Skip to content

alirezarezvani/claude-skills / finance/skills/stock-analysis

stock-analysis

Unverified

Analyze, research or value a listed stock or company (US, global, NSE or BSE) with verified data, ratios, scores and valuation. Use whenever asked if a stock is a buy, what it is worth, or how a company looks.

FileKindSize
SKILL.mdskill md32.7 KB
evals/evals.jsonconfig13.2 KB
evals/fixtures/synthetic-ar-excerpt.mdmarkdown5.7 KB
evals/fixtures/synthetic-drhp-excerpt.mdmarkdown3.4 KB
examples/README.mdmarkdown1.9 KB
examples/forensic-analysis-example.mdmarkdown8.8 KB
examples/standard-analysis-example.mdmarkdown15.6 KB
references/01-data-sourcing.mdmarkdown38.5 KB
references/02-core-factors.mdmarkdown49.3 KB
references/03-earnings-quality.mdmarkdown40.1 KB
references/04-balance-sheet-and-cashflow.mdmarkdown56.6 KB
references/05-returns-and-dupont.mdmarkdown49.8 KB
references/06-valuation.mdmarkdown53.7 KB
references/07-forensic-red-flags.mdmarkdown57.5 KB
references/08-governance.mdmarkdown67.0 KB
references/09-risk-and-macro.mdmarkdown52.7 KB
references/10-peer-set.mdmarkdown45.5 KB
references/11-scoring-rubric.mdmarkdown36.8 KB
references/12-report-template.mdmarkdown26.6 KB
references/13-situations.mdmarkdown60.6 KB
references/14-accounting-comparability.mdmarkdown48.6 KB
references/15-document-diligence.mdmarkdown76.1 KB
references/16-market-mechanics-and-tax.mdmarkdown48.6 KB
references/17-process-and-epistemics.mdmarkdown59.6 KB
references/18-forensic-mode.mdmarkdown16.5 KB
references/19-ipo-mode.mdmarkdown15.6 KB
references/20-challenge-pass.mdmarkdown14.1 KB
references/21-data-integrity-tools.mdmarkdown7.2 KB
references/sectors/_index.mdmarkdown67.8 KB
references/sectors/auto.mdmarkdown47.5 KB
references/sectors/aviation-hotels.mdmarkdown49.3 KB
references/sectors/banks.mdmarkdown38.6 KB
references/sectors/biotech-clinical.mdmarkdown47.8 KB
references/sectors/chemicals-cement.mdmarkdown46.2 KB
references/sectors/exchanges-payments.mdmarkdown41.2 KB
references/sectors/fmcg-consumer.mdmarkdown35.7 KB
references/sectors/holdco-assetmgr.mdmarkdown41.4 KB
references/sectors/infra-capitalgoods.mdmarkdown43.0 KB
references/sectors/insurance-brokers-services.mdmarkdown49.9 KB
references/sectors/insurance.mdmarkdown38.0 KB
references/sectors/it-saas.mdmarkdown44.3 KB
references/sectors/metals-mining.mdmarkdown49.4 KB
references/sectors/mortgage-reit-specialty-finance.mdmarkdown43.5 KB
references/sectors/nbfc.mdmarkdown40.3 KB
references/sectors/oil-gas.mdmarkdown43.5 KB
references/sectors/people-businesses.mdmarkdown46.3 KB
references/sectors/pharma-healthcare.mdmarkdown40.6 KB
references/sectors/rail-freight.mdmarkdown66.2 KB
references/sectors/realestate-reit.mdmarkdown44.3 KB
references/sectors/retail-ecommerce.mdmarkdown48.3 KB
references/sectors/semiconductors.mdmarkdown43.2 KB
references/sectors/shipping-logistics.mdmarkdown47.2 KB
references/sectors/telecom-media.mdmarkdown43.1 KB
references/sectors/utilities-power.mdmarkdown44.9 KB
references/sectors/waste-environmental.mdmarkdown46.5 KB
scripts/benchmarks.jsonconfig200.5 KB
scripts/lint_report.pyscript47.8 KB
scripts/ratios.pyscript127.4 KB
scripts/score.pyscript96.8 KB
scripts/valuation.pyscript37.9 KB
scripts/verify_data.pyscript72.5 KB

references/sectors/telecom-media.md

View on GitHub
# Telecom, towers, broadcasting, media and OTT — sector playbook

Use this when: the company sells connectivity or audience attention — mobile and fixed-line operators, tower and fibre infrastructure vehicles and InvITs, cable/DTH distribution, TV broadcasters, radio, print, film production and exhibition, music/IP libraries, and streaming (SVOD/AVOD) platforms.

These are licence-gated, capital- or content-intensive subscription and advertising businesses in which almost every line of a generic screener is driven by accounting policy, regulation or the position in the investment cycle rather than by underlying economics. Telecom is an invest-then-harvest business where the return profile is structurally awful for four to six years after each technology cycle and then jumps; media is a business where the single largest cost — content — flows through the P&L on a schedule management chooses. In both, the operational KPIs (ARPU, churn, active-subscriber ratio, tenancy, ratings share, paid subs) lead the financials by two to four quarters, so they carry more information than the current income statement. Nothing in this sector can be analysed on a consolidated basis: a group containing mobile, towers, DTH, broadcast and a cash-burning OTT arm must be pulled apart before a single ratio is computed.

