few restaurants but owns the land under franchised locations. Revenue is mostly franchise fees plus rent. Gross margin is extremely high, but comparing it directly to Starbucks (mostly company-owned) is entirely wrong. Always read the Business Model section before comparing metrics.
6. F&B (Food & Beverage Manufacturing)
Representative names: PepsiCo, Mondelez, General Mills, Kraft Heinz, Tyson Foods
Food manufacturing sits at the intersection of Consumer Staples and commodity risk. Revenue is stable (people keep eating through a recession) but margin gets squeezed by the price of agricultural inputs, energy, and packaging.
Input Cost Exposure & Hedging
Commodity Exposure - the price of wheat, sugar, palm oil, milk feeds directly into COGS. Check the footnotes: what percentage does the company hedge, does it use derivatives?
Gross-to-Operating Margin Spread - the gap between gross margin and operating margin shows how large marketing/advertising spend is. F&B spends 5-10% of revenue on advertising.
Portfolio Mix & Innovation
Innovation Revenue % - revenue from products launched under 3 years ago. Sector average is 10-15%. PepsiCo often leads thanks to its "better-for-you" portfolio.
Price/Mix vs Volume - similar to Staples but matters more because F&B has higher elasticity. Pushing price too hard = losing share to private label.
Kraft Heinz wrote down $15.4B of goodwill in 2019 - nearly half its market cap. That was the result of an eye-wateringly priced 2015 merger. When reading an F&B company post-M&A, always compare goodwill to shareholders' equity. If goodwill exceeds equity, the company is "standing on thin air".
7. Healthcare (Pharma, Biotech, Medical Devices)
Representative names: Johnson & Johnson, Pfizer, Eli Lilly, Amgen, UnitedHealth Group, Intuitive Surgical
Healthcare is the most idiosyncratic sector for reading financial statements. A pharma company can lose money for 10 straight years and still be worth billions - because the value sits in the pipeline (drugs in development), not current revenue. Conversely, a "big pharma" posting enormous profits today may be sitting on a time bomb called the patent cliff.
Patent Cliff & Revenue Concentration
Patent Cliff - when a patent expires, generics flood in, revenue collapses. Humira (AbbVie): from $21.2B (2022) down to $9B (2024) after losing exclusivity. Pfizer's Lipitor fell 71% in one year. The 2025-2030 window sees roughly 200 drugs lose patent protection, about 70 of them blockbusters.
Revenue Concentration - what percentage of revenue comes from the top 1-3 drugs? If more than 40% comes from one product about to lose patent protection = existential risk. Look in the 10-K's "Product Revenue by Product" section.
R&D Intensity & Phase Success Rates
R&D/Revenue - Big pharma: 15-25%. Pre-revenue biotech: 100%+ (all expense, no revenue). A sudden drop in R&D at a large pharma company = cutting future investment, a red flag.
Pipeline Value - assessed via rNPV (risk-adjusted NPV): each drug candidate is priced separately by phase. Phase 1: ~10% chance of success. Phase 3: ~50-60%. Approved: near certain. A biotech with one Phase 3 candidate is worth something entirely different than a biotech with one Phase 1 candidate.
Payer Mix & Reimbursement
Payer Mix - is revenue coming from Medicare, Medicaid, private insurance, or self-pay? Medicare/Medicaid pay less than private insurers. If payer mix shifts toward government payers = margin gets squeezed.
Days in AR - how long it takes to collect payment from insurers/hospitals. Outpatient: 30-45 days. Hospital: 45-60 days. Rising = collection trouble or a high denial rate.
Recurring Revenue % - medical device makers (Intuitive Surgical, Abbott) run a razor/blade model: sell the machine cheap, make money on consumables/service. Recurring revenue above 60% = stable.
Medical Loss Ratio & Enrollment
MLR (Medical Loss Ratio) = Claims Paid / Premiums Collected. The ACA requires MLR ≥ 80-85%. Low MLR = high profit but regulatory risk. Too-high MLR = medical costs aren't under control.
