Stock valuation frameworks that hold up.

Intrinsic value asks one simple question: what is this company actually worth? Not what the market says it's worth today, but what you'd pay if you owned the entire business forever.

Microsoft trades at $378 per share as of today. The company earned $12.05 per share last year. You're paying 31x earnings for a software giant that grew revenue 13% annually over the past decade. Is that a fair price? Too high? A bargain?

Market prices move on emotion, headlines, and momentum. Intrinsic value stays anchored to cash flows and growth rates. When the two diverge, you get opportunity. Amazon traded below its intrinsic value for years after the dot-com crash, despite growing revenue 25% annually.

Two methods dominate valuation work. Discounted cash flow (DCF) models project future cash flows and discount them back to today's dollars. This forward-looking approach works well for mature companies with predictable earnings. Multiples analysis compares price-to-earnings, price-to-sales, or enterprise value ratios across similar companies. Faster but less precise.

Take Nvidia at $450 per share, trading at 66x earnings. That seems expensive until you consider the company's AI chip dominance drove 126% revenue growth last quarter. A DCF model assuming 30% annual growth for five years, then 10% thereafter, might justify today's price. Assume 15% growth instead, and the stock looks overvalued.

The math matters less than the thinking. Every stock price implies a growth rate. Your job is deciding whether that implied growth rate makes sense given the company's competitive position, market size, and management execution.

A DCF model answers one question: what should you pay today for all the cash a company will generate forever? The math is simple. Add up every dollar of future cash flow, discount it back to today's dollars, and you get a fair value.

Three inputs drive everything. Growth rate: how fast will cash flows increase? Discount rate: what return do you demand for the risk? Terminal value: what's the company worth when your forecast period ends?

Here's the problem: terminal value typically accounts for 60-80% of your final answer. You're betting most of your valuation on what happens after year 10, which nobody can predict.

Take Microsoft trading at $380. Assume it generates $80 billion in free cash flow, growing 8% annually for 10 years, then 3% forever. Use a 9% discount rate. The DCF spits out roughly $380 per share.

Now change that terminal growth rate from 3% to 4%. The stock is suddenly worth $480. Drop it to 2%? Now it's $320. One percentage point moved the fair value by $60-100 per share.

This isn't a bug—it's the model working as designed. Small changes in forever assumptions create massive valuation swings. The discount rate is equally sensitive. Bump Microsoft's cost of capital from 9% to 10%, and you lose another $50 per share.

A DCF doesn't tell you what Microsoft is worth. It tells you what Microsoft is worth if your assumptions are correct. The model is brutally honest math applied to your best guesses about an unknowable future.

Use DCFs to stress-test your thinking, not to generate precise price targets. The exercise of building the model teaches you more than the output itself.

Price-to-earnings ratios get all the attention, but they miss a crucial point: GAAP earnings aren't cash. Netflix reported $4.49 per share in 2023 earnings but generated $6.9 billion in free cash flow—nearly double its net income. Depreciation, stock compensation, and timing differences make P/E ratios unreliable for cash generation analysis.

EV/EBITDA solves this by stripping out capital structure differences. When Microsoft acquired Activision for $68.7 billion, they compared enterprise values to EBITDA across gaming companies—not P/E ratios. This multiple ignores whether a company funds growth with debt or equity, making it perfect for acquisition comparisons.

EV/Revenue makes sense only in two scenarios: pre-profit companies or businesses with predictable margins. Salesforce trades at 7.2x revenue because SaaS margins are stable once you reach scale. But using revenue multiples for cyclical industrials like Caterpillar would be meaningless—their margins swing wildly with economic conditions.

Here's what most investors miss: every multiple is shorthand for growth plus margin expectations. A 25x P/E only looks expensive if you ignore the growth rate behind it.

Consider Walmart and Amazon, both trading at 27x earnings in early 2024. Identical P/E ratios, but Amazon was actually cheaper. Walmart's revenue grew 6% annually while Amazon's AWS division alone grew 13%. Amazon's higher growth rate justified the same multiple, making it the better value despite identical P/E ratios.

The key insight: never evaluate a multiple in isolation. A 15x P/E looks cheap until you realize the company's revenue is shrinking 10% annually. Context—growth rates, margin trends, competitive position—determines whether any multiple represents value or a trap.

Every price target needs to survive two brutal questions: what has to be true for this target to work, and what could go wrong that would break it? Most analysts skip this step. Don't.

Build three scenarios, not three numbers. Your bull case isn't just 20% higher revenue—it's a specific story about market expansion, pricing power, or operational leverage. Your bear case explains exactly how the business model breaks down. Salesforce trades at 45x earnings because investors believe in recurring revenue growth. If that growth slows to single digits, the multiple collapses.

Focus on the inputs that kill valuations. Terminal growth rate assumptions drive 60% of DCF value in most models. Dropping from 3% to 2% can slash your target by 15%. Margin assumptions matter more than revenue forecasts for mature companies. Capital expenditure requirements often get understated, especially for asset-heavy businesses.

Consider Zoom's 2021 story. Analysts set $600 price targets assuming 40% gross margins would expand as the company scaled. The bull case required enterprise customers to pay premium prices for advanced features. Instead, competition intensified and Zoom's margins peaked at 37%. The stock fell 85% as growth slowed and profitability disappointed.

Test your assumptions against historical precedent. Has any SaaS company maintained 80% gross margins at $10 billion revenue? Have retailers sustained 15% operating margins during recessions? If your model requires unprecedented performance, your price target needs extraordinary evidence.

Stress-testing isn't pessimism. It's the difference between a defensible investment thesis and expensive wishful thinking.

The hardest part of valuation isn't building complex models or projecting cash flows. It's walking away when the math tells you a good company trades at a bad price.

Even perfect execution on your analysis means nothing if you overpay. Tesla in late 2021 illustrates this perfectly. The company was executing flawlessly—ramping production, expanding globally, dominating EV market share. Every fundamental metric supported the bull case. Yet at 200 times earnings and $1.2 trillion market cap, the stock was priced for perfection. No room for error, no margin of safety. Investors who bought at $400 watched it fall 75% despite the company hitting most targets.

Three warning signs demand discipline. First, when your upside scenario barely justifies the current price. Second, when you need heroic assumptions—50% annual growth for a decade—to make the numbers work. Third, when minor disappointments could crater the stock because expectations are sky-high.

The most dangerous investments aren't obvious frauds or failing businesses. They're excellent companies trading at prices that assume everything goes right for years. Netflix in 2021, Zoom during peak pandemic, Amazon at the dot-com peak—all great businesses that destroyed wealth at the wrong entry points.

This discipline hurts. You'll watch stocks rise after you walk away. You'll look foolish when momentum continues. But preservation of capital matters more than missing out. The best analysts are as disciplined about what they don't buy as what they do.