## Contents

- [Why the generic ratio set fails here](#why-the-generic-ratio-set-fails-here)
- [The metrics that actually matter](#the-metrics-that-actually-matter)
- [How to value companies in this sector](#how-to-value-companies-in-this-sector)
- [Peer set construction](#peer-set-construction)
- [Sector-specific red flags](#sector-specific-red-flags)
- [Cycle and structural context](#cycle-and-structural-context)
- [India vs global notes](#india-vs-global-notes)
- [Checklist](#checklist)

## Why the generic ratio set fails here

Do not report any of the following without the correction stated alongside. Several of them do not merely add noise — they invert.

**P/E and EPS — near-useless, frequently undefined.** Telecom carries enormous non-cash depreciation and amortisation: network assets on 10–25 year lives, plus spectrum amortised over the 20-year licence term. An operator can be strongly cash-generative and still post accounting losses for years; several large listed telcos have done exactly that through a 4G/5G build. Media is worse, because content and film libraries are amortised on management-chosen curves, making EPS a policy output rather than an observation. A consolidated media P/E also blends a profitable linear broadcast business with a streaming arm that is loss-making *by design* during scale-up. Negative or violently volatile EPS makes P/E undefined, and where it is defined it usually ranks the cycle rather than the business.

**OPM / EBITDA margin — not comparable across companies or across time.** Since IFRS 16 / Ind AS 116 (FY20 in India), tower and fibre lease rentals moved out of opex and into depreciation and lease interest, inflating reported telecom EBITDA margins by roughly 800–1,200 bps with zero change in cash flow. The perverse consequence: an operator that sold its towers and leased them back now reports a *higher* margin than one that owns them, despite identical or worse economics. Any cross-company or pre/post-2019 margin comparison is invalid unless computed on **EBITDAaL** (EBITDA after lease costs), now the standard European reporting convention. Mix compounds the problem — an integrated telco with handset/device sales, wholesale carriage or DTH is structurally lower-margin than a pure mobile play, with no difference in quality.

**Revenue is not clean either (India-specific).** Reported revenue is gross revenue, while licence fee (8% of AGR including USOF) and spectrum usage charge are levied on **Adjusted Gross Revenue** — a regulatory cost wedge of roughly 8–11% of service revenue with no developed-market equivalent. Comparing an Indian telco's margin to a US or EU carrier's without adjusting for this compares regulatory regimes.

**ROCE / ROE — distorted at both ends, and pro-cyclical in the wrong direction.** The denominator includes spectrum (an intangible with no standalone earning power), right-of-use lease assets, and in media large acquisition goodwill. The numerator is depressed during multi-year build-outs. Because the sector is invest-then-harvest, trailing ROCE is structurally poor for years after each technology cycle and then steps up — so screening on trailing ROCE systematically buys the top of the cycle and sells the bottom. Where accumulated losses have driven book equity negative, ROE is not low; it is mathematically meaningless.

**D/E — misleading in both numerator and denominator.** Book equity is corrupted by accumulated losses, revaluations and past capitalisation of forex differences. More importantly, headline borrowings routinely exclude the largest real obligations: deferred spectrum payment liabilities to the government, AGR dues, lease liabilities and vendor/supply-chain financing. Replace D/E entirely with **fully-loaded net debt / EBITDAaL** rebuilt from the notes.

**FCF — swings from deeply negative to strongly positive with no change in business quality.** Network capex is lumpy and spectrum auctions are episodic. You must separate maintenance capex from growth capex, and treat spectrum outlays as a distinct quasi-regulatory item rather than as capex. Use **(EBITDAaL − capex) / EBITDAaL** as the operating cash conversion measure and keep spectrum on its own line.

**Current ratio — inverted.** Prepaid telecom runs structurally negative working capital because customers pay in advance. A current ratio well below 1.0 is normal and is a sign of a good business model, not distress. The same is true of subscription media with annual prepayment.

**Asset turnover, inventory turnover and P/B — meaningless for the asset-light half of the sector.** For a broadcaster, music label or production house, the real asset — the IP library, the ratings franchise, talent relationships — is either fully amortised or was never capitalised. P/B on such a company measures acquisition history. Conversely, telecom asset turnover of 0.3–0.6x is normal for the second-most capital-intensive sector in the market and reads as "inefficiency" to a generic screen.

**Dividend payout and yield — invert across the capex cycle.** Telcos suspend or cut dividends during spectrum and technology-build years and pay out heavily in harvest years, so the yield signal is highest exactly when the reinvestment need is lowest and vice versa. Do not treat a suspended dividend during a 5G build as a quality signal in either direction.

**Any consolidated ratio on a group.** Mobile, towers, DTH, enterprise/data-centre, broadcast, exhibition and OTT have different margins, different capital intensity, different cyclicality and different multiples. Analyse and value them separately or do not analyse them at all.

