A company — what is it actually worth?

Valuation is the most fundamental and most often misused tool in investing. Aswath Damodaran has a line: “Every valuation is wrong; the question is how wrong, and in what direction.” This guide organizes valuation’s five method categories (DCF, multiples, DDM, asset-based, sum-of-the-parts) + the operational workflow + special methods for early-stage companies + 15 classic traps by degree of operability. Read alongside A guide to reading US earnings filings, and you can upgrade from “I can read a filing” to “I can put a price on it.”

§01 · Framework — five ways to value a company

“What is a company worth” has five totally different answers depending on the angle, and each angle maps to a category of valuation method. Understanding the five angles matters more than memorizing any formula — because the same company can come out at 3× different numbers under different angles. The methods aren’t wrong; the angles are different.

1. Discounted future cash flows · DCF. “A company = all the free cash flow it will produce in the future, discounted at risk.” Cleanest in theory, most assumption-laden in practice — get growth rate, discount rate, and terminal value all wrong and the answer can be off 10×. Best for mature companies that produce stable FCF. e.g. Apple, Coca-Cola, Microsoft.
2. Relative valuation · Multiples. “Peers / history are at 15×, so this should be too.” P/E, EV/EBITDA, EV/Sales, P/B — quick and easy, but assumes the comp set is reasonable. If the whole market is expensive, all peers are expensive, and relative valuation amplifies that structural bias. The most common method in sell-side reports.
3. Dividend discount · DDM. “Shareholders actually receive dividends, so a company = the present value of all future dividends.” Best for steady-payout, near-zero retention utilities, banks, tobacco. Largely fails for tech and growth equities.
4. Asset-based valuation. “What’s left after selling all assets and paying down debt.” NAV / Liquidation Value. Suited to real estate, shipping, financials, resources — companies whose book assets are “relatively real.” For asset-light tech firms, the books are basically a stack of desks plus software, and NAV is essentially meaningless.
5. Sum-of-the-parts · SOTP. “A multi-business company = each business valued separately + net cash − holding discount.” Alphabet = Search + YouTube + Cloud + Waymo + Other Bets + cash. Segment-disclosure quality directly determines SOTP credibility.
6. Real options. “A company holds an optional ‘do / don’t do’ future project, and that optionality itself is valuable.” Biotech pipelines, natural-resource exploration rights, undeveloped land. An extension of Black-Scholes — beats DCF in projects with high uncertainty + large optionality.

Bottom Line · Valuation isn’t a single answer; it’s a range.

Professional valuation produces 2-3 methods + sensitivity analysis, outputting a “$90-$130” range, not a sub-decimal “$104.73.” Precision is illusion; a defensible range is the truth. Be skeptical of any sell-side report giving you a “precise single number” — that’s just an anchor.

§02 · DCF — three flavors of cash-flow discounting

DCF (Discounted Cash Flow) is the “theoretical origin” of valuation — intrinsic value = future cash flows discounted to today. But the same DCF framework has three different flavors, each requiring a different discount rate. Mixing flavors is one of the most common DCF mistakes.

General form. Value = Σ CF_t / (1+r)^t + TV/(1+r)^n. Three flavor choices: FCFF → WACC → Enterprise Value · FCFE → Cost of Equity → Equity Value · Dividends → Cost of Equity → Equity Value. From Enterprise Value to share price: EV − Net Debt − Minority Interest + Non-operating Assets = Equity Value ÷ Diluted Shares = intrinsic share price.

FCFF (Free Cash Flow to Firm) — cash flow available to all capital providers (equity + debt). Calculated as FCFF = EBIT × (1 − Tax) + D&A − Capex − ΔNWC. Discount with WACC to get Enterprise Value. Recommended for beginners, since it’s not affected by capital-structure changes.

FCFE (Free Cash Flow to Equity) — cash flow available only to shareholders. Calculated as FCFE = FCFF − Interest × (1 − Tax) − Net Debt Repaid. Discount with Cost of Equity to get Equity Value directly. Best for companies with stable capital structure.

