README.md
# Business Narrative
Research and structure the qualitative business story of a public company the way Aswath Damodaran does in *Narrative and Numbers*, then translate that story into the valuation inputs a model consumes.
## What it does
- Researches current filings / IR material (never stale memory) into a fact base
- Builds four story pillars: income structure, business model & moat, industry & TAM, growth & reinvestment quality
- Classifies the company's corporate life-cycle stage and the dominant value driver
- Runs Damodaran's possible / plausible / probable test plus a "this time is different" red-flag check
- Outputs a Narrative Brief ending in a **story-to-numbers map** — suggested ranges for growth, margins, reinvestment, risk, and terminal posture, plus method/SOTP signals — that hands off to `company-valuation`
## Triggers
`what's the story on X`, `understand the business`, `business model of X`, `how does this company make money`, `investment narrative`, `bull case / bear case`, `is the growth story credible`, `what's the moat`, `revenue mix and segments`, `TAM for X`, `what has to be true for this to work`. Also runs as Step 2 of `both-stock-analysis`.
## Prerequisites
- Web access for current filings, IR material, and industry/TAM research
- No special libraries; this is a research-and-reasoning skill
## Output
A Narrative Brief: one-line description, four pillars, life-cycle stage, the 3 P's verdict, and the story-to-numbers map (driver → input → default → story-implied range → justifying pillar) with path signal, SOTP signal, and a confidence level.
## Reference Files
- `references/narrative_framework.md` — Life-cycle → driver → story-type mapping, the value-driver bridge in depth, the 3 P's ladder, and the "this time is different" checklist
- `references/research_checklist.md` — What to gather per pillar, the US/Thai primary-source map, the segment-data note, and pre-handoff quality checks
## Disclaimer
For research and educational purposes only. Not financial advice.
references/narrative_framework.md
# Narrative Framework — Story to Numbers (Damodaran)
How to keep a story disciplined and convert it into valuation inputs. Read alongside Steps 3–5 of `SKILL.md`.
## Contents
- The core loop
- Corporate life cycle → dominant driver → story type
- The value-driver bridge (in depth)
- The 3 P's ladder
- "This time is different" red-flag checklist
---
## The core loop
Damodaran's process: develop a narrative about the business → test the narrative (is it possible, plausible, probable?) → convert each surviving narrative element into a value driver → value the company → keep the feedback loop open (let the numbers challenge the story and vice versa). The narrative constrains the numbers; the numbers discipline the narrative. Neither rules alone.
Every story element must terminate at one of four drivers, or it is not part of the valuation:
- **Cash flows** — what the business throws off after taxes and reinvestment.
- **Growth** — how fast revenue/earnings rise.
- **Reinvestment efficiency** — return on the capital reinvested to fund growth (ROIC vs WACC).
- **Risk** — the cost of capital that discounts it all.
If you cannot route a claim ("great management", "strong brand") to a driver (lower cost of capital? higher margin? higher ROIC?), it is rhetoric, not value.
---
## Corporate life cycle → dominant driver → story type
| Stage | Revenue/earnings shape | Dominant driver | Typical story type | Valuation posture |
|---|---|---|---|---|
| **Young / startup** | Tiny revenue, losses | Total market size; survival | "Capture a huge TAM" | Mostly narrative; value is option-like. Demand unit-economics proof. |
| **High growth** | Fast revenue, thin/negative margins | Growth + whether ROIC will clear WACC at scale | "Scale into profitability / the platform wins" | DCF with care; lean on EV/Revenue + Rule of 40; scrutinise the path to ROIC > WACC. |
| **Mature growth** | Decelerating growth, margins firming | Margins, reinvestment discipline | "Durable compounder" | DCF-friendly; the ROIC−WACC spread is the thesis. |
| **Mature stable** | Low growth, high/steady margins, cash return | Cash flow, capital allocation | "Cash cow / dividend & buyback" | DCF + dividend lens; watch capital-allocation discipline. |
| **Decline** | Falling revenue | Asset value, cash extraction speed | "Melting ice cube / liquidation value" | Asset-based and conservative terminal; beware value traps. |
| **Cyclical / commodity** | Swings with the cycle | Normalised mid-cycle earnings | "Buy the trough, sell the peak" | Normalise through the cycle; do not extrapolate peak or trough. |
The stage tells you which numbers to fight over. Arguing margins for a pre-revenue company, or TAM for a cash cow, wastes effort on the wrong driver.