## The metrics that actually matter

Ranges below are **indicative only**. They vary by market, market structure (number of competitors), regulatory regime, technology cycle and reporting period, and any of them can be wrong for a specific company in a specific year. Comparison against a tight peer set, against regulator-published industry data (TRAI in India, FCC/Ofcom/BEREC elsewhere) and against the company's own eight-to-twenty-quarter history overrides every absolute band here.

| Metric | Definition / how to compute | Indicative healthy range | Why it matters |
|---|---|---|---|
| **ARPU and its trend/composition** | Service revenue ÷ average subscribers, monthly. Split prepaid/postpaid, mobile/fixed, consumer/enterprise. Check every quarter for definitional changes (inclusion or exclusion of interconnect, device, or enterprise revenue). | India ₹180–250/month post the 2024 tariff repair, with management ambition materially higher; US postpaid $45–55; Western Europe €10–16. **Direction matters more than level** — sustained 8–12% YoY ARPU growth is the entire Indian telecom thesis. | In a saturated market with 85–100% SIM penetration, subscriber growth is over and essentially all incremental EBITDA comes from ARPU. Incremental revenue on an already-built network drops through at 70–85%, making ARPU the highest-operating-leverage variable in the model. It is also the cleanest read on whether oligopoly pricing discipline is holding. |
| **Monthly churn rate** | Disconnections per month as % of average base. Split prepaid vs postpaid and, where disclosed, voluntary vs involuntary. | India prepaid 2.0–3.5%/month normal, >4% signals distress; postpaid (India/US) 0.8–1.2%; FTTH/broadband 1.0–1.5%; pay-TV/DTH 1.5–2.5%; standalone SVOD 3–6%. | Churn sets subscriber lifetime (≈1/churn) and therefore acquisition payback. At 2% monthly churn a subscriber lasts ~50 months; at 4%, ~25 months — halving lifetime value while acquisition spend is unchanged. Rising churn alongside rising ARPU tells you a tariff hike is being rejected. Never read the two separately. |
| **Active subscriber ratio (VLR %) and quality of net adds** | Visitor Location Register subscribers — those actually active on the network — as % of reported subscribers. **Published monthly by TRAI in India, per operator per circle.** Elsewhere, use the reported-subs vs revenue-generating-units gap. | India 90%+ is strong; the strongest operators run 96–100%, weak ones 80–88%. Watch the trend at least as much as the level. | Reported subscriber counts are the most gameable number in telecom — inactive SIMs, multi-SIM users and dealer-loaded connections inflate the base and mechanically depress ARPU. VLR% is an independent, regulator-published cross-check on whether headline net adds are real paying customers. India is the only major market with this disclosure; use it aggressively. |
| **Revenue market share vs subscriber market share (RMS vs SMS)** | Share of industry AGR/service revenue versus share of industry subscribers, by circle. TRAI publishes circle-level AGR quarterly in India. | RMS > SMS indicates a premium subscriber mix; the leading Indian incumbent runs several hundred bps of RMS above its SMS. A widening RMS–SMS gap marks a share gainer in value rather than volume. | Subscriber share rewards whoever gives away the most; revenue share is the only share that converts into profit. With ~80% fixed costs, the operator gaining RMS compounds a margin advantage. Circle-level RMS also reveals where an operator has genuine network density versus where it is subscale and structurally loss-making. |
| **Capex intensity, split maintenance / growth / spectrum** | Cash capex ÷ revenue, with spectrum payments broken out separately (spectrum is a licence cost, not network capex). | Steady-state telecom 13–18% of revenue; active 5G or FTTH build 22–30%; sustained above 30% means the business is not self-funding. Media/broadcast 3–6% — in media, **content spend replaces capex** and should be measured instead. | Determines whether EBITDA ever becomes shareholder cash. Two telcos with identical EBITDA margins are worth very different amounts if one needs 15% of revenue to stand still and the other 28%. Capex intensity is also where earnings management hides (see red flags on capitalised opex). |
| **EBITDAaL margin (EBITDA after lease costs)** | EBITDA less right-of-use depreciation and lease interest — i.e. the pre-IFRS 16 / Ind AS 116 economic margin. Most European operators and Indian sell-side models publish it; if not disclosed, rebuild it from the lease note. | Indian mobile: 40–48% reported EBITDA ≈ 33–40% EBITDAaL. Developed-market mobile 30–38% EBITDAaL. Towercos 55–70% (a different business). Cable/broadband 40–50%. | The only margin comparable across operators with different tower-ownership structures and across the IFRS 16 boundary. It also keeps EV/EBITDA honest — if you use reported post-IFRS-16 EBITDA you must add lease liabilities to EV, and mixing the two conventions is the single most common valuation error in the sector today. |
| **Fully-loaded net debt / EBITDAaL** | Net debt **including** deferred spectrum liabilities to the government, AGR dues, lease liabilities, vendor/supply-chain financing, and any perpetual or hybrid instruments — divided by EBITDAaL for consistency. | Below 2.5x comfortable; 2.5–3.5x manageable for a stable incumbent with contracted cash flows; above 4x is a solvency question, not a valuation question. Towercos and infra vehicles sustain 4–6x on contracted revenue. | Telecom failures are always leverage failures, never demand failures: the assets are long-lived and revenue is sticky, so what kills operators is a maturity wall meeting a spectrum auction. Deferred spectrum and AGR dues are legally senior government obligations with hard moratorium expiry dates; excluding them understates leverage by turns, not decimals. |
| **Operating FCF conversion: (EBITDAaL − capex) / EBITDAaL** | Share of operating profit surviving the network reinvestment requirement, before interest, tax and spectrum. | Mature developed telco 45–60%. Indian telco mid-5G-cycle 20–40%, rising above 50% as the build completes. Persistently below 20% excluding growth capex means the business is a treadmill. | Replaces FCF yield in a sector where reported FCF is dominated by lumpy spectrum. It answers the only question that matters for a capital-intensive business: after paying to keep the network competitive, is anything left? It is also the input to EV/OpFCF, the primary developed-market telecom multiple. |
| **Data usage per subscriber and network/plan mix** | GB per user per month; % of base on 4G/5G; postpaid share of base; fixed-broadband attach or converged-household share. | India 25–33 GB/user/month, the highest globally, growing 15–20% YoY; developed markets 15–25 GB. Postpaid mix: India 5–8%, US 55–70%, Europe 40–60%. 5G handset share above 30–40% of the base is typically where monetisation begins. | Data growth is the demand driver that eventually forces either capacity capex or tariff increases — the tension between the two *is* the telecom investment cycle. Postpaid and 5G mix are the leading indicators of ARPU: moving a prepaid user to postpaid roughly doubles ARPU and cuts churn by more than half, raising revenue and lowering acquisition cost simultaneously. |