Three variables drive 90% of the result: 1) Growth rate — how fast over the next 10 years? 2) Discount rate (WACC) — what return is required for this company’s risk? 3) Terminal value — what is the perpetuity value 10 years out? The first two are in §03, terminal value in §04.

Key DCF terms

1. FCFF · Free Cash Flow to Firm / Unlevered free cash flow. FCFF = EBIT × (1−T) + D&A − Capex − ΔNWC. Cash flow to “the firm as a whole.” Differs from OCF by adding back after-tax interest (no debt-cost deduction).
2. FCFE · Free Cash Flow to Equity / Levered free cash flow. FCFE = FCFF − Interest × (1−T) − Net debt repaid. Cash flow to shareholders. Maps directly to “distributable returns to equity.”
3. NWC · Net working capital. NWC = AR + Inv − AP. Receivables + inventory − payables. Increases in NWC consume cash, equivalent to incremental capex.
4. Reinvestment Rate. Reinv = (Capex + ΔNWC − D&A) / EBIT × (1−T). The share of earnings reinvested to sustain growth. Combined with ROIC, sets sustainable growth.
5. ROIC · Return on Invested Capital. ROIC = NOPAT / Invested Capital. Measures how much the company earns per $1 of invested capital. ROIC > WACC is the definition of value creation.
6. g · Long-term growth rate. g = ROIC × Reinvestment Rate. The mathematical ceiling on a company’s “sustainable growth.” A perpetual rate above g_GDP makes terminal value go to infinity.

⚠ The most common DCF mistake · Don’t treat EBITDA as cash flow.

“EBITDA × multiple = valuation” is a shortcut in PE work and some reports, but EBITDA is not cash flow — it doesn’t subtract capex, working-capital, or taxes. A company with $1B EBITDA and $800M capex has roughly $100M of FCF. Discount real cash-generating ability, not “approximations of cash flow.”

§03 · Discount rate — WACC and cost of equity

The discount rate is the “most sensitive + most often handwaved” parameter in DCF. A 1% change in WACC can move a 10-year DCF by 20-30%. Below: standard formulas + practical estimation methods for each parameter.

1. WACC · Weighted Average Cost of Capital. WACC = (E/V) × Ke + (D/V) × Kd × (1−T). Weighted blend of cost of equity and cost of debt. E = market cap of equity, D = market value of debt, V = E + D. T = marginal tax rate.
2. CAPM · Capital Asset Pricing Model. Ke = Rf + β × (Rm − Rf). Cost of equity = risk-free rate + beta × equity risk premium. The most general way to estimate cost of equity.
3. Rf · Risk-free rate. 10Y US Treasury yield is the standard for US companies. Emerging markets need a Country Risk Premium added (Brazil ~ +300 bps). Examples: 2026-04 US 10Y ~ 4.2%.
4. β (Beta). Stock’s systematic risk relative to the market. 1.0 = moves with market; >1 = bigger moves; <1 = milder. In practice, regress 2-5 years of weekly data. Examples: Apple β ~ 1.2; utilities ~ 0.6; biotech ~ 1.5.
5. ERP · Equity Risk Premium. Excess return of the market over the risk-free asset. Damodaran’s monthly implied ERP is the industry benchmark. US long term 5-6%, developed Europe 5-7%, emerging markets 8-12%.
6. Kd · Cost of debt. Average rate on existing debt OR YTM on newly issued bonds. Don’t use a low coupon from 10 years ago — use the YTM at current market prices.
7. Tax Shield. Interest is tax-deductible, so the effective cost of debt = Kd × (1−T). US corporate marginal tax rate generally 21-25%.
8. Capital Structure weights. Use market values, not book. If cash > debt (Apple, Google), D/V can be 0 or negative (negative net debt is generally treated as 0).
9. Unlevered β. βu = βl / [1 + (1−T)(D/E)]. “Pure business beta,” stripped of capital structure. When comparing peers, you must unlever first, then re-lever to the target’s structure.