Membership/Enrollment Growth - UnitedHealth, Cigna, Humana measure growth by member count. Falling enrollment plus rising premiums = losing market share.
Low net margin but high ROE - not a paradox. Health insurers often carry a net margin of just 3-5% (UNH ~6%, Cigna ~3%), looking "thinner" than even Walmart. But ROE runs 15-25%+ because of the same mechanism as Costco/Walmart: asset-light + leverage + enormous volume. They don't need factories or inventory - they just collect premiums, pay claims, and keep the spread. Shareholders' equity is small (thanks to continuous buybacks plus low capex needs), so even though net profit per dollar of revenue is low, net profit per dollar of equity is very high. The DuPont formula explains it clearly: ROE = Net Margin × Asset Turnover × Equity Multiplier. Low margin is offset by high turnover (large revenue relative to total assets) and high leverage (large assets relative to equity). Don't write off UNH or Elevance just because margin sits at 4% - look at ROE and cash flow generation instead.
You cannot use a normal P/E or DCF for pre-revenue biotech. A company with no revenue, losing $200M/year, but holding a Phase 3 drug for a rare disease, can be worth $5B+ if the FDA approves it. The correct method is Sum-of-the-Parts (SOTP): value each drug candidate separately, risk-adjust by phase, add it up plus cash on hand minus debt. Don't look at negative Net Income and conclude "this company is bad".
8. A Special Case: Negative Shareholders' Equity
Representative names: McDonald's, Starbucks, Yum! Brands, Philip Morris, Domino's, AutoZone
Open McDonald's Balance Sheet and you'll see something strange: Shareholders' Equity of about negative $1.8 billion (2025). By the textbook, negative equity means bankruptcy. But McDonald's is perfectly healthy, has paid a steady dividend for nearly 50 years, and runs a 47% operating margin. What's going on?
Buybacks + Debt = Equity Disappears
These companies generate enormous, stable cash flow from franchise fees, rent, or addictive products (tobacco). They use that cash flow plus additional borrowing to buy back stock and pay dividends, year after year, with total payouts far exceeding cumulative earnings.
McDonald's has bought back stock and paid dividends worth hundreds of billions of dollars over two decades. Retained Earnings has been "drained" and gone negative. This is deliberate financial engineering, not a sign of distress.
ROE, P/B, D/E Become Meaningless
ROE = Net Income / Equity. When equity is negative, ROE is also negative - not because the company is losing money (Net Income is positive!), but because the denominator is negative. A ROE of -150% at McDonald's doesn't mean anything.
P/B = Price / Book Value. Negative book value = negative or undefined P/B. Useless.
D/E = Debt / Equity. Negative equity = negative or infinite D/E. Also useless.
The three most textbook-familiar ratios simply do not apply to this group of companies.
Instead of ROE, use ROIC (Return on Invested Capital) or ROA (Return on Assets). McDonald's ROIC runs ~25-30%, ROA ~14-16% - double Starbucks. Instead of D/E, use Net Debt / EBITDA to assess debt-servicing capacity (McDonald's runs ~3-4x, entirely manageable). Instead of P/B, use EV/EBITDA or FCF Yield.
When Is Negative Equity OK vs Dangerous?
OK when: Operating cash flow is strong and stable. Interest coverage ratio is high (above 5x). The business model is capital-light (franchise, licensing, subscription). Cash flow comfortably services debt and maintains the dividend. Investment-grade credit rating.
Dangerous when: Equity is negative because of accumulated losses (retained earnings negative from real losses, not from buybacks). Operating cash flow is weak or negative. Debt is rising to plug operating losses, not to return capital. Credit rating has been downgraded.
Quick way to tell them apart: look at Treasury Stock on the Balance Sheet. If Treasury Stock is the largest item dragging equity negative = buyback-driven, usually safe. If there's no significant Treasury Stock yet equity is still negative = real losses.
McDonald's Is Really A Real Estate Company
95% of McDonald's 41,822 restaurants are franchised. McDonald's owns the land and leases it to franchisees: collecting roughly 10% of revenue as rent plus a 4% franchise fee. Revenue is mostly rent + royalty, not burger sales.