| **Subscriber acquisition cost and payback months** | Gross SAC (handset subsidy + dealer commission + activation + marketing) ÷ gross adds; payback = SAC ÷ (ARPU × contribution margin). Cross-check against whether SAC is expensed or capitalised. | Payback under 12 months is strong; 12–24 months acceptable for postpaid/FTTH with low churn; beyond 24 months the model depends on churn staying low, which it rarely does. | Ties churn and ARPU into a single return-on-marketing test. It is also the item most often quietly capitalised — if SAC moves to the balance sheet, EBITDA rises with no change in cash. Compare SAC × gross adds against the reported selling and marketing line to detect this. |
| **Homes passed, homes connected, penetration of homes passed** | Premises the fibre/cable network can serve, premises actually taking service, and the ratio. Also track cost per home passed and per home connected. | Penetration of homes passed: 30–40% is roughly breakeven, 45–60% a good return, below 25% after three years means the build was mis-targeted. Indian FTTH ARPU ₹500–700 against build cost of ₹8,000–15,000 per home passed requires high take-up. | Fixed-line economics are entirely a penetration story: the cost is incurred when the street is passed and the incremental cost of connecting a home is small. Returns are hyper-sensitive to take-up, not to ARPU. Companies love to disclose homes passed (an input they control) and stay vague about homes connected (the output that determines returns). |
| **Tower tenancy ratio, EBITDA per tower, contract structure** | Tenants per tower; cash EBITDA per tower per month; plus weighted-average remaining contract life, escalators, exit penalties and tenant concentration. | Tenancy 1.7–2.2x India, 2.0–2.5x US towercos; below 1.5x the asset is uneconomic. Each incremental tenant carries roughly 90% incremental margin. Contracted revenue backlog of 5+ years is the norm. | A tower is a fixed-cost asset whose entire profitability is the tenancy ratio — the second and third tenants are almost pure profit. Tenancy trend also tells you whether operators are consolidating networks (bad) or densifying for 5G (good). Tenant concentration is a counterparty risk no financial ratio captures: a towerco leaning on a financially stressed anchor tenant faces write-offs and forced renegotiation, not a tenancy problem. |
| **Content cost as % of revenue, and the amortisation policy** | Cash content/programming spend vs the P&L content amortisation charge, plus the disclosed amortisation curve (accelerated vs straight-line) and assumed library life. | Indian general-entertainment broadcasters 30–40% of revenue; sports-rights businesses 60–90%+ and structurally lower-margin; global streamers 45–60%. Cash spend and amortisation should be roughly neutral over a full cycle. | Content is the media equivalent of capex, but unlike capex it flows through the P&L on a schedule management chooses — making the cash-spend vs amortisation gap the single largest lever on reported media profit. Comparing two broadcasters on EBITDA margin without comparing amortisation policy compares accounting policies, not businesses. |
| **Revenue mix: advertising vs subscription; ad growth vs nominal GDP** | Split of media revenue between cyclical advertising and recurring subscription/carriage; ratio of ad revenue growth to nominal GDP growth over 5+ years. | Advertising grows 1.0–1.3x nominal GDP over a cycle but falls 15–30% in a downturn. Subscription share above 50% materially de-risks the model. Indian TV broadcasters have historically run 60–75% advertising; print 65–75% advertising plus circulation. | Advertising has near-100% operating leverage in both directions and two-to-three times the cyclical amplitude of GDP, so an ad-heavy media company must be valued on mid-cycle earnings, never trailing. In India the subscription line is additionally regulated — TRAI's New Tariff Order capped channel pricing and bouquet discounts, structurally compressing it, a policy risk with no developed-market analogue. |
| **Viewership/ratings share and ad-rate realisation (CPM/GRP)** | Share of TV and digital viewership by genre and language (BARC in India, Nielsen in the US), plus effective CPM or rate per GRP achieved. Track over 8+ quarters. | Genre leadership matters far more than aggregate share: the #1 channel in a genre commands a 20–40% ad-rate premium over #2, and #3 onwards is often loss-making. Stable-to-rising share in core genres is the requirement. | Media is winner-takes-most at the genre level — advertisers buy reach, so ratings share converts non-linearly into revenue share. A broadcaster losing ratings reports flat revenue for two to three quarters because rate cards lag, then falls off a cliff. Ratings are the leading indicator the financials will not show for a year. |
| **Streaming/OTT unit economics** | **Paying** subscribers (not MAU, registered users or downloads), monthly ARPU, annual content spend ÷ average paying subs, and segment contribution margin after content and marketing. | Content cost per paying sub must trend below roughly half of annual ARPU for the model to work. Scaled global streamers run content near 45% of revenue; Indian SVOD ARPU of ₹50–150/month against globally-priced content is why most of it is structurally loss-making. Contribution margin should turn positive within 4–5 years of launch. | Streaming inverts media economics: content cost is fixed and global, revenue is per-subscriber and local, so the business only works above a scale threshold — below it, losses grow *with* subscribers. Companies substitute MAU and engagement metrics for paid subs precisely when paid conversion is failing. Insist on the paying number, the ARPU and the segment margin, and check the OTT arm is not buried inside a profitable broadcast segment. |
| **Spectrum holdings, expiry schedule and cost per MHz** | MHz held by band and circle/market, weighted-average remaining licence life, deferred payment schedule and moratorium expiry dates, and price paid per MHz per capita versus recent auction clearing prices. | No material expiry inside 3 years without a disclosed, funded renewal plan. Sub-1GHz (700/850/900 MHz) coverage spectrum is worth a multiple of mid-band per MHz because it determines rural and in-building coverage economics. | Spectrum is the sector's recurring solvency event and its main source of value destruction — auction-inflated prices are how operators end up earning below cost of capital for a decade. Spectrum depth also determines capex: an operator short of mid-band spectrum must build more sites for the same capacity, permanently raising capex intensity. |

