If you ask three Wall Street analysts what a stock is worth, you get three numbers and three precise spreadsheets defending each one. The honest answer is that intrinsic value is a range, not a number, and any retail investor can produce a defensible range in 20 minutes using four simple methods that do not require a finance degree.
This post walks through the four. We then run all four on Microsoft (MSFT) and show how the answers triangulate.
What "intrinsic value" actually means
Buffett's definition: the discounted value of the cash that can be taken out of a business between now and judgement day. Two consequences fall out:
- Intrinsic value is forward-looking. It depends on what the business will do, not what it has done.
- Intrinsic value is a range, because the future is uncertain.
Any single-number "intrinsic value" is misleading. The right answer to "what is MSFT worth" is a range like "between 350 and 480 USD per share, depending on assumptions, with the centre of the range around 415."
Method 1: The Graham Number
The Graham number is the simplest valuation method in the value-investing canon. Benjamin Graham defined it in "The Intelligent Investor" (1949). It is a quick screen, not a precision tool, but it surprisingly often catches the right zip code.
The formula:
Graham Number = sqrt(22.5 x EPS x Book Value per Share)
The 22.5 comes from Graham's twin caps on what a defensive investor should pay: P/E at most 15, P/B at most 1.5. Multiply them: 15 x 1.5 = 22.5. The square root of (22.5 x EPS x BVPS) is the price at which you are right at the boundary of both caps.
Worked example on Microsoft (MSFT):
- MSFT EPS (TTM): roughly 12.50 USD
- MSFT book value per share: roughly 38 USD
- Graham number = sqrt(22.5 x 12.50 x 38) = sqrt(10,687) = 103.4 USD
Microsoft trades around 415 USD. The Graham number says Microsoft is roughly four times Graham's defensive-investor ceiling. That is the punchline for almost every high-quality growth compounder in 2026: by Graham's strict 1949 rules, the entire S and P 500 is wildly overvalued. The Graham number is a starting point, not a verdict. It tells you the stock is not in deep-value territory by classical metrics.
When the Graham number is useful: cyclicals at the bottom of their cycle, financials trading near book value, beaten-up industrials. When it is not useful: software, platforms, consumer brands, anything with intangibles that drive the value but do not appear on the balance sheet.
Method 2: Simplified DCF (Discounted Cash Flow)
DCF projects free cash flow forward, discounts each future year back to present at a discount rate, and sums the present values. Most retail investors think DCF requires a 200-row spreadsheet. It does not. A 3-stage simplified DCF takes 10 lines.
The formula:
Intrinsic Value = sum of [FCF_year_n / (1 + r)^n] for n = 1 to 10
+ Terminal Value / (1 + r)^10
Where Terminal Value = FCF_year_10 x (1 + g) / (r - g)
r = discount rate (8 to 10 percent for stable large caps)
g = terminal growth rate (typically 3 percent, capped at long-run GDP)
Worked example on Microsoft:
- Current FCF: roughly 75 billion USD
- Assume FCF grows 11 percent per year for the next 10 years (recent CAGR has been 13 percent, modest haircut for size)
- Discount rate: 8.5 percent (large investment-grade tech)
- Terminal growth: 3 percent
Year 1 FCF: 75 x 1.11 = 83 billion. Year 10 FCF: 75 x (1.11)^10 = 213 billion.
Sum of present-valued years 1 to 10: roughly 1,030 billion.
Terminal value at year 10: 213 x 1.03 / (0.085 - 0.03) = 213 x 18.7 = 3,983 billion.
Present value of terminal: 3,983 / (1.085)^10 = 1,766 billion.
Total intrinsic value: 1,030 + 1,766 = 2,796 billion USD.
Microsoft market cap is roughly 3,100 billion. The DCF says Microsoft is roughly 10 percent expensive on these assumptions. Change the 11 percent FCF growth assumption to 13 percent (matching recent CAGR), and the DCF result goes above 3,500 billion and Microsoft looks fair.
This is the famous DCF problem: the result is precise but the precision is fake. The growth rate and discount rate dominate the answer. Use DCF as one estimate, not the truth.
Method 3: Owner Earnings Yield (Buffett's preferred number)
Owner earnings is Buffett's term for the cash that can be extracted from the business without impairing its competitive position. The formula (from Buffett's 1986 letter):
Owner Earnings = Net Income
+ Depreciation and Amortization
- Maintenance Capex (not growth capex)
+/- Working Capital changes
We walk through this in detail in /blog/what-is-owner-earnings-buffett/.