The result: operating margin above 45%, low CapEx (franchisees fund their own fit-out and staffing), extremely predictable cash flow. This is why McDonald's can carry heavy leverage and still be safe - cash flow is stable enough to service the debt.
Starbucks is the opposite: mostly company-owned (though licensed stores are growing). Higher CapEx, lower margin, but also negative equity from a similar buyback strategy. Higher risk than McDonald's because of weaker operating leverage.
When you come across a company with negative shareholders' equity, run through 4 questions:
1) Is equity negative because of Treasury Stock (buybacks) or negative Retained
Earnings (losses)?
2) Is Operating Cash Flow positive and growing?
3) What is Net Debt / EBITDA? Below 4x is usually OK.
4) What is Interest Coverage (EBIT / Interest Expense)? Above 5x is comfortable.
9. A Special Case: "Cash" That Isn't The Company's
Representative names: Interactive Brokers ($IBKR), Robinhood ($HOOD), Charles Schwab, Coinbase
Open Interactive Brokers' Balance Sheet: Total Assets ~$200 billion (Q4 2025). Market cap is only ~$80 billion. P/B looks incredibly low, "cash and securities" look enormous. At a glance, it looks like the company is sitting on a mountain of cash. But in reality: most of the "assets" on the balance sheet are cash and securities belonging to customers, not the company.
Customer Money Is Both An Asset And A Liability
When a customer deposits $100,000 into an IBKR account, that money appears simultaneously in two places on the balance sheet: Assets (Cash & Securities - Segregated) and Liabilities (Customer Credit Balances / Payable to Customers).
IBKR has ~$144B in customer credit balances. This money does not belong to IBKR - the company only holds it under SEC Rule 15c3-3 (the Customer Protection Rule). It must be segregated and cannot be used for the company's own operations.
The result: real equity is only about $16.6B against $200B total assets. An equity/assets ratio of ~8% - but this isn't bank-style dangerous leverage, because most of the assets and liabilities are simply a pass-through of customer money.
P/C, P/B, ROA - All Wrong
Price-to-Cash looks extremely low - but that "cash" belongs to customers, matched by an equivalent liability. It can't be used for dividends, buybacks, or M&A.
P/B looks low if calculated on total assets - but if calculated on real equity ($16.6B) against market cap (~$80B), P/B is actually ~4.8x - not cheap at all.
ROA looks extremely low (because the denominator is inflated by customer assets) - but ROE above 20% is very strong. This is why ROE is the correct metric for a broker-dealer, not ROA.
Instead of P/B on total assets, use P/E or P/B on tangible equity (stripping out customer assets/liabilities). Instead of ROA, use ROE or Pre-tax Margin. Assess growth via DARTs (Daily Average Revenue Trades), Client Accounts, Customer Equity (total customer deposits - a real measure of "size"), and Net Interest Income (IBKR earns interest on customers' idle cash - similar to a bank's NIM).
Other sectors that mislead in similar ways
Broker-dealers aren't the only case. Many sectors have a balance sheet "bloated" by holding assets/cash that don't belong to them, or have an accounting structure that renders ordinary metrics meaningless:
Float: Money "Borrowed" From Customers
Insurance Float = premiums collected minus claims not yet paid. Berkshire Hathaway has ~$171B of float (2024). This money sits on the balance sheet as investment assets, but will have to be paid out to policyholders when claims come in.
Trap: Enormous total assets, a huge investment portfolio - but that's money "borrowed" from policyholders. A low P/B doesn't mean cheap. Combined Ratio (loss ratio + expense ratio) is the core metric: under 100% = underwriting profit (collecting more in premiums than paid out in claims). But combined ratio ignores investment income - which makes up 30-50% of real profit.
Representative names: Berkshire, Progressive, Allstate, AIG
Net Income Is Meaningless, FFO Is The Real Number
Real estate usually appreciates over time, but GAAP requires depreciating real estate every year. The result: Net Income is artificially suppressed by depreciation that doesn't reflect reality.
