Every valuation ends in a number with decimals. Almost none of them have earned the decimals. The Path 9 scorecard stops at a question it cannot answer: what is a fair price? This module answers it the honest way, as a range with the assumptions showing.
One company, dated numbers, every step you can redo. I use Tata Consultancy Services (TCS), a Nifty 50 company with almost no debt and cash flows that are easy to read. The figures come from its annual report for FY26 (the year to 31 Mar 2026) and NSE prices of 31 Aug 2026, kept more than 30 days old as SEBI's education norm asks. It is a worked example of method, not a view on the stock.
- Read the priceTurn it into four ratios and see what each leaves out.
- Place it in a rangeAgainst its own history and its sector, on one date.
- Build a DCFFive inputs, each with its source.
- Run it backwardsFix the price and solve for the growth it assumes.
- Write the rangeAnd what would change your mind. Never one number.
3By the end you can
- Build P/E, P/B, EV/EBITDA and dividend yield from one price and one annual report.
- Run a five input DCF and a reverse DCF, and see how much rests on assumptions.
- Write a valuation as a range, with the inputs and what would change it.
Four ratios from one price
TCS closed at ₹2,399.30 on 31 Aug 2026 (NSE). Times 361.81 crore shares, that is a market value of ₹8,68,091 crore. Four ratios come out of it.
P/E ratio, on FY26 earnings of ₹136.01 a share
TCS FY26, NSE 31 Aug 2026
P/B ratio, on book value of ₹296.40 a share
TCS FY26, NSE 31 Aug 2026
EV/EBITDA, on EBITDA of ₹72,398 crore
TCS FY26, NSE 31 Aug 2026
dividend yield, on ₹110 declared for FY26
TCS FY26, NSE 31 Aug 2026
| Ratio | How it is built | TCS | What it leaves out |
|---|---|---|---|
| P/E | Price divided by earnings per share | 2,399.30 / 136.01 = 17.6. Before one off charges, 16.4 | FY26 profit carried ₹4,526 crore of one off charges before tax |
| P/B | Price divided by book value per share | 2,399.30 / 296.40 = 8.1 | Brands and people are not on the balance sheet. Return on equity was 48.7% on average equity |
| EV/EBITDA | Market value plus debt minus cash, divided by operating profit before depreciation | 8,32,478 / 72,398 crore = 11.5 | Tax, capital spending and what leases cost |
| Dividend yield | Dividends per share divided by price | 110 / 2,399.30 = 4.6%. Without the ₹46 special dividend, 2.7% | It is before tax, and a special dividend may not repeat |
Enterprise value (EV) is the price of the whole operating business: market value ₹8,68,091 crore, plus lease liabilities ₹11,283 crore, minus cash, bank balances and investments ₹46,896 crore, which is ₹8,32,478 crore. EBITDA is revenue of ₹2,67,021 crore less operating costs of ₹1,94,623 crore.
The same day NSE showed the Nifty 50 at a P/E of 20.36, a P/B of 2.92 and a dividend yield of 1.17%. TCS had the lower P/E and a P/B 2.8 times higher: one business that earns a lot on very little book value.
3 terms this section turns on
A ratio is a range, not a number
Is 17.6 low? Against what? Only a range answers. Over the 472 trading days from 7 Oct 2024 to 31 Aug 2026, NSE's Nifty 50 P/E ran from 19.62 to 23.62, median 21.86. The Nifty IT index, TCS's sector, ran from 16.91 to 36.41, median 26.48.
Nifty IT P/E fell from 34.15 on 29 Nov 2024 to 19.90 on 31 Aug 2026. The Nifty 50 P/E moved between 19.62 and 22.97 at month ends over the same two years.
NSE Indices daily P/E, last session of each month from Oct 2024 to Aug 2026. Index P/E as published by NSE.
The sector multiple fell 42%, from 34.15 on 29 Nov 2024 to 19.90 on 31 Aug 2026. Whatever you call normal for IT, there was a day in these two years when the index stood at least a third away from it. TCS at 17.6 sat below the sector (19.90) and the Nifty 50 (20.36). That says where a number sits, never why.
Read a multiple three ways: against the company's own range, against its sector on the same date, and against what the business earns on its capital. Always with a date on it.
A simple DCF in five inputs
A discounted cash flow (DCF) adds up the cash a business will free up, each year's cash cut back for the wait. Two of the five inputs are facts from the annual report. Three are my assumptions, and those three decide the answer.
Two terms to hold on to from 9.8
| Input | TCS | Where it comes from |
|---|---|---|
| Starting free cash flow | ₹47,948 crore | FY26 cash from operations of ₹52,094 crore less ₹4,146 crore of capital spending |
| Growth, years 1 to 10 | 7% a year | My assumption. In FY26 free cash flow grew 6.6% and revenue 4.6% |
| Growth after year 10 | 4% a year | My assumption: a rate that does not outgrow the economy |
| Discount rate | 12% | My assumption: the 10 year government bond yield of 6.97% on 31 Aug 2026 plus 5 points for owning shares, rounded |
| Net cash, shares | ₹35,613 crore, 361.81 crore | Cash and investments of ₹46,896 crore less leases of ₹11,283 crore; annual report |
The centre case gives ₹2,229 a share, 7% below the 31 Aug price. Do not read that as a verdict. The second number matters more: 51% of the value comes from the years after year ten. Half of what you computed rests on a growth rate nobody can observe.
Why ranges beat point targets
Now move the inputs. One point less on the discount rate lifts the value 14%; one point more cuts it 11%. Three points less growth cuts it 18%; three more lifts it 23%. Here are nine defensible sets.