---
## The value-driver bridge (in depth)
For each driver, the question the story must answer, and how it sets the number:
**Growth → revenue growth path.** Decompose into TAM growth × share change × price. A credible growth path fades toward GDP over the explicit window (no company outgrows the economy forever). Set the near-term rate from the story (market growth + share gains), the fade from competitive realism, and the terminal from the home economy's nominal GDP. Red flag: a path that implies the company eventually exceeds plausible market share, or grows faster than the end market.
**Profitability → operating margin trajectory.** Margins are set by pricing power, scale economics, and mix. Expanding margins need a *mechanism*: operating leverage on fixed costs, a shift toward higher-margin segments, or genuine pricing power from the moat. Anchor the ceiling to the best comparable operator — assuming margins above any peer in history is a flag. Mix shift (Pillar 1) is the most defensible source of expansion.
**Reinvestment → capex % and ΔNWC, plus the quality test.** Growth requires reinvestment; the amount is set by capital intensity and growth ambition (sales-to-capital ratio). The *quality* test is decisive: reinvestment only creates value if the incremental return (ROIC) exceeds WACC. Capex rising while FCF also rises signals productive reinvestment; capex rising with FCF flat/falling signals low-return projects. Capital-light models reinvest little and convert more revenue to cash.
**Risk → cost of capital (beta, ERP, country premium).** Business risk (cyclicality, customer concentration, regulatory exposure), operating leverage, financial leverage, and geography set the discount rate. More cyclical / more levered / more emerging-market exposure → higher cost of capital. The country risk premium add-on (per the markets reference in the orchestrator) handles geography. Do not let an exciting growth story quietly lower the discount rate.
**Terminal → terminal growth and terminal margin.** The terminal value encodes what the company *becomes*. A durable compounder holds a positive ROIC−WACC spread into perpetuity (use cautiously — competition erodes spreads); a fading business converges ROIC toward WACC; a cyclical settles at mid-cycle. Cap terminal growth at nominal GDP and never above the risk-free rate.
---
## The 3 P's ladder
Force every key claim down this ladder before it becomes an input:
1. **Possible** — is there any path to this outcome? Filters out the physically/logically impossible only.
2. **Plausible** — is there a credible mechanism and a precedent (this company before, or a comparable elsewhere)?
3. **Probable** — given competition, capacity constraints, and base rates, how likely, and over what horizon?
Most hype dies between "plausible" and "probable." A story can be possible and even plausible while still being improbable — and improbable outcomes do not deserve base-case weight. Assign the optimistic case to the bull scenario, not the base.
---
## "This time is different" red-flag checklist
Flag and push back when the implied numbers require any of these:
- **Impossible share** — revenue that implies the company takes an implausible fraction of (or more than) the total market.
- **Unprecedented margins** — operating/net margins above the best operator the industry has ever produced, with no structural reason.
- **Perpetual hyper-growth** — growth sustained well beyond any comparable company's run, or above the end market indefinitely.
- **Permanent moat** — a positive ROIC−WACC spread assumed to never compress despite the returns inviting competition.
- **Free growth** — growth with no commensurate reinvestment (revenue rises but capital/working capital does not).
- **Risk that vanishes** — a discount rate that drops as the story gets more exciting rather than reflecting real business risk.
For each flag, state what the company would concretely have to do to earn the optimistic number, then decide whether that belongs in the base case or only the bull case.
references/research_checklist.md
# Research Checklist & Source Map
What to gather for the four pillars, where to get it, and how to avoid stale-memory errors. Read alongside Step 1 of `SKILL.md`.