## How to value companies in this sector

Value at the **enterprise level, on cash-flow-based multiples, with a fully-loaded EV** — and normalise to mid-cycle. Trailing numbers value the peak of the ad cycle or the trough of the capex cycle.

**EV/EBITDA (on EBITDAaL) is the anchor for telecom**, for the same reason it is for utilities: enormous depreciation makes earnings meaningless, capital structures vary wildly, and enterprise value is the only level at which spectrum-heavy, debt-laden and debt-free operators are comparable. Two disciplines are non-negotiable:

1. Use **EBITDAaL**, or if you use reported post-IFRS-16 EBITDA, **add lease liabilities into EV**. Mixing the two is the most common valuation error in the sector.
2. Build a fully-loaded EV: market cap + net debt + **deferred spectrum liabilities + AGR dues** + lease liabilities + minorities − market value of listed stakes.

Indicative multiple ranges (they move with rates and cycle; treat as orientation, not targets): developed-market incumbents 5–7x EV/EBITDAaL, US carriers 6–8x, Indian telcos 9–13x (a growth premium reflecting the ARPU-repair thesis and a consolidated three-plus-one market structure), towercos 10–15x, Indian broadcasters 10–16x, print 4–7x, exhibition 7–10x on EV/EBITDAaL.

**EV/OpFCF — EV ÷ (EBITDAaL − capex) — is the sharper telecom multiple** and the one European specialists actually trade on, because it penalises operators whose EBITDA is expensive to maintain. Anything cheap on EV/EBITDA but expensive on EV/OpFCF is a capex trap; run both, always.

**DCF is more defensible here than in almost any other sector and should be the primary intrinsic method.** Subscriber revenue is contractually recurring and highly predictable, asset lives are long, and terminal value is genuinely reachable. Model it as a **subscriber build** — subs × ARPU, with churn, capex intensity and an explicit spectrum renewal schedule — rather than as percentage-of-revenue assumptions. Discount at a WACC reflecting actual leverage, and separately test whether **spectrum-inclusive ROIC exceeds WACC**: the sector's long-run failure mode is earning below the cost of capital on auction-inflated spectrum, and a DCF that ignores the next auction is a DCF of a business that will not exist.

**Sum-of-the-parts is mandatory wherever the entity is not a single-market mobile operator.** Indian structures make this unavoidable — an integrated incumbent may combine India mobile, a separately listed African business, a stake in a listed towerco, an enterprise/data-centre arm, a DTH business and a partly-owned regional subsidiary, each on its own multiple. The largest Indian mobile business exists only inside a diversified conglomerate and must be valued at an implied EV/EBITDA within a group SOTP. Media conglomerates similarly need broadcast, films, music IP and OTT valued on different bases.

**Value loss-making OTT arms separately and honestly** — on EV/Sales against a credible path to contribution margin, or on LTV/CAC, or as a *negative* (a quantified cash drain with option value). Folding a structurally loss-making streaming business into a group EBITDA multiple either destroys the multiple or hides the burn; both are wrong.

**EV per subscriber and EV per MHz (or MHz-pop) are the replacement-cost / transaction benchmarks.** Use them for distressed or loss-making operators where every earnings-based multiple is undefined. They set a floor and are how sector M&A is actually priced. **EV per tower and EV per tenancy** do the same for infrastructure; **EV per screen** for exhibition; **EV per home passed** for fibre.

**Infrastructure vehicles — towers, fibre InvITs, data centres — are valued on yield, closer to REIT convention than to telecom convention:** distribution or AFFO yield, and EV/EBITDA benchmarked as a spread over the risk-free rate. In India these increasingly sit in InvIT structures where the operative test is distribution yield versus the 10-year G-sec spread, plus a line-by-line read of the NDCF bridge.