WACC reference for major industries

Source: Damodaran 2025 industry tables · rough estimates · in practice use the company’s own data.

§04 · Terminal value — TV is 60-80% of valuation

In a 10-year DCF, terminal value typically accounts for 60-80% of total value. In other words — your assumption about “what happens 10 years out” matters several times more than your forecast for “next quarter”.

1. Gordon Growth · GGM. TV = FCF_n+1 / (WACC − g). Assumes the company grows at perpetual g from year n+1. g is generally the long-term nominal GDP growth (2-3%), never higher. As WACC and g get close, TV blows up — extremely sensitive.
2. Exit Multiple method. TV = EBITDA_n × Exit Multiple. Assumes the company is sold at the end of year n at the peers’ current EV/EBITDA multiple. More common in PE, but implicitly assumes “the multiple in 10 years equals today’s” — a strong assumption.

The two methods should cross-check each other — if your Exit Multiple TV implies an 8% growth rate but Gordon’s g is 3%, you owe an explanation for the 5% gap.

Common errors: 1) g > GDP — mathematically impossible forever; any company growing faster than the economy must dilute eventually. 2) Year-10 FCF spike — to engineer the desired TV, the final year FCF is artificially inflated. By the end of the projection, FCF should have entered “steady state” (ROIC ≈ WACC). 3) Capex = depreciation — perpetually, capex must ≥ D&A, otherwise assets get “depreciated to nothing” over an infinite horizon.

Damodaran golden rule: TV / Enterprise Value > 80% = warning; > 90% = redo the model. You’re really valuing “10+ years from now,” not the company today.

⚠ TV sensitivity example · A 1% shift in g changes TV by 40%+.

• WACC = 9%, g = 2%: TV multiple = 1/(9%-2%) = 14.3× FCF
• WACC = 9%, g = 3%: TV multiple = 1/(9%-3%) = 16.7× FCF (+16%)
• WACC = 9%, g = 4%: TV multiple = 1/(9%-4%) = 20× FCF (+40%)
• WACC = 9%, g = 5%: TV multiple = 1/(9%-5%) = 25× FCF (+75%)

Always run a sensitivity matrix — a 3×3 of WACC 8%-10% × g 1%-3% — and present a range, not a number.

§05 · Multiples — how to use multiples right

Relative valuation is what 90% of sell-side reports use — simple, fast, intuitive. But “simple = easy to get wrong.” Below: mainstream multiples + fit + comp selection + common traps in one pass.

Mainstream multiples

Three-step relative-valuation workflow

1. Pick the right comps. “Same industry” doesn’t equal “comparable.” Need similarity in business model + scale + growth + geography + profit stage. Shopify is better compared with BigCommerce than with Amazon; Tesla’s better compared with BYD + Rivian than with Toyota.
2. Adjust to comparable basis. Different companies’ “EBITDA” can differ by 30% (SBC added back? leases handled how? one-offs stripped?). Use a single source’s adjusted data (Damodaran / CapIQ / Bloomberg), or normalize on your own.
3. Use the median, not the mean. One or two outliers skew the mean. The median is more stable. Even better, take the 25%-75% quartile range, not a single point.
4. Apply “market premium / discount”. If peers are at 15× and the target grows 20% faster → 18× makes sense; if it grows slower → 12×. Quantify the difference using PEG or “growth + margin adjustment”; don’t just hand-wave ”+/- 10%.”
5. Compare with the company’s own history. Beyond peers, also compare to the company’s own historical median. Current P/E vs. its 5-year median z-score is an objective cheap/expensive signal.
6. Cross-check across multiples. Relying on P/E alone is easy to mislead with one-offs. Watch P/E + EV/EBITDA + P/S + FCF Yield together: if all point the same way, conclusion is sturdier; if they diverge, dig into why.