## The cardinal rule
Revenue mix, strategy, guidance, and competitive position change every quarter. Do **not** narrate them from memory. Pull current figures from primary sources, and date every figure you cite. When memory and a filing disagree, the filing wins.
---
## What to gather (mapped to the four pillars)
**Pillar 1 — Income structure**
- Revenue by reportable segment, product line, and geography (latest full year + most recent quarter).
- Gross/operating margin by segment if disclosed; otherwise segment operating profit.
- Year-over-year mix shift (which segment is gaining share of revenue and of profit).
**Pillar 2 — Business model & moat**
- Unit economics: average selling price, volumes, cost structure (fixed vs variable).
- Capital intensity: capex as % of revenue, asset turnover.
- Cash conversion: operating cash flow vs net income, working-capital cycle.
- Moat evidence: pricing trends, customer retention/churn, market-share stability, switching costs, regulatory licences, patents.
**Pillar 3 — Industry & TAM**
- Market size and growth rate (cite the source and its date; triangulate two sources if possible).
- Competitive structure: top players and shares, whether the market is consolidating or fragmenting.
- Regulatory and policy backdrop, especially for banks, telecom, utilities, energy, healthcare.
- The company's share and its trajectory.
**Pillar 4 — Growth & reinvestment quality**
- Stated growth strategy and capital plans (capex guidance, M&A, new markets/products).
- Historical incremental returns: did past reinvestment raise ROIC, or dilute it?
- Funding: is growth self-funded from FCF, or reliant on debt/equity issuance?
---
## Source map
### US listings
- **10-K / 10-Q** (SEC EDGAR) — segments, MD&A, risk factors, the authoritative numbers.
- **8-K** — material events, earnings releases.
- **Investor relations site** — quarterly decks, fact sheets, segment KPIs.
- **Earnings-call transcript** — management's framing of strategy and guidance.
### Thai listings (SET)
- **56-1 One Report** — the annual report + disclosure (the Thai analogue to the 10-K); the primary source for segment revenue, business structure, and risk.
- **SET / SETTRADE company pages** — filings, financial highlights, shareholder structure.
- **Quarterly financial statements + MD&A** filed with the SET.
- **Company IR site** — opportunity-day decks and presentations (often the clearest segment breakdowns).
### Either market
- Web search for current TAM, competitive landscape, regulatory news, and recent strategy — but verify hard numbers against filings.
- Trade publications and regulator sites for industry structure.
---
## Segment data note
yfinance does **not** expose reporting-segment revenue/margins. For Pillar 1 and for the SOTP signal you hand to valuation, segment numbers must come from the 10-K / 56-1 One Report / IR deck. If segments are not separately disclosed, say so — and flag that SOTP is not feasible, so the valuation should rely on consolidated DCF + relative.
---
## Quality checks before handing off
- Every figure has a date and a source.
- The revenue mix sums to ~100% and matches the latest filing.
- TAM and share are internally consistent (implied revenue ≤ TAM).
- You can name the single biggest threat to the moat and the single assumption the story most depends on.
- Confidence (high/low) is set and justified for the downstream scenario ranges.
SKILL.md
---
name: business-narrative
description: >
Research and structure the qualitative business story of a public company the way Aswath
Damodaran does in "Narrative and Numbers," then translate it into the valuation inputs a
model consumes — growth path, margins, reinvestment, and risk. Use this skill before
valuing a company, or whenever the user asks: "what's the story on NVDA", "understand the
business", "business model of CPALL", "how does this company make money", "investment
narrative", "bull case or bear case", "is the growth story credible", "what's the moat",
or "revenue mix and segments". It builds four story pillars (income structure, model and
moat, industry and TAM, growth and reinvestment quality), classifies the life cycle,
applies the possible/plausible/probable test, and outputs a Narrative Brief plus a
story-to-numbers map that hands assumption ranges to company-valuation. Step 2 of the
both-stock-analysis pipeline. Always research current filings and IR material — never
narrate revenue mix, strategy, or guidance from stale memory.