FFO (Funds From Operations) = Net Income + Depreciation/Amortization - Gains on Sales. This is the correct "EPS" for REITs. P/E is meaningless for REITs - use P/FFO or P/AFFO instead.
Representative names: Realty Income, Prologis, American Tower, Digital Realty
Deposits = A Special Kind Of "Debt"
Similar to broker-dealers: customer deposits are liabilities on a bank's balance sheet, but also the "raw material" for lending. $3.4 trillion in total assets (JPMorgan) sounds massive - but most of it is loans (funded by deposits) plus securities. Equity is only ~$340B (~10%).
A very high D/E ratio (~9-10x) is normal for a bank. Applying an ordinary company's D/E standard (under 2x) to a bank leads to the wrong conclusion.
The right metrics: CET1, NIM, Efficiency Ratio (covered in the Banking section)
Regulated Returns & Rate Base
Utilities are regulated: the government permits a fixed rate of return (usually 9-11% ROE) on the rate base (total infrastructure investment). EPS looks low, P/E looks high - but earnings are extremely stable and predictable.
Trap: Payout ratio is usually above 70-80% (because earnings are suppressed by heavy depreciation). Using an ordinary payout ratio would make you think the dividend isn't sustainable. More correct: use FCF payout or AFFO payout.
Representative names: NextEra, Duke Energy, Southern Company, Enterprise Products
Customer Crypto Assets On The Balance Sheet
After SAB 121 (2022), listed crypto exchanges (Coinbase) had to record customer crypto assets and a corresponding liability on the balance sheet. Total assets balloon, but don't reflect the company's actual assets.
Similar to IBKR: real revenue comes from transaction fees, staking revenue, and interest on custodial cash - not from the crypto assets sitting on the balance sheet.
Representative names: Coinbase, Crypto.com (pre-IPO)
The General Rule
When a company's balance sheet looks "too cheap" (low P/B, mountain of cash), ask:
1) Is this "cash" money the company truly owns and can use freely?
2) Do the assets include holdings on behalf of customers (customer deposits, segregated cash, custodial assets)?
3) Is there a matching liability on the balance sheet?
If the answer to (2) or (3) is "yes" - strip out the customer assets/liabilities before calculating any ratio.
10. Foreign Companies On US Exchanges: ADRs And The "China Discount"
Not every stock on the NYSE/Nasdaq is a US company. Hundreds of foreign companies trade in the US through an ADR (American Depositary Receipt) - a certificate issued by a US bank representing shares of the underlying foreign stock. Understanding this structure is essential before reading their financial statements.
What is an ADR?
An ADR works like this: a depositary bank (usually JPMorgan, BNY Mellon, or Citibank) buys the underlying shares on a foreign exchange, holds them in custody, and issues "receipts" (certificates) that trade on a US exchange. Each ADR can represent 1 underlying share, or 10, or 0.1 - depending on the ADR ratio.
Unsponsored ADRs are created by a broker/dealer without any cooperation from the foreign company. The company hasn't signed an agreement, has no reporting obligation, and provides no information to US investors. There can be multiple different unsponsored ADRs for the same company.
Trades OTC only. No voting rights. No SEC reporting obligation. Financial information may not be in English or may not follow US GAAP/IFRS. Low liquidity, wide spreads. If you're a beginner, avoid these entirely.
Why Chinese stocks look "cheap" - and why they really aren't
Alibaba trades at a P/E of ~11x. JD.com ~9x. Baidu ~10x. Compared to Amazon (~60x), Google (~25x), Meta (~25x) - these look "incredibly cheap". But a low P/E doesn't mean cheap. It reflects risk the market is pricing in, and that risk is very real.
You Don't Own The Company - You Own A Shell
China bans foreigners from directly owning companies in many sectors (internet, media, telecom). To IPO in the US, Chinese companies use a VIE (Variable Interest Entity) structure:
When you buy $BABA, you're buying shares of Alibaba Group Holding Limited - a shell company in the Cayman Islands. This Cayman entity has "contractual arrangements" with the real operating company in China. But you do not own equity in the real Chinese company. No voting rights. No claim on assets in bankruptcy. 83% of Chinese VIEs are registered in the Cayman Islands or British Virgin Islands.