§06 · Dividend discount — DDM / Gordon / H-Model

Dividend discount is the “original” DCF — proposed by John Burr Williams in 1938’s The Theory of Investment Value. The core: shareholders only actually receive dividends, so the company is worth the present value of all future dividends.

1. GGM · Gordon Growth Model. P = D_1 / (r − g). First-order approximation under “dividends grow perpetually at g.” Simplest, most classic. Best for mature dividend payers. The denominator r-g is extremely error-sensitive.
2. Two-Stage DDM. High-growth phase, then perpetuity. In reality companies don’t go straight to perpetuity. Use 5-10 years of g1, then g2 in perpetuity. Good fit for newly dividend-paying companies (mature tech entering payout).
3. H-Model. P = D_0(1+g_L)/(r−g_L) + D_0×H×(g_H−g_L)/(r−g_L). Closed-form solution where growth rate linearly transitions from g_H to g_L (H = half the transition length). More realistic than the two-stage model.
4. Payout Ratio. Payout = Dividend / EPS. Share of earnings paid out. Utilities 60-80%, banks 40-50%, growth tech 0-20%.
5. Sustainable Growth. g = ROE × (1 − Payout). Mathematical constraint of reinvested earnings. g cannot exceed ROE × retention, otherwise external financing is required.
6. Yield + Growth. Total Return ≈ D/P + g. Approximate decomposition of long-term shareholder return. Most blue chips = 2-3% yield + 5-7% earnings growth = 7-10%.

When to use DDM · Dividends > 50% of earnings, with stable dividend history > 10 years.

• Fits: utilities, telecom, tobacco, consumer staples, banks, mature insurance, REITs
• Doesn’t fit: growth tech, biotech, cyclicals, startups, non-payers
• Substitute: for non-payers, switch to FCF-based DCF (treat FCF as “distributable but undistributed dividends”)

§07 · Asset-based — NAV · Liquidation · Replacement

When a company’s primary value lies in the assets it owns (not in future cash flow), asset-based valuation is the most direct. Typical cases: real estate, shipping, resource mining, certain financial holding companies.

1. NAV · Net Asset Value. NAV = Σ fair value of assets − liabilities. “Suppose all assets are sold today at fair value.” REITs, PE funds, holding groups all use this. Note fair value ≠ book value — real estate uses cap rates, securities use mark-to-market.
2. Liquidation Value. The “fire sale” price. Typically 20-50% below NAV (forced-sale discount). Absolute floor of valuation. Benjamin Graham’s “cigar butts” were companies trading below liquidation value.
3. Replacement Cost. “How much would it take to rebuild this company from scratch.” Tobin’s Q = Market Cap / Replacement Cost. Q < 1 = cheaper to acquire than build new — M&A opportunity.
4. Book Value. Shareholders’ equity on the balance sheet. Directly meaningful for banks / insurers / REITs where “the assets are the business”; nearly useless for tech / consumer-goods firms.
5. Tangible Book. TBV = Book Value − Goodwill − Intangibles. Book value with goodwill stripped out. Core metric for bank valuation; P/TBV < 1 has historically been the cheap zone.
6. EPV · Earnings Power Value. EPV = Normalized NOPAT / WACC. Popularized by Bruce Greenwald — “value generated by current operations alone, assuming no growth.” EPV > NAV = growth value present; EPV < NAV = idle assets.
7. Hidden Assets. Items significantly understated on the books: land / property at historical cost, listed-equity holdings, unused NOLs. Japanese companies famously have lots of hidden assets, targeted by activist shareholders.
8. DCF of Assets. For natural-resource companies (oil, mining): discount each barrel / ton of reserves at current price minus extraction cost. Standard for oil-major NAV valuation.
9. P/NAV. Standard valuation metric for REITs, holding companies, resource firms. REIT historical mean 0.9-1.1×; mining stocks in bull markets 1.5-2×, in bear markets 0.5-0.7×.