**Media splits by sub-sector:**
- *Mature, low-capex, high-payout broadcasters* — the one corner of this sector where **P/E and dividend yield genuinely work**, provided you have checked content amortisation policy and related-party content sourcing first.
- *Content and IP owners (music, film libraries)* — library NPV plus a slate DCF; the back catalogue is an annuity, the new slate is a portfolio of options.
- *Streaming and digital-first media* — EV/Sales against a path to contribution margin, or LTV/CAC. Current earnings are negative by design.
- *Exhibition* — EV/EBITDAaL and EV per screen; IFRS 16 lease liabilities dominate the balance sheet and must be in EV.
- *Print* — value the cash and property separately from the declining operating business; a low P/E on a structurally shrinking circulation base is not cheapness.

**What NOT to use:** P/E on any loss-making or heavily-amortising telco; P/B on asset-light media; reported EV/EBITDA without a lease convention decision; EV/EBITDA on an entity with a large loss-making segment inside the consolidated EBITDA; dividend yield on a telco mid-capex-cycle; trailing multiples on any advertising-exposed business.

## Peer set construction

Compare like with like or do not compare. The sub-sector splits below must **never** be mixed in a single peer table.

- **Mobile operator vs integrated telco vs pure fixed/cable.** Device and handset revenue, wholesale carriage and DTH all dilute margin without diluting quality. Where possible compare on service revenue only, excluding equipment.
- **Telecom operator vs towerco vs fibre InvIT vs data centre.** These are three different asset classes: an operator is an operating business with technology risk; a towerco is a contracted-cash-flow annuity valued like infrastructure; an InvIT is a yield instrument. Their EBITDA margins (40% vs 60%+ vs 90%) and their multiples are not comparable in either direction.
- **Market structure is part of the peer definition.** A three-player market with tariff discipline and a five-or-six-player market with a price war are different industries. Never benchmark an operator in a consolidating market against one in a fragmenting one.
- **Position in the market matters more than country.** A #1/#2 operator with scale and RMS above SMS has structurally different economics from a subscale #3/#4 operator, which is usually loss-making at the circle/regional level regardless of national reporting. Compare leaders to leaders.
- **Lease convention and reporting standard.** Verify every peer is on the same EBITDA definition (pre- or post-IFRS 16). Pre-FY20 Indian data is pre-Ind AS 116 and is not comparable to post-FY20 without restatement.
- **Regulatory-cost regime.** Indian operators bear licence fee and SUC on AGR; most developed markets do not. Adjust before comparing margins across borders.
- **Media: split by revenue model and genre.** Ad-led general entertainment, subscription-led niche/sports, regional-language broadcasters, print, radio, music IP, film production, exhibition and OTT are separate peer sets. A regional-language broadcaster with genre leadership and a national broadcaster with fragmented share are not comparables even at similar revenue.
- **Sports-rights businesses are their own category.** Content cost at 60–90% of revenue, cliff-edge rights renewal risk, and lumpy multi-year amortisation make them non-comparable with general entertainment.
- **Streaming: split by ownership.** A standalone SVOD, a studio-owned streamer subsidised by a library, and a telco-bundled OTT have completely different acquisition costs and churn. Bundled OTT subs are not equivalent to directly-acquired paying subs.
- **Own history is the strongest peer.** For a company whose market structure has changed (a merger, an entrant exiting, a tariff repair), pre-change history may be a worse comparator than current peers. Say which you are using and why.

## Sector-specific red flags

**Leverage and obligations**
- Headline "net debt" that excludes deferred spectrum payment obligations, AGR dues, lease liabilities or vendor financing. Indian operators have historically presented debt ex-spectrum; these are legally senior government dues with hard moratorium expiry dates. Always rebuild leverage yourself from the notes.
- Tower/fibre sale-and-leaseback or InvIT monetisation presented as **deleveraging**. Cash debt falls, lease liabilities rise, and an owned asset has become a permanent rental obligation, often at a worse economic cost. Check total obligations, not reported borrowings.
- Spectrum or licence expiry inside two to three years with no disclosed funding plan, or management calling a renewal auction "manageable" without quantifying it. Spectrum is the sector's recurring solvency event.
- Negative net worth sustained only by promoter or government support, or by conversion of dues into equity. At that point conventional ratios are undefined and the work is a solvency-and-dilution analysis, not a valuation.

**Subscriber and revenue quality**
- Rising subscriber base with a falling VLR/active ratio, or a quiet change in the definition of "subscriber" (moving from a 30-day to a 90-day activity window). This is the sector's most common cosmetic fix: it flatters net adds and deflates ARPU simultaneously.
- ARPU "growth" delivered by shedding low-value subscribers rather than by pricing. Read ARPU alongside **absolute service revenue** — if ARPU rises while service revenue is flat or falling, the base is shrinking and the improvement is arithmetic, not commercial.
- Growing gross adds alongside rising churn — churning the same customers through the front and back door inflates reported adds while acquisition cost is expensed repeatedly. Read gross adds, net adds and churn together, never in isolation.
- Registered users, downloads, MAU or "reach" substituted for paying subscribers and audited circulation, precisely when paid conversion is deteriorating.