The biggest risk: the Chinese government could declare the VIE contracts illegal at any time - and foreign investors lose everything.
Government Intervention - Unpredictable
The Ant Group case (2020): Jack Ma criticized China's financial system at a conference. Two days later, the government cancelled Ant Group's $34.5 billion IPO - what would have been the largest IPO in history. Ma "disappeared" for 3 months. Ma's voting stake in Ant was cut from over 50% to 6.2% as required. Alibaba lost ~$400B in market cap.
The Tech Crackdown case (2021): Beijing suddenly tightened rules on edtech (the entire sector lost 90%+), gaming (playtime limits), ride-hailing (Didi was forced to delist from the US). No timeline, no compensation, no negotiation.
When the government is the risk - no financial statement can protect you.
Auditing & The HFCAA
HFCAA (Holding Foreign Companies Accountable Act): a US law allowing the SEC to delist a foreign company if the PCAOB isn't permitted to inspect its auditor for 3 consecutive years. 150+ Chinese companies once faced delisting risk.
In 2022, the PCAOB reached an agreement to inspect auditors in China - temporarily resolving the issue. But that agreement can be pulled back at any time by either side. The SEC currently lists 286 Chinese companies on US exchanges.
Before 2022, the PCAOB had never fully inspected audit workpapers in China - meaning that for roughly 20 years, US investors bought Chinese stocks whose auditors had never been independently inspected.
Capital Controls & Fund Transfers
China maintains capital controls - money doesn't move freely across borders. A Chinese company can be highly profitable but still find it hard to move profits abroad to pay dividends to ADR shareholders.
Dividend withholding tax: China collects a 10% tax on dividends sent abroad. On top of that, there's tax in the receiving country (depending on the tax treaty). Dividends get "cut" twice before reaching the investor.
Currency risk: RMB/USD volatility plus the risk of a government-driven RMB devaluation during a trade war.
The low P/E of Chinese stocks isn't "the market being wrong" or "a missed opportunity". It accurately reflects the risks above: the VIE structure could be voided, government intervention is unpredictable, auditing isn't yet fully trustworthy, capital flows are controlled.
The correct way to evaluate it: treat a low P/E like a high yield on junk bonds - high yield because of high risk, not because the market "forgot" to price it. If you accept that risk with your eyes open - fine. But don't buy just because "P/E is 10 while Google's is 25, so it's twice as cheap".
Part IV - EBITDA, GAAP vs Non-GAAP, And The Tricks Used To Hide Losses
What is EBITDA - and why is it both useful and dangerous?
Put simply: EBITDA is profit before subtracting interest expense, taxes, and asset depreciation. It tries to measure "how much the company earns from its core operations", setting aside capital structure (debt vs equity), tax regime, and how assets are allocated.
EBITDA is useful in two cases:
- Comparing companies in the same sector but with different capital structures (Company A borrows a lot, Company B borrows little - EBITDA strips out that difference)
- M&A valuation - EV/EBITDA is the most common multiple because it "neutralizes" different tax jurisdictions and debt structures
But EBITDA is also extremely easy to abuse:
"Every time you see the word EBITDA, you should substitute the word 'bullshit earnings'."
Warren Buffett added: "People who use EBITDA are either trying to con you or they're conning themselves." The reason: depreciation is a real expense. A broken truck has to be replaced. An old server has to be swapped out. A degraded plant has to be repaired. Removing depreciation from the equation = pretending assets never need replacing.
Asset-light businesses
SaaS, marketplace, and platform companies - few fixed assets, low depreciation. EBITDA and operating income sit close together. In this case, EBITDA is a reasonable proxy for cash generation.
It's also fine when comparing across borders (different tax regimes) or for quick screening within the same sector.
Capital-intensive businesses
Airlines, oil & gas, telecom, manufacturing - huge depreciation, mandatory maintenance CapEx. Positive EBITDA with negative FCF is completely normal. WorldCom (2002) capitalized $3.8B of operating expense as an asset, pushing it into depreciation instead of expense - EBITDA looked clean, but it was actually the largest accounting fraud in US history at the time.
