DCF Valuation in India: Skip the Spreadsheet Theatre

The first DCF most people build has forty rows of revenue assumptions, each one tuned over a weekend. Then they nudge the terminal growth rate from 4% to 5%, the discount rate from 12% to 11%, and the answer moves more than every line item they agonised over combined. That is the day the uncomfortable thing about discounted cash flow lands: most of the labour goes into the part that barely moves the number.

A DCF feels like precision engineering. It is closer to a mood ring for two numbers you mostly guessed — and a third number you have no business trusting.

Asian Paints, Infosys and the other companies named below are illustrations of how the inputs behave, not a view on any stock.

What does a DCF actually assume?

Strip away the spreadsheet and a discounted cash flow model says one plain thing: a business is worth the cash it will hand its owners over its life, with future cash discounted back because a rupee in FY35 is worth less than a rupee today.

That logic is sound. The trouble is the inputs. You need three things — the cash the business throws off for the next five to ten years, a number for everything after that (the terminal value), and a discount rate to drag it all back to the present. The first you can study. The other two you largely assert.

Here is the part the tutorials skip. The explicit forecast — those years of revenue, margin and capex you slaved over — is usually the smallest part of the answer. Take Infosys's actual modelled inputs from Inve's own DCF engine: a 3% near-term growth rate, a five-year explicit window, a 5% terminal growth, a 10% discount rate (Inve DCF model, dcf-baseline-v1, through Q4 FY26). Run those numbers and the terminal value — the single figure standing in for everything after year five — works out to roughly 79% of the total. Hindustan Unilever's modelled inputs (5% growth, seven years, 9% discount rate) land at about 77%. The bit that takes nine-tenths of your effort decides one-fifth of the answer.

So the honest reading is this. A DCF's output is not a valuation. It is a bet on two assumptions — what the business is worth forever, and what return you demand — dressed up in forty rows of fake confidence. Move the terminal growth a point, or the discount rate a point, and the "intrinsic value" swings by a fifth or more without you touching a single thing the business actually does.

That is not a flaw to fix. It is the nature of the tool.

Why does terminal value swallow the whole model?

Think of the terminal value as the part of the iceberg under the water. The explicit five-to-ten-year forecast is the tip you can see and argue about. The terminal value is everything below — the assumption that the business keeps compounding cash long after your forecast window ends.

Mathematically it dominates because most of a healthy company's cash flows arrive after your window, and the terminal value is your one-number stand-in for all of them. A small change to a number representing infinity is a large change to the total. That is exactly why Infosys's modelled inputs throw 79% of the weight into the terminal number: a short five-year explicit window and a terminal growth (5%) sitting close to the discount rate (10%) push almost everything past the horizon.

Two constraints fence in that terminal number, and both are easy to violate. First, keep the terminal growth rate and the discount rate on the same basis — both nominal. The discount rates used here (the 9-14% range in this article, and the 11-14% most investors land on) are nominal, so the perpetual growth rate must be nominal too. The ceiling is long-run nominal GDP growth — per the NISM-XV research curriculum, a company cannot compound faster than the economy forever — and India's nominal GDP runs roughly 10-11%. A prudent perpetual rate sits well below that ceiling, which is why Inve's model uses a conservative 5% nominal terminal for these names and drops the cyclicals lower (Patanjali sits at 3%). The 5% is a deliberately cautious choice, not the GDP cap. The trap for a beginner is mixing a real growth rate (say 5-6%) with a nominal discount rate (11-14%): that mismatch will badly understate value, because you are discounting at the inflation-inclusive rate while growing at the inflation-stripped one. Second, the perpetuity only behaves when that growth rate stays strictly below the discount rate (g < k); push them close together and the (k − g) denominator shrinks, which is exactly why a tiny move in either input near that boundary produces an outsized swing.

Watch the two dominant inputs do their work, holding everything operating fixed. The grid below is illustrative — ₹100 crore of free cash flow growing 8% for ten years, then settling into terminal growth, same business in every cell — but the arithmetic is real and reproducible. Only the discount rate and terminal growth move.

Intrinsic value (₹ crore) — terminal growth across, discount rate down (illustrative)

The same company is "worth" anywhere from ₹1,441 crore to ₹2,193 crore — a 52% spread — and we never changed what the business does. We only changed two inputs you cannot observe and can barely defend. Anyone who hands you a single DCF price target with two decimal places is selling false precision. The number after the decimal is theatre.