**Accounting**
- A step-up in capitalised network costs, capitalised interest, or capitalised subscriber acquisition costs (handset subsidies, dealer commissions). Moving opex into capex lifts EBITDA immediately and defers the charge into depreciation years later. Watch for capex rising while network deployment metrics (sites added, km of fibre laid) do not.
- Lengthening content amortisation lives, or switching from an accelerated to a straight-line curve. Also watch cash content spend running persistently far above the P&L amortisation charge — the library is either being built or the charge is understated, and only the impairment test will eventually tell you which.
- Presenting post-IFRS 16 / Ind AS 116 EBITDA alongside peers' pre-standard figures, or celebrating margin expansion in the adoption year. The 800–1,200 bps uplift is pure accounting.
- "Other income", treasury and investment gains, spectrum trading gains, or asset disposal profits carrying the reported profit line. Strip to core operating cash generation.
- Receivable days above 100–120 in media — government advertising receivables in print, agency receivables in broadcast — and "unbilled revenue" growing faster than revenue.
- Revenue recognised from barter, ad-for-equity or advertising-for-stake deals, common among media houses with startup clients. Non-cash revenue booked at full margin that reverses when the counterparty fails.

**Segment and governance (India-weighted)**
- A loss-making OTT/digital business consolidated into a profitable broadcast segment with no separate segment disclosure. Insist on segment-level EBITDA; the **absence** of the disclosure is itself the signal.
- Related-party content sourcing — film, music or programming rights bought from promoter-owned production houses without a demonstrated arm's-length test. Check group loans and advances to promoter entities buried in "other non-current assets"; the mid-cycle collapse of a large Indian media group followed exactly this pattern.
- High or rising promoter share pledging in Indian media and telecom holding structures, which converts a business problem into a forced-selling problem.
- Concentration risk in infrastructure: a towerco or fibre company with a large revenue share from a financially stressed anchor tenant. The real risk is receivable write-offs and contract renegotiation, not tenancy trends.

**Leading indicators of trouble**
- Ratings/viewership share declining while revenue holds up. Rate cards lag ratings by two to three quarters, so financials look fine right until they do not. Treat BARC/Nielsen share loss as a leading indicator that **overrides** current earnings.
- Regulatory changes treated as one-offs when they are structural resets: TRAI's New Tariff Order on channel pricing, AGR definition changes, licence-fee or USOF revisions, must-carry and carriage-fee rules.
- Capex guidance cut sharply in a competitive market. It buys short-term FCF and cash flow at the cost of network quality, which shows up as churn 18–24 months later.

## Cycle and structural context

**The telecom capex cycle drives everything.** Each generation (3G, 4G, 5G, next) follows the same arc: spectrum auction → heavy build with collapsing ROCE and negative FCF → coverage parity, price war and consolidation → tariff repair and margin expansion → harvest with high FCF and dividends → next auction. Where a company sits on that arc determines which metrics are informative. In the build phase, ignore ROCE, P/E and FCF and focus on capex efficiency, net debt trajectory and funding cover. In the harvest phase, ignore the flattering trailing multiples and ask when the next auction is and what it will cost.

**Consolidation is the value event.** Telecom margins are a function of the number of credible competitors. Markets going from five-plus players to three tend to see ARPU repair and multiple expansion; markets where a well-funded entrant appears see the reverse, sometimes for a decade. Track competitor count, competitor funding capacity, and whether the weakest player is being kept alive by government forbearance — a subscale competitor unable to invest but unable to exit is the worst structure for the whole industry.

**Structural threats to media are not cyclical.** Linear TV viewing and print circulation are in secular decline in most markets; advertising is migrating to a small number of global digital platforms that capture the majority of incremental ad spend. Do not model a broadcaster's ad revenue as merely cyclical without testing whether the audience base itself is eroding. Cord-cutting in developed markets, and the shift from pay-TV to bundled OTT in India, permanently compress carriage and subscription revenue for distributors.

**Telecom's own structural threats:** OTT messaging and voice destroyed the voice and SMS revenue lines a decade ago and the sector has been repricing data ever since; satellite direct-to-device and fixed wireless access are the current edge cases. Net-neutrality regimes prevent operators capturing application-layer value, which is why telecom persistently earns less than the ecosystem it enables.

**Regulation is a primary earnings variable, not a footnote.** Spectrum auction design and reserve prices; licence fee and revenue-share definitions; interconnect/termination rates; the New Tariff Order in India; must-carry and carriage rules; content and broadcasting ownership caps; data-protection and content-takedown regimes; and merger approval thresholds. Any of these can reset the revenue base permanently. Read the current consultation papers, not just the enacted rules — the regulator's direction of travel is usually visible a year ahead.

**Rates matter more than for most sectors.** These are long-duration, highly leveraged, yield-held assets. Rising rates hit valuation twice: through the discount rate on very long cash flows and through refinancing cost on structurally high debt. Infrastructure vehicles and InvITs trade explicitly on a spread to the sovereign.

**Content cost inflation is the media cycle.** Sports rights and premium content costs ratchet up with each renewal while advertising is cyclical, so media margin compression is often a rights-renewal event rather than a demand event. Check the renewal calendar for every material rights contract.

## India vs global notes

**Regulators and disclosure.** India: TRAI (tariffs, subscriber and AGR data, New Tariff Order for broadcasting), DoT (licences, spectrum, AGR assessment), MIB (broadcasting and content), and BARC for TV ratings. TRAI publishes monthly subscriber and VLR data and quarterly performance indicator reports with circle-level AGR — a level of independent operational disclosure no other major market provides. Use it as the primary cross-check on company-reported subscribers, ARPU and market share. Globally: FCC and Nielsen in the US, Ofcom in the UK, BEREC/national regulators in the EU, with disclosure in 10-K/10-Q (EDGAR) or annual reports and much less granular regulator-published operating data.

