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/infra-capitalgoods.md

View on GitHub
# Infrastructure, EPC, capital goods and defence — sector playbook

Use this when: the company builds, engineers or equips physical assets to order — EPC and civil contractors, roads/water/urban-infra developers, T&D and railway/metro specialists, industrial and process capital-goods makers, electrical equipment, machinery, defence and shipbuilding PSUs and private primes, and their global analogues (E&C services firms, concession operators, industrial OEMs, defence primes).

This sector breaks a generic checklist at the definitional level, not the benchmark level. Revenue and margin are not observations — they are outputs of management's estimate of cost-to-complete under percentage-of-completion accounting, trued up years later. Most Indian infra groups are two incompatible businesses stapled together (an asset-light contracting arm and a set of 15–30 year concession SPVs), so every consolidated ratio describes a company that does not exist. And the binding constraint on growth is usually not balance-sheet debt but bank-guarantee headroom, which never appears in D/E. Analyse this sector as: **order book quality → working capital → cash conversion → guarantee headroom**, with margin percentage treated as the least reliable number on the page.

All ranges below are **indicative only**. They move with the capex cycle, contract mix, client type, commodity regime and country. A company's own 5–10 year history and its closest sub-sector peers override every absolute band in this file.

## 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

Four structural reasons, then the metric-by-metric damage.

**(1) Revenue and margin are accounting estimates, not facts.** Under Ind AS 115 / IFRS 15 (and ASC 606) over-time recognition — usually cost-to-cost input method — revenue and margin are derived from management's estimate of total contract cost-to-complete. A 200bps "margin improvement" can be a spreadsheet change rather than an operating event. Reported OPM is an *opinion* that gets corrected years later, often through an onerous-contract provision. Treat any margin move that is not corroborated by cash collection as unproven.

**(2) Margin is not comparable across companies.** Reported OPM is a function of contract structure, not efficiency: item-rate vs lump-sum turnkey (LSTK); how much client-supplied free-issue material (steel, cement, cable) sits inside the scope; how much of the work is subcontracted pass-through; whether price-variation/escalation clauses exist. A contractor with 40% pass-through content shows 6% OPM on identical economics to a peer showing 12%. Developed markets solve this explicitly — Fluor, AECOM, WSP, Stantec report **Net Service Revenue** (revenue net of pass-through subcontractor and procured cost) and compute margin on NSR. Indian companies rarely disclose the split, so raw OPM comparisons across large diversified EPC, T&D, roads and water contractors are close to meaningless without a scope-mix adjustment. **EBITDA per unit of capital employed and cash conversion beat margin percentage.**

**(3) Consolidated ROCE and D/E blend two different businesses.** The EPC arm may earn 20–25% ROCE; a HAM/BOT SPV earns a contracted 12–16% equity IRR back-ended over 20 years with near-zero accounting return in years 1–5. The blended figure describes no real business. Split into **core/standalone EPC ROCE** (EBIT of contracting ÷ [net working capital + net fixed assets used in EPC], *excluding* investments in and loans to SPVs) and **asset-level project/equity IRR**. The same split breaks D/E: SPV debt is non-recourse and self-liquidating against a contracted annuity; parent recourse debt is not. A 2.0x consolidated D/E can be materially safer than a 0.8x one. Equity-accounted JVs — very common in EPC consortia and defence offset structures — keep both revenue and capital entirely off the face of the statements.

**(4) D/E and net debt understate true leverage by design.** EPC runs on **non-fund-based limits**: performance bank guarantees, advance BGs against mobilisation advances, retention BGs, LCs. These routinely run 1.5–3x fund-based debt, never appear in D/E, and convert instantly into funded debt on invocation. Add customer mobilisation advances (interest-free customer financing), acceptances / channel finance / supply-chain finance parked in trade payables instead of borrowings, and corporate guarantees issued to SPVs. Interest during construction is capitalised, so interest coverage is flattered too.

Metric by metric:

- **OPM / EBITDA margin** — an estimate, and non-comparable across contract structures (above). Never rank contractors on it.
- **ROCE** — arithmetically valid but economically meaningless on a consolidated infra group. Recompute on core EPC capital only.
- **EV/EBITDA** — usable for concession assets and product capital goods; wrong for the consolidated group, and wrong for E&C services where acquisition intangible amortisation should *not* be added back (use EV/EBITA there).
- **D/E and net debt/EBITDA** — systematically understated; blind to BGs, LCs, corporate guarantees, acceptances and mobilisation advances. Compute recourse and non-recourse separately, then add a third line for off-balance-sheet contingent exposure.
- **FCF** — structurally negative for a *healthy growing* contractor, because every new order consumes working capital before it produces cash. Positive FCF frequently signals a shrinking order book, not quality. Screening on FCF here selects for decline. Use cumulative CFO/EBITDA instead.
- **Current ratio** — inflated by contract assets and retention money that may never convert to cash, and distorted by contract liabilities (mobilisation advances) that are customer financing, not a normal payable. It says almost nothing about liquidity; BG headroom does.
- **P/E** — near-useless. Earnings are lumpy and cycle-dependent; a single arbitration award or claim recognition can be 30–60% of PAT; loss years from onerous