Where this section sits in module 9.8
- 13% rate, 4% growth₹1,630
- 12% rate, 4% growth₹1,821
- 13% rate, 7% growth₹1,980
- 11% rate, 4% growth₹2,067
- 12% rate, 7% growth₹2,229
- Price, 31 Aug 2026₹2,399
- 13% rate, 10% growth₹2,415
- 11% rate, 7% growth₹2,552
- 12% rate, 10% growth₹2,739
- 11% rate, 10% growth₹3,158
DCF value per share under each set of assumptions, growth for years 1 to 10 then 4% forever. Illustration of method, not a valuation of TCS, a target or a recommendation. FY26 annual report and NSE, 31 Aug 2026.
A spread that wide does not tell you what to do. That is the information: the data cannot pin the answer down. One number hides this. Hand someone ₹2,229 and they hear precision. Hand them ₹1,630 to ₹3,158 and they hear what you actually know.
There is a rule behind the habit. In India a price target is a research service under the SEBI (Research Analysts) Regulations 2014, the work of registered research analysts. RupeeCase Learn is education, so it shows the method and gives no target.
Run it backwards: what does the price assume?
A reverse DCF keeps the price and solves for the unknown. It asks what you must believe to pay today's price. For TCS, hold the other inputs and solve for growth in years 1 to 10:
| Discount rate | Growth a year, years 1 to 10, that the price needs | Then |
|---|---|---|
| 11% | 6.1% | 4% forever |
| 12% | 8.1% | 4% forever |
| 13% | 9.9% | 4% forever |
Check it: type 8.08 into the growth box above and the value per share returns to about ₹2,399. Now set the numbers beside FY26, when revenue grew 4.6%, free cash flow 6.6% and profit before one off items 8.8% (annual report). At 12% the price asks for growth in that neighbourhood for a decade, then 4% forever. That is neither a bargain nor a warning. It is the belief inside the price, written where you can argue with it.
Last, write what would change your answer. Revenue growth below the implied rate for two years running? A bond yield two points higher, lifting your discount rate? Module 9.7 calls these thesis invalidation triggers. They outlive the number.
How value tilts use these numbers
A systematic value tilt does not build a DCF per stock. A DCF carries three guesses, so 500 stocks would carry 1,500. A tilt ranks on ratios that are measured, not argued over.
- Rank. Sort the whole universe on P/E, P/B, EV/EBITDA and dividend yield, same date, same definition.
- Blend. Each ratio fails somewhere: P/B for asset light software, P/E at the peak of an earnings cycle. A blend fails less.
- Filter. Drop weak balance sheets and unstable earnings, so cheap does not mean cheap for a reason.
- Wait. Value can lag for years. From 13 Mar to 31 Aug 2026 the Nifty200 Value 30 rose 0.4% while the Nifty200 Momentum 30 rose 10.4% and the Nifty200 Quality 30 rose 8.4% (NSE closes). Past returns are not a guide to the future.
On 31 Aug 2026 that index stood at a P/E of 9.30, a P/B of 1.28 and a yield of 2.04%, against 20.36, 2.92 and 1.17% for the Nifty 50 (NSE). Module 3.2 covers the factor itself.
Common mistakes
MythA DCF tells you what a stock is worth.
FactA DCF turns five inputs into one number. For TCS, one point on the discount rate moves it 11% to 14%, and half of it comes from beyond year ten.
- Tuning until it agrees. If you change inputs until the DCF matches the price, you have measured your own patience. Write the inputs and sources first.
- Quoting a point. ₹2,229 is one cell of a grid. Show the grid.
- Comparing with no date or peer. A multiple means something only beside a date and a comparable.
- Counting a special dividend as income. TCS declared ₹46 of its ₹110 as a special payout.
- Valuing a lender like a software firm. Banks borrow money to lend money. Use P/B with return on equity and asset quality, not EV/EBITDA.
RupeeCase strategies do not value companies. They rank NSE index universes on price momentum, or hold a fixed rule, and none of the 18 uses a P/E, a P/B or a DCF as a signal. Nifty 50 Equal Weight holds all 50 names with no selection at all. What carries over is the discipline: the rule is written before the stock is seen, and the rebalance date is fixed. None of the cards carries a value tilt. The fixed calendar is explained in the tutorial on rebalance cadence. Compare the cards on the strategies page, starting with Nifty 50 Equal Weight.
Key takeaways
- Four ratios from one price, each with what it leaves out.
- A multiple means something only beside a date, a sector and the return on capital.
- In the TCS centre case 51% of DCF value sits beyond year 10.
- Nine defensible input sets gave ₹1,630 to ₹3,158 a share: write the range.
- A reverse DCF shows the belief inside the price: about 8.1% growth a year at 12%.
- A price target is a research service under SEBI rules; Learn shows the method only.
Questions readers ask
Which is better, P/E or EV/EBITDA?
Neither. P/E is simple and counts interest and tax. EV/EBITDA compares companies with different debt, but it ignores capital spending, so it flatters businesses that need heavy plant. Read them together, and for a bank use P/B with return on equity instead.
Can I use ₹2,229 as a target for TCS?
No. It is one cell of a grid built on inputs I chose, and a price target is a research service for registered analysts. Rebuild the box with your own inputs and sources, and keep the range.
How often should I redo a valuation?
When the facts change: after each annual report and quarterly result, or when bond yields move a point. The price changes every second. Your inputs should not.
₹100 in Focused Largecap 20 became about ₹321 in the model
Rules and figures in this module were last verified on 19 Oct 2026. Tax and regulatory rules change; check the source before acting.