Beyond plain EBITDA, you'll encounter many variants:
EBITDAX = EBITDA + Exploration Expenses (oil & gas sector)
Adjusted EBITDA = EBITDA plus whatever the company wants to exclude (SBC,
restructuring, litigation, M&A costs...)
"Community Adjusted EBITDA" = WeWork's infamous invention - excluding even
marketing, G&A, and development costs. It turned a $769M EBITDA loss into a "profit" of $233M.
WeWork went bankrupt 4 years later.
GAAP vs Non-GAAP: Two versions of the truth
When reading a US company's earnings release, you'll see two sets of numbers:
Non-GAAP isn't inherently bad - sometimes it gives a clearer picture (for example, excluding a one-time restructuring charge). But the problem is each company chooses what to exclude on its own. There's no unified standard. And the most common trick: excluding SBC.
SBC: a "nonexistent expense" worth billions of dollars
Stock-Based Compensation (SBC) is paying employees in stock instead of cash. Under GAAP, SBC is a real expense - it reduces Net Income. But because SBC "doesn't cost cash", most tech companies exclude it from non-GAAP earnings. The result: non-GAAP EPS runs 31% higher on average than GAAP EPS across the sector.
The problem: SBC is a real expense - it's just paid not by the company, but by existing shareholders, through dilution. When a company issues new shares to employees, every old share you hold represents a smaller slice of the company.
This is how many tech companies "turn a loss into a profit" in the headline:
Step 1: Report a GAAP loss (because it counts SBC, restructuring, impairment).
Step 2: Exclude SBC from "Adjusted" earnings → suddenly it's a profit.
Step 3: Add SBC back into Operating Cash Flow (since SBC is non-cash) → CFO looks great.
Step 4: Declare "FCF positive" → investors are pleased.
But in reality: employees are still being paid. That cost has simply moved from cash flow to share dilution. If the company didn't pay in stock, it would have to pay in cash - and CFO would be much lower.
Dilution: the number the earnings release won't highlight
Every quarter, check Diluted Shares Outstanding (usually at the bottom of the Income Statement or in the EPS section). If this number grows more than 2-3%/year, SBC is "eating" into your ownership stake:
Gross Dilution vs Net Dilution
Gross dilution = total new shares issued to employees.
Net dilution = gross dilution minus share buybacks. Many tech companies claim "we buy back stock to offset dilution". But if buybacks total $2B while SBC issues $3B of new shares, net dilution is still positive. Check the diluted share count over 3-5 years - if it isn't shrinking, buybacks are just "chasing SBC".
How Much SBC Is "Normal"?
S&P 500 average: SBC ~2% of revenue.
Mature tech (Microsoft, Apple): 5-8% of revenue. Dilution stays low thanks to large buybacks.
Growth SaaS (CrowdStrike, Datadog): 15-25% of revenue. Acceptable if revenue growth exceeds 30% and the trend is declining.
Red flag: SBC above 30% of revenue with no declining trend. Or absolute SBC growing faster than revenue → each new dollar of revenue "costs" more dilution.
How far is too far when it comes to "adjusting"?
The SEC requires every non-GAAP metric to come with a reconciliation to GAAP table in the earnings release. This is where you see exactly what management has excluded. Read it.
Items commonly excluded, ranging from reasonable to suspicious:
If the gap between GAAP earnings and non-GAAP earnings keeps widening over time while operating cash flow isn't growing correspondingly - that's a serious red flag. The company is redefining "adjusted" more and more loosely to cover up deteriorating results.
Quick check: take FCF - SBC (instead of plain FCF). If this number is consistently negative, the company isn't actually making money - it's just shifting cost from cash to dilution.
"Other Income": when Net Income balloons because of one accounting line
SBC makes non-GAAP look better than GAAP. But there's an opposite trick - even more subtle - that makes GAAP itself look better: mark-to-market gains on equity investments. This shows up on the "Other Income (Expense), Net" line of the Income Statement, right above Net Income, and can be many times larger than profit from the core business.