§08 · Sum-of-the-parts — SOTP for multi-business companies

When a company has multiple business lines with different growth and valuation logic, forcing one consolidated multiple “averages out” the differences. SOTP (Sum-of-the-Parts) values each business separately, then sums and adjusts.

Standard process

1. Split financials by segment disclosure — revenue / op income / segment assets from 10-K Segment Reporting
2. Pick a fitting valuation method per business — mature cash-flow businesses use DCF or EV/EBITDA; high-growth use EV/Sales; early-stage use comparable IPO pricing
3. Sum the Enterprise Values
4. Add non-operating assets — cash, available-for-sale securities, land banks, equity stakes (e.g. Tencent’s Meituan/JD holdings)
5. Subtract debt + minority interests + restricted items
6. Apply holding / non-controlling discount — empirically 10-20% “holding discount” (the market doesn’t pay top multiple for every line)

Example: hypothetical Alphabet SOTP

When SOTP fits · Multi-business + meaningful segment differences + potential spin-off value.

• Internet majors: Alphabet / Meta / Amazon / Alibaba / Tencent — main business + cloud + ads + investments
• Traditional conglomerates: GE / Siemens / Honeywell / Berkshire — multiple business lines
• Media / telecom: Disney / Comcast / Verizon — content + distribution + network infrastructure
• Holding companies: Berkshire / SoftBank / Alibaba — portfolio-driven structure

§09 · Startup valuation — VC / early stage / unicorns

DCF on early-stage companies is nearly meaningless — no stable cash flow, no real comparables, even growth is a guess. The VC industry developed a completely different toolkit, with the core idea “back-out from exit valuation + discount for risk.”

1. VC Method · Venture Capital Method. 1) Exit-year revenue × exit multiple = Exit Value · 2) Exit Value / target return multiple = pre-money. Most common in VC. Assume exit in 5 years × 20× return → reverse-engineer today’s pre-money. All assumptions live in “exit-year revenue” and “target multiple.”
2. Berkus Method. Common pre-seed. Assign $0-500K to each of 5 items: 1) Sound Idea 2) Prototype 3) Quality Team 4) Strategic Relationships 5) Product Rollout / Sales. Max valuation $2.5M; suited to no-revenue stage.
3. Scorecard. Reference the median seed valuation in the same region / stage, then apply a premium / discount across 5-6 dimensions (team, market, product, competition, funding need). Popularized by Bill Payne.
4. RFS · Risk Factor Summation. Start from an “industry baseline valuation,” then add or subtract +$500K each across 12 risk factors (management, stage, legislation, manufacturing, marketing, funding, competition, technology, litigation, international, reputation, exit). Proposed by Ohio TechAngels.
5. Comparables. Use PitchBook / CB Insights / Crunchbase to look up median valuations from recent same-stage / same-vertical raises. AI vertical 2023-2025 carried a 3-5× premium.
6. First Chicago. Probability-weighted across 3 scenarios (success / survival / failure). Typical: Success 40% × $500M + Survival 30% × $50M + Failure 30% × $0 = $215M. Realistic approach that accounts for failure rates.
7. Pre / Post Money. Post-Money = Pre-Money + Investment. “Pre-money $20M, raise $5M” → post-money $25M, new investor owns 20%. Always check whether announcements are pre- or post-money.
8. SAFE / Convertible. Y Combinator’s Simple Agreement for Future Equity. “Valuation cap + discount” mechanism defers pricing, convenient for early stage. Converts at the next round at the cap or discount.
9. Unicorn Discount. Stanford research: announced unicorn valuations on average overstate by 50% (because the headline simply linearly extrapolates the latest preferred round, ignoring liquidation preferences). The gap between “paper” and “real” valuation is large.

§10 · Workflow — from a 10-K to a price range

Putting theory into practice: given a 10-K, how do you produce a credible valuation range in 4-8 hours? Below is a standard workflow.