---
# Business Narrative
A number without a narrative is a guess; a narrative without numbers is a fairy tale. This skill builds the disciplined story side of a valuation and hands it, as concrete assumption ranges, to the numbers side. Work in the spirit of Damodaran: every claim about the business must eventually attach to one of four value drivers — **cash flows, growth, reinvestment efficiency, and risk (cost of capital)** — or it is decoration.
The deliverable is a **Narrative Brief** ending in a **story-to-numbers map**. That map is the whole point: it is what makes the downstream DCF assumption-driven rather than default-driven.
**Disclaimer:** Research and educational output only. Not financial advice.
---
## What you produce (output contract)
Hand the next step (valuation) a brief with these parts. The final table is mandatory — it is the handoff.
1. **One-line business description** — what the company sells and to whom, in plain language.
2. **Four pillars** — income structure, business model and moat, industry and TAM, growth and reinvestment quality (sections below).
3. **Life-cycle stage** — young/growth/mature/decline, and what that implies for which driver dominates.
4. **The 3 P's verdict** — is the implied story possible, plausible, probable? Plus any "this time is different" flags.
5. **Story-to-numbers map** — the table that converts the story into suggested inputs and ranges for the value drivers, each justified by a pillar.
Keep it tight (roughly one page of prose + the map). Depth lives in the reference files.
---
## Step 1: Research the company on current sources
Do not reconstruct revenue mix, strategy, or guidance from memory — these change every quarter. Pull current figures from primary sources first: the latest annual report / 10-K (Thai listings: the 56-1 One Report), recent quarterly filings and the MD&A, the investor-relations deck, and the most recent earnings call. Use web search to fill gaps and to read the competitive landscape and TAM.
`references/research_checklist.md` lists exactly what to gather, the Thai/US source map, and how to find segment-level numbers (which yfinance does not expose). Read it now.
The output of this step is a fact base — segment revenue and margins, unit economics, capex, balance-sheet posture, market shares, regulatory exposure. The next steps interpret it.
---
## Step 2: Build the four story pillars
### Pillar 1 — Income structure (where the money actually comes from)
Break revenue into segments / products / geographies, and attach a margin to each. Profit usually concentrates somewhere different from revenue — find it. State: the revenue mix, the gross/operating margin by segment, and which segment is the real profit (and growth) engine. A consolidated number hides the company; the mix reveals it.
### Pillar 2 — Business model and moat
How do unit economics work, and what protects them? Cover pricing power, the moat (network effects, switching costs, scale economies, brand, cost advantage, regulatory licence), capital intensity, and how revenue converts to cash. Be specific about *durability*: a moat that is widening justifies different numbers than one that is eroding. Name the single biggest threat to the moat.
### Pillar 3 — Industry and TAM
Structural tailwinds and headwinds, competitive dynamics (consolidating or fragmenting?), regulatory shifts, the total addressable market, and where the company sits in it (share, and whether share is rising). Distinguish a large TAM the company can actually capture from a large TAM that invites competition and compresses returns.
### Pillar 4 — Growth and reinvestment quality
Where does future revenue come from, and — the Damodaran test — does the company earn returns **above its cost of capital** on the capital it reinvests to get that growth? Growth that earns below WACC destroys value; only growth that out-earns its cost is worth paying for. Tie this to the ROIC−WACC spread the valuation step computes. State whether growth is reinvestment-funded (capex/acquisitions) or capital-light, and whether the reinvestment is productive (capex rising *with* FCF, not against it).
---
## Step 3: Classify the corporate life cycle
Place the company on its life cycle, because the stage tells you which driver carries the value and which numbers to scrutinise. Full mapping (stage → dominant driver → typical story type → valuation posture) is in `references/narrative_framework.md`.
| Stage | What carries value | Watch most |
|---|---|---|
| Young / pre-revenue | Total addressable market, narrative | Survival, funding, unit economics proof |
| Growth | Revenue growth + reinvestment quality | Whether ROIC clears WACC as it scales |
| Mature | Margins, cash return, capital discipline | Moat durability, reinvestment restraint |
| Decline | Asset value, cash extraction | Value traps, terminal assumptions |
---
## Step 4: Run the 3 P's test
State the implied story explicitly, then stress it (Damodaran's possible → plausible → probable ladder):
- **Possible** — could this happen at all? (Almost anything is.)