In Q1 2026, Alphabet posted "blowout" results: Net Income up 81% YoY to $62.6 billion, diluted EPS up 82%. Every headline read "blowout quarter". But look at the Other Income, Net line on the Income Statement: +$37.7 billion - mostly net unrealized gains on non-marketable equity securities.
Translation: Alphabet invests in private AI startups (like Anthropic). When these startups raise a new funding round at a higher valuation, Alphabet has to mark up the value of its investment on the balance sheet to fair value. The gain flows straight into Other Income, then down into Net Income.
Over 60% of this quarter's $62.6B Net Income sits entirely inside an accounting spreadsheet. Not a single dollar of actual cash came into the company. Operating Income (profit from the core business: search, ads, cloud) only rose 30% - a "real" picture that's healthy but far less impressive than the headline 81%.
This isn't fraud. It's US GAAP standard practice (ASU 2016-01, effective 2018): companies must mark non-marketable equity securities to fair value every quarter, recording unrealized gains/losses directly in Net Income. Before 2018, these holdings sat outside the Income Statement (recognized only when sold). After 2018, they became a line that is highly volatile and non-cash, right inside the "bottom line".
Gains Are Created, Losses Are Too - It Cuts Both Ways
When a startup's valuation rises, Net Income balloons. When valuation falls (a down round, a market crash), Net Income collapses. In 2022, both Alphabet and Microsoft recorded multi-billion-dollar impairment losses on exactly these investments when the tech market corrected.
The consequence: Net Income becomes highly volatile in ways that don't correspond to operating quality. One "blowout" quarter might simply reflect one startup successfully raising a round, not the parent company running its business better.
Private Valuations Are "Estimates"
A listed stock has a clear market price. But non-marketable equity (a private startup) has no daily price - fair value is estimated based on the most recent funding round, an internal DCF, or comparable transactions.
Anthropic raises a round at a $X billion valuation - Alphabet has to mark its investment up proportionally. But is that valuation really "correct"? During the 2024-2026 AI bull market, AI startup valuations are suspected of running too high. If that's true, Alphabet's Net Income is resting on figures that could be marked down sharply.
When Net Income spikes suddenly, always check Other Income before celebrating:
1) What percentage of Net Income is Other Income? If above 20%, this is an
"unusual" quarter that shouldn't be extrapolated forward.
2) What makes up most of Other Income? Interest income (from cash holdings) is
recurring and "real". Unrealized gains on equity securities are non-cash and volatile.
3) Compare to Operating Income: this is the true measure of the
core business. Operating Income up 30% while Net Income is up 81% → the gap comes from
financial "noise", not business quality.
4) Cross-check against Cash Flow from Operations: unrealized
gains don't show up in CFO (because they're non-cash). If Net Income is up 81% but CFO is only
up 25-30%, that gap is exactly the mark-to-market effect.
The general rule: Operating Income measures how well the company earns from its core business. Net Income measures what else the company gained or lost from financing, investing, and one-time items. When the two diverge sharply - especially when Net Income far exceeds Operating Income in a given quarter - don't trust the headline. Dig into Other Income to understand where that "excess" came from. It could be real cash (a gain on sale of a business), or it could be pure accounting (mark-to-market). These two types carry completely different implications for future valuation.
Part V - 12 Common Mistakes When Reading Financial Statements
Confusing Profit With Cash Flow
The most classic mistake. A company reporting a $500M profit while Operating Cash Flow is completely negative can absolutely happen - and has happened many times (WeWork, Luckin Coffee). Accrual accounting allows revenue/expense to be recognized before cash actually moves. Always cross-check Net Income against CFO.
Looking At Only One Quarter
A single good or bad quarter rarely tells the full story. Did revenue spike from one large contract? Did margin rise from a temporary cut in R&D? Always look at the trend across at least 4-8 quarters (1-2 years) to separate signal from noise.
Comparing Metrics Across Different Sectors
"Apple's P/E is 30, JPMorgan's is 12, so JPM is cheaper" - wrong. The two sectors have completely different growth profiles, capital intensity, and risk. Only compare within the same sector, same business model. Apple vs Microsoft. JPM vs Bank of America. Exxon vs Chevron.