Isn't the discount rate just a formula?

This is where most retail DCFs quietly cheat. The discount rate is taught as a CAPM calculation — risk-free rate plus beta times an equity premium — which makes it look objective. It isn't. It's a judgment about how much you trust the cash flows, smuggled in as a Greek letter.

Watch what Inve's own model does across the FMCG and IT names it covers, and the judgment shows through the arithmetic. Nestlé India and Hindustan Unilever — slow, predictable, market-leading — get a 9% discount rate. Infosys, low-leverage but with single-digit growth, gets 10%. Asian Paints gets 12% (Inve DCF model, dcf-baseline-v1). Nothing in a textbook formula explains that three-point gap. The rationale field does, in plain words: Asian Paints shows "earnings volatility and mixed track record," so the model assigns "a higher discount rate of 12%" and the lowest confidence of any of the fourteen names it scores. A higher discount rate is not a calculation. It is the model saying I trust these numbers less — and charging for it.

You can see why it earns that distrust in the company's own quarterly print. Asian Paints' standalone net profit ran ₹1,209 crore in the March-2024 quarter, then ₹597 crore in September 2024, then ₹694 crore in March 2025 — roughly halving and recovering on barely-moving sales (Inve data, quarterly financials through Q3 FY26). Discount the future cash of a business whose recent past zig-zags like that, and a single point on the rate is the right place to bury your unease. The number that looks most scientific in the whole model is the one you should hold most loosely.

How do you run a DCF backwards instead?

Here is the reframe that makes the tool useful. Stop asking the DCF "what is this worth?" Ask it the question it can actually answer: "what does the market price already assume?"

That is a reverse DCF. Instead of feeding in your growth guess to get a value, you feed in today's price and solve for the growth and margin the market has already baked in. The implied number is still a range, not a point — it moves with the discount rate and terminal growth you assume — but it gives you a bar. Then you ask one falsifiable question: has this management ever credibly guided, let alone delivered, that bar?

This flips a fragile forecasting exercise into a checkable claim. Forecasting forward, you are one analyst guessing the future against the entire market. Working backward, you let the price state its own assumption, then go to the concall record and test it. The three names below show the test at three altitudes, each with a real guided number to check against.

Reverse DCF — the implied bar vs what management actually guided (illustrative)

Notice the gap in the bottom row. Management guided 8-10% volume; Inve's model fed in 6%, "significantly discounting the high single-digit volume growth to account for… execution risk" (Inve DCF model rationale). That four-point haircut is the whole game — and the next section is why the model was right to take it.

These worked numbers are hypothetical illustrations of how a reverse DCF reads, not a buy, sell or hold view on EMMVEE, Titan, Asian Paints or any other stock. When the market's implied growth runs ahead of what management has actually guided, that gap is a question to go research — does the delivery record support the bar? — not a signal to sell or short. The point of the exercise is to send you to the concall, not to a trade.

Why does the implied-growth question come down to management?

A reverse DCF is only as trustworthy as the guidance you test it against. So look at what guidance is actually worth in the aggregate.

Across 13,280 management commitments tracked over 1,519 listed Indian companies, roughly 930 were ghosted — stated once on a call, then never mentioned again on any later one (Inve data, 2026-06-24). Around 35% of companies — 534 of them — have at least one such commitment that simply went quiet. That is before counting the ones that were missed, quietly diluted, or pushed out a year.

Sit with what that means for your model. If the market is pricing in 14% growth and management has guided 14%, you have not validated the price. You have validated that management said a number. The reverse DCF is the question; management's delivery record is the answer key — and on more than nine hundred commitments across a third of these companies, the answer key has a blank where the answer should be.

This is the part of valuation that no amount of spreadsheet craft can fix. You can model cash flows to the rupee. You cannot model whether the people guiding those cash flows mean it.

What does a guided number actually do over a year?

Watch one travel. In its Q4 FY24 call, Asian Paints set a target of "double-digit volume growth in FY25" (Inve Promise Tracker, ASIANPAINT). A forward modeller plugs in 10% and moves on. Here is what the company itself then did with that number, quarter by quarter, in its own words:

• Q1 FY25: reiterated double-digit — but the quarter came in at 7%.
• Q2 FY25: officially downgraded to "single digit for the full year," after Q2 volume was flat (0%).
• Q3 FY25: single-digit reiterated; the quarter did 1.6%.
• Q4 FY25: full-year volume growth landed at 2.5%.