- **Plausible** — is there a credible mechanism and precedent?
- **Probable** — given competition, capacity, and base rates, how likely is it?
The discipline is forcing a story down this ladder before it becomes a number. Then run the **"this time is different" checklist** in `references/narrative_framework.md`: total-market sizes that imply impossible share, margins above the best operator in history, growth sustained far past any comparable, or a moat assumed permanent. Flag anything the numbers cannot support, and say what the company would have to *do* to earn the optimistic case.
---
## Step 5: Translate the story into numbers (the bridge)
This is the handoff. Convert the narrative into suggested inputs and ranges for the value drivers, and say how each differs from a naive default and why. The downstream `company-valuation` skill defaults to mechanical values (historical-CAGR growth, 3-year-median margins, market beta); your job is to replace those with narrative-driven ranges where the story warrants it.
Produce this table:
| Value driver | Valuation input it sets | Naive default | Story-implied range | Which pillar justifies it |
|---|---|---|---|---|
| Growth | Revenue growth path (5-yr) | Hist. CAGR / analyst +1y | e.g. 8–12% fading to GDP | TAM × share (Pillar 3, 4) |
| Profitability | Operating margin trajectory | 3-yr median | e.g. expand 200–400 bps on mix/scale | Model & moat (Pillar 1, 2) |
| Reinvestment | CapEx % of revenue, ΔNWC | 3-yr median | e.g. elevated while ROIC > WACC | Reinvestment quality (Pillar 4) |
| Risk | Beta / ERP / country premium | Market beta, base ERP | e.g. nudge up for cyclicality/leverage | Model risk, geography (Pillar 2, 3) |
| Terminal | Terminal growth & margin posture | ~GDP | Compounder vs fading vs cyclical | Life cycle (Step 3) |
Also emit two **method signals** for the valuation step:
- **Path signal** — stable cash flows → DCF-friendly; pre-revenue / hyper-growth → lean on EV/Revenue + relative; bank/insurer → P/B, P/TBV; REIT → P/FFO. (Mirrors `company-valuation`'s method-applicability table.)
- **SOTP signal** — if the company has 2+ segments with genuinely distinct economics (different growth, margin, capital intensity), flag SOTP, because a blended multiple will misprice it.
Close with confidence: **high** if the four pillars agree and the 3 P's hold comfortably; **low** if the story leans on a single fragile assumption or fails "probable" — and tell the valuation step to widen its scenario ranges accordingly.
---
## Output format
```
# Business Narrative — [Company] ([Ticker])
**In one line:** [what they sell, to whom]
## Income structure
[revenue mix + margin by segment; where profit really sits]
## Business model & moat
[unit economics, moat type + durability, biggest threat]
## Industry & TAM
[tailwinds/headwinds, competitive structure, TAM, share trajectory]
## Growth & reinvestment quality
[growth sources; does reinvestment earn > WACC?]
## Life-cycle stage
[stage → dominant driver]
## The 3 P's test
[possible / plausible / probable verdict + "this time is different" flags]
## Story → numbers map
[the bridge table + path signal + SOTP signal + confidence]
```
---
## Caveats
- The brief is only as current as the sources — date your figures and prefer primary filings.
- Narratives are accountable to numbers: if a pillar implies a number the company has never achieved, say so rather than smoothing it over.
- Confidence is part of the deliverable; an honest "low" with wide ranges beats false precision.
- Not financial advice.
---
## Reference Files
- `references/narrative_framework.md` — Damodaran's narrative-to-numbers method: life-cycle → story-type → driver mapping, the value-driver bridge in depth, and the "this time is different" red-flag checklist.
- `references/research_checklist.md` — What to gather, the US/Thai primary-source map, where to find segment data, and how to avoid stale-memory errors.