Hours 1-2 · Foundations

• Read 10-K Business Section + MD&A + Risk Factors (1 hour)
• Pull 3 years + latest quarter financials (Rev / Gross / OpInc / FCF)
• Identify core segments / product lines + economics of each
• Assess current business-model maturity (growth / steady / decline)
• List 3-5 comparable companies

Hours 3-4 · Build

• Forecast 5-10 years of revenue (by segment or top-down driver)
• Forecast gross margin path (converging to peer / own historical median)
• Forecast operating margin (OpEx ratio evolution)
• Compute FCF (NOPAT + D&A − Capex − ΔNWC)
• Estimate WACC (CAPM + current capital structure)
• Pick a terminal-value method (Gordon or Exit Multiple)

Hours 5-6 · Cross-check

• Independently value with multiples (P/E, EV/EBITDA, EV/Sales)
• Compare with peer median + own historical range
• If DCF vs Multiples differ > 20%, explain why
• Sensitivity matrix: WACC ±1% × g ±1%
• Plot the valuation range (most-bear / base / most-bull)
• Map current share price onto the range

Hours 7-8 · Documentation + judgment

• Write a one-page “key assumptions” summary (3 most sensitive variables)
• List “what to verify” (what to watch for next earnings)
• Issue Buy / Hold / Sell + trigger conditions
• Set “how I’d know I’m wrong” signals (Bearish Catalysts)
• Position size = f(valuation discount, conviction, portfolio correlation)
• Archive: revisit in 6 months and track assumptions vs. reality

§11 · Common traps — 15 classic mistakes

1. Terminal value > 85%. You’re really valuing “forever,” not the company. Either extend the projection to steady state, or lower perpetual g.
2. Perpetual growth > nominal GDP. No company can outgrow the economy forever. Cap g at 3% (US long-run nominal GDP ~4%).
3. Extrapolating the high-growth phase 10 years out. “30% CAGR for 5 years, so 30% for the next 10” → absurd valuation. Every company’s growth must decay — the S-curve is the rule.
4. Capex perpetually below D&A. “Assets wear down but no money is reinvested.” In perpetuity Capex must be ≥ D&A.
5. Treating SBC as costless. Stock-based comp is non-cash but dilutes equity — a real cost. When using FCF, don’t add SBC back, or count dilution into share count.
6. Picking friendly comps. Tesla against Toyota, NIO, or Amazon? Comp choice determines the answer. Comp selection must be defensible.
7. Ignoring one-offs. TTM net income includes a $10B one-time tax refund — extrapolating to perpetuity = severe overvaluation. Use normalized / adjusted figures.
8. Hand-waving WACC. “~10% works for tech” — but the real answer must come from CAPM. Each 1% WACC change moves DCF 20-30%.
9. Using book values for capital-structure weights. WACC’s E/V and D/V are market-value weights, not book. Use latest share price × shares outstanding.
10. Same WACC across all stages. Early-stage high risk should be discounted higher; steady-state should be discounted normally. A single WACC for the entire model underestimates early-stage risk.
11. Miscomputing EV in EV/EBITDA. EV = Market Cap + Net Debt + Minority − Associates. Using Total Debt or forgetting Minority is the common error.
12. Ignoring lease obligations. Post-ASC 842, long-term operating leases are on-balance-sheet. When computing EV, include Operating Lease Liability in Net Debt, otherwise retail / airlines understate EV significantly.
13. Circular “Exit Multiple” reasoning. Use “industry historical multiple” to set the exit multiple → reverse-prove “it’s worth this multiple.” Cross-check with Gordon.
14. “Precise to the decimal”. “Fair value = $127.83” is absurd. The right output is “$100-$145, midpoint $120.” Precision is illusion.
15. Reverse-engineering target prices. The original sin of sell-side: start from “$300 target,” then back out assumptions. Model first, judgment second; reversed, it becomes magic.