Double-digit became 2.5%. The modeller who anchored on the opening guide built a thesis on a number the company quietly walked four-fifths of the way down — and never had to formally retract, because each step looked reasonable on its own. That is the trap: guidance is a negotiating posture, not a measurement. Some managements lowball to beat; some open high and fade. The reverse DCF doesn't ask you to trust the number — it asks you to check the market's implied bar against the stated bar and the history of clearing it.

It happens in the other direction too. Infosys opened its FY26 constant-currency revenue guidance at 0-3% in its Q4 FY25 results (the 0-3% opener is on the company's own Q4 FY25 investor materials), then walked it up across the next three quarterly calls — the kind of step-up Promise Tracker logs guide-by-guide from the concall archive, where each quarter's revised band can be checked against the company's quarterly results and investor presentations (available on the exchanges and at infosys.com/investors). A forward DCF built in April 2025 on the initial guide was stale within a few quarters, not because the modelling was bad, but because the input was never a fact. A guided number is a claim with a direction and a track record — and the direction is the tell.

What does management do when the number gets uncomfortable?

It goes quiet. And the quiet is more honest than the guidance.

Back to Asian Paints. In Q4 FY25 management guided "₹700-800 crore per year" of standalone capex for FY26 and FY27. Through Q1 and Q2 FY26 it was reaffirmed and on track. Then in the Q3 FY26 call — eight hundred-odd lines of transcript — the standalone capex figure simply does not appear; the only mention of "capex" is a passing reference to private-sector demand (Asian Paints Q3 FY26 concall). Inve's record marks it ghosted: stated, tracked, then dropped from the conversation (Inve Promise Tracker, ASIANPAINT).

A forward DCF that capitalised that ₹700-800 crore into the asset base and modelled the returns on it is now resting on a number management stopped repeating. You would never know from the model. You would only know from the transcript. That is the asymmetry the reverse DCF is built to exploit — and the forward DCF is built to hide.

Where this argument could be wrong

The honest steelman for the forward DCF is real, so state it plainly. For a genuinely stable, predictable compounder — a Nestlé India, where Inve's model assigns its highest-confidence inputs (9% rate, 7-year window) — the cash flows can be bounded, and a forward DCF run as a set of explicit scenarios (bear / base / bull), not a single point, is a legitimate way to think about a range of value. The terminal-value problem doesn't vanish, but for a business whose volumes barely move, the terminal assumption is far less heroic than it is for a 25%-growth solar manufacturer like EMMVEE. And a reverse DCF is not magic either: it still embeds an assumed discount rate and terminal growth, so its "implied growth" is a band, not a verdict. If you anchor too hard on a single implied number, you have just rebuilt the false precision you set out to escape.

So the claim is narrow, not absolute. A DCF is dangerous when it is sold as a precise answer, and useful when it is run as a question. Both directions can be abused. Neither replaces reading the concall.

So what do you do with a DCF?

Build it. But build it to interrogate the price, not to produce one. Run it backwards, read off the growth the market already assumes, and then do the unglamorous work: pull the last several quarters of guidance and check whether management has ever credibly guided — and delivered — what the price now demands.

A DCF that hands you a single "fair value" to two decimals is theatre. A DCF that tells you "this price needs 16% growth for a decade — and the closest real guide, Titan's 15-20% jewellery CAGR, is a band this management has only sometimes hit, while a peer like Asian Paints opened at double-digit and delivered 2.5%" is an investment thesis. One pretends to know the future. The other turns valuation into a falsifiable bet — the only honest kind.

The owner's question is the one to end on. Forget the price target. Ask: what must I believe about this management's next five years of guidance — and their record of meaning it — for the price I'm paying to make sense? If you can answer that from the concall record rather than the spreadsheet, you have done the real valuation. The rest is theatre.

If you want to see whether a company's guided numbers actually held — the walk-ups, the quiet drops like Asian Paints' vanished capex line, the questions management dodged on the call — that delivery record is what Promise Tracker surfaces, pulled straight from the concall archive, so you are not re-reading eight transcripts by hand. The point isn't a grade. It's whether the growth your reverse DCF needs has ever shown up.

For deeper checks on the inputs themselves, it pays to read the cash flow statement properly and to know which management guidance is worth trusting before you anchor a model on it.

Frequently asked questions