**India-specific cost and liability structure (no developed-market analogue).**
- **AGR / licence fee:** 8% of Adjusted Gross Revenue including USOF, plus spectrum usage charges. This is a revenue-share tax on the top line.
- **AGR dues:** the Supreme Court's ruling on the definition of AGR created large retrospective liabilities with defined instalment schedules. These are senior government dues and belong in net debt.
- **Deferred spectrum payments:** auction purchases are paid over 16–20 years after a moratorium. Headline "net debt" often excludes them; you must add them back, with the moratorium expiry date noted as a cliff.
- **New Tariff Order (broadcasting):** TRAI caps on channel à-la-carte pricing and bouquet discounting structurally compressed broadcaster subscription revenue and reset the distributor–broadcaster split. There is no US or EU equivalent.

**Accounting.** Ind AS 116 mirrors IFRS 16 (mandatory FY20 in India), so Indian pre-FY20 EBITDA is not comparable to post-FY20. US GAAP retains an operating-lease P&L charge for ASC 842 operating leases, so US carriers' reported EBITDA is *already* closer to EBITDAaL than an IFRS peer's — a direct cross-border comparability trap. Content amortisation policy is disclosed under ASC 926 for US film/TV entities with useful specificity; Indian disclosure is typically thinner, so read the significant-accounting-policies note and the concall Q&A.

**Units and conventions.** Indian filings use crore and lakh; ARPU is quoted monthly in rupees, while US carriers quote monthly ARPU/ARPA in dollars and European operators often quote it excluding VAT. India reports subscribers by **circle** (22 licensed service areas) — circle-level economics vary enormously and a national average conceals subscale, loss-making circles. Indian quarterly reporting is not required to include full segment cash flows; use the annual report, CARO reporting on related-party and pledge disclosures, and the earnings call transcript, where operational KPIs (churn, VLR, 5G rollout, content spend) are usually disclosed only verbally.

**Ownership and governance.** Indian telecom and media are promoter-controlled; check promoter holding, pledge percentage, and the structure of holding companies between the promoter and the listed entity. FDI in telecom is permitted to 100% (with security-clearance conditions) and broadcasting has sector-specific caps; a change in foreign ownership rules is a live corporate-action variable. Related-party content transactions are the dominant Indian media governance risk — check the RPT schedule and the audit committee's approval language, not just the aggregate number.

**Market maturity.** India runs the world's highest data usage per user at among the world's lowest ARPUs, so the investment case is a price-repair thesis with enormous operating leverage. Developed markets run the reverse: high ARPU, saturated demand, and the investment case is cost efficiency, convergence bundling and capital return. Do not import a target multiple or a healthy-range band across that divide without stating the adjustment.

## Checklist

- [ ] Identify sub-sector precisely (mobile / integrated / fixed / tower / InvIT / broadcast / print / music-IP / exhibition / OTT) and refuse consolidated ratios on multi-segment groups.
- [ ] Recompute EBITDAaL; state the lease convention used and confirm every peer is on the same one.
- [ ] Rebuild net debt fully loaded: borrowings + deferred spectrum + AGR dues + lease liabilities + vendor financing + hybrids; compute net debt/EBITDAaL.
- [ ] Pull ARPU and absolute service revenue together — confirm ARPU growth is pricing, not base shrinkage.
- [ ] Read churn and ARPU jointly; compute implied subscriber life (1/churn) and acquisition payback.
- [ ] For India, cross-check reported subscribers against TRAI VLR % and circle-level RMS vs SMS.
- [ ] Split capex into maintenance, growth and spectrum; compute capex/revenue and (EBITDAaL − capex)/EBITDAaL.
- [ ] Map the spectrum expiry and deferred-payment schedule for the next five years, with the funding plan.
- [ ] For fixed/fibre, get homes connected, not just homes passed; compute penetration of homes passed.
- [ ] For towers, get tenancy ratio, contracted backlog, escalators and tenant concentration; assess anchor-tenant credit.
- [ ] For media, compare cash content spend against the P&L amortisation charge and read the amortisation policy note for changes.
- [ ] Split media revenue into advertising vs subscription; value ad-exposed earnings on mid-cycle, never trailing.
- [ ] Check genre-level ratings share over 8+ quarters; treat share loss as overriding current revenue.
- [ ] For OTT, demand paying subs, ARPU, content cost per paying sub and segment contribution margin — reject MAU substitutes.
- [ ] Confirm the loss-making digital arm is separately disclosed; absence of segment disclosure is itself a flag.
- [ ] Test for capitalised opex: capex rising without matching network build metrics; SAC moving to the balance sheet.
- [ ] Value on EV/EBITDAaL **and** EV/OpFCF; investigate any gap between the two.
- [ ] Run a subscriber-build DCF with explicit churn, capex intensity and spectrum renewal; test spectrum-inclusive ROIC vs WACC.
- [ ] Build an SOTP for any group with listed stakes, towers, DTH, enterprise or OTT; value infra vehicles on distribution yield vs the sovereign.
- [ ] India: check promoter pledge, related-party content sourcing, group loans in "other non-current assets", and the concall for KPIs absent from the filings.
- [ ] State the market structure (number of credible competitors, direction of consolidation) and where the company sits in the capex cycle before quoting any multiple.
add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks.Stars: 115kUpdated Claude CodeCodexVerifiedInstall