Owner earnings yield is the inverse, divided by market cap:
Owner Earnings Yield = Owner Earnings / Market Cap
Worked example on Microsoft:
- Estimated owner earnings (using a 5-year average to smooth): roughly 78 billion USD
- Market cap: roughly 3,100 billion USD
- Owner earnings yield = 78 / 3,100 = 2.52 percent
Buffett's published thresholds (rough):
- Above 8 percent: clearly cheap
- 5 to 8 percent: fair price for a wonderful business
- 3 to 5 percent: stretched, requires extreme business quality
- Below 3 percent: expensive on absolute terms
Microsoft at 2.52 percent is below the 3 percent floor. It is also below the 10-year Treasury yield of roughly 4.5 percent, which is the absolute floor Buffett invoked in his 2025 letter.
Translation: Microsoft is genuinely high quality. Almost every Buffett scorer rates Microsoft as a wonderful business on Moat, Durability, Health, and Management. The valuation pillar, however, is Partial Fit at this price. You are paying for the quality and then some.
Method 4: PE times Growth (Lynch's PEG)
Peter Lynch's PEG ratio is the simplest growth-adjusted valuation:
PEG = (P/E ratio) / (EPS growth rate, expressed in percentage points)
Lynch's thresholds:
- PEG below 1.0: attractive
- PEG between 1.0 and 1.5: fair
- PEG above 1.5: overpaying for growth
Worked example on Microsoft:
- MSFT P/E ratio (forward): roughly 33
- 5-year forward consensus EPS growth: roughly 14 percent
- PEG = 33 / 14 = 2.36
PEG above 2.0 is the territory Lynch said you should walk away from unless the business quality is truly exceptional. Microsoft's quality is exceptional, but Lynch's own books make clear he would not pay PEG 2.36 even for an exceptional business. He would wait.
When the PEG is useful: stable-growth quality businesses with predictable EPS growth. When it is not: cyclicals, turnarounds, anything where EPS swings widely year to year.
Triangulating the four results
Microsoft, four estimates:
| Method | Result | What it says |
|---|
| Graham number | 103 USD vs 415 USD price | Deep-value framework, says expensive by 4x |
| DCF (8.5%, 11%, 3%) | 2,796 B vs 3,100 B mcap | Roughly 10% expensive on these assumptions |
| Owner earnings yield | 2.52% | Stretched (target was 5%+) |
| PEG | 2.36 | Overpaying for the growth (Lynch threshold 1.5) |
Three of four methods agree Microsoft is expensive on absolute terms in 2026. The DCF is the most generous (and most assumption-dependent). The owner earnings yield is the cleanest single number; it says you are paying about double what Buffett would historically tolerate.
The triangulation: Microsoft is a wonderful business at a stretched price. Not catastrophically overvalued. Not a bargain. Owning Microsoft at 415 USD requires confidence that the next decade of FCF growth comes in at least at the 11 to 13 percent rate, or that the discount rate falls (interest rates drop), or both.
This is exactly the answer the Buffett-Fit Score on /buffett/msft/ returns: Strong Fit on Moat, Durability, Health, and Management; Partial Fit on Valuation. Three of four valuation methods agree.
Why all four, instead of just one
Each method has a structural blind spot:
- Graham number assumes book value reflects the value of assets. Useless for asset-light businesses.
- DCF assumes you can forecast 10 years of FCF. Garbage in, garbage out.
- Owner earnings yield assumes maintenance capex can be estimated. Judgement call.
- PEG assumes EPS growth is sustainable. Useless for cyclicals.
Use all four. Where they agree, you have a defensible estimate. Where they disagree, the disagreement itself is informative (a business that looks cheap on Graham and expensive on PEG is usually a cyclical at a peak).
Where invest-like fits
Every stock page on invest-like at /buffett/[ticker]/ surfaces all four computed valuations, plus a fifth (reverse-DCF, which inverts DCF to compute the implied growth required to justify the current price; see /blog/how-to-value-a-stock-2026/ for the full walkthrough).
The Buffett-Fit Score combines all five into a single Valuation pillar grade. The verdict (Strong Fit, Partial Fit, Wait) is the synthesis. The point of computing five is so that no single assumption dominates the conclusion.
Quick checklist for any new stock
- Compute Graham number. Is the stock within 50 percent of it?
- Compute owner earnings yield. Is it above 5 percent?
- Compute PEG. Is it below 1.5?
- Run a simplified DCF. Is the result within 20 percent of market cap?
If three of four say "cheap to fair," you have a candidate. If three of four say "expensive," walk away unless the business quality is genuinely exceptional.
Disclosure
Educational tool. The Microsoft numbers above are illustrative and use approximate fundamentals. Actual numbers move quarter to quarter. The four valuation methods are standard practice in value investing; the synthesis is the judgement.
Author: Zaid Ghazal, founder of invest-like, Kiel, Germany. Not a registered investment adviser. Full methodology at /methodology/. DCF and reverse-DCF documentation at /methodology/dcf/.