Skipping The Footnotes
Footnotes hold the truth that headlines don't say. Did revenue recognition policy change? Are there enormous operating lease obligations hidden off-balance-sheet (before ASC 842)? Suspicious related-party transactions? If you only read 3 pages, read the footnotes.
Trusting Non-GAAP Without Verification
"Adjusted EBITDA" might exclude SBC, restructuring, litigation, even "strategic investment losses". Every company defines it differently. Always read the "reconciliation to GAAP" section at the end of the earnings release to see exactly what's being excluded.
Treating Revenue As Everything
Revenue up 40% sounds impressive - but if gross margin falls from 70% to 50%, the company is "buying" revenue by cutting prices or selling lower-margin products. Revenue growth without margin expansion (or at least margin stability) is unsustainable growth.
Forgetting SBC When Calculating FCF
At tech companies, SBC often runs 15-25% of revenue. Because SBC is a non-cash expense, it gets added back into CFO, making FCF look higher. But SBC is a real cost - it dilutes shareholders. Calculate "FCF after SBC" = FCF minus SBC expense for a more accurate figure.
Not Understanding Seasonal Patterns
Retail Q4 (holiday season) is always strongest. Enterprise tech tends to close big deals at fiscal year-end. Comparing Q4 to Q3 (sequentially) doesn't mean much - always compare YoY (this year's Q4 vs last year's Q4). Except when looking for inflection points.
Treating Gross Margin The Same Across Sectors
Software: 70-85%. Grocery retail (Walmart, Costco): 23-27%. Energy: 30-50% depending on oil price. Banking: doesn't use gross margin at all. A grocery chain with 25% gross margin can be healthier than a SaaS company with 65% gross margin. Context is everything.
Ignoring Accounts Receivable Growth
If Accounts Receivable grows faster than Revenue (DSO - Days Sales Outstanding rising), there are two possibilities: (1) the company is loosening payment terms to "buy" revenue, or (2) customers are running into financial trouble and can't pay. Both are red flags.
Confusing Buybacks With Shareholder Value
A company buying back $10B of stock sounds like it's paying shareholders. But if it's simultaneously issuing $8B via SBC, net buybacks are only $2B. And if the purchase happens at too high a price (P/E of 50x when growth is decelerating), the buyback destroys value. Check the diluted share count over the years - if it isn't shrinking, buybacks are just offsetting dilution.
Reading Absolute Numbers Instead Of Ratios
"Apple made $97B, Ford made $1.8B, so Apple is 54 times better" - meaningless. You need to compare margin (Net Margin: Apple ~25%, Ford ~2-3%), return on capital (ROIC), and growth rate. Absolute numbers only tell you size, not quality.
Part VI - Warning Signs (Red Flags)
Beyond the 12 mistakes above, there are signs in a financial statement that, when you see them, you should stop and dig deeper before making a decision:
Part VII - A Quick 15-Minute 10-K Checklist
You don't need to read all 200 pages. With 15 minutes, here's the most effective flow:
Takeaways
The Income Statement is the "accounting version" of the truth. The Cash Flow Statement is the "cash version". The Balance Sheet is "financial health". These three views must be consistent with each other - when they aren't, that's where you need to dig deeper.
NIM for banks, ARR for SaaS, SSS for retail, RRR for oil & gas. Using the wrong metric for the wrong sector leads to the wrong conclusion. The first step is always understanding the business model before looking at the numbers.
The headline says "revenue up 20%". The footnotes say "due to a change in revenue recognition policy". The MD&A says "due to an acquisition, organic growth was only 3%". Three very different stories. The truth is usually in the smallest print.
No company has a perfect report. Good financial-statement-reading skill is the ability to spot when the pieces don't line up: profit rising while cash falls, revenue rising while receivables balloon, margins rising while market share falls.
03 Discussion
Leave a note
A considered space for questions, counterpoints, and useful additions. Civil, on-topic, signed.
Reader notes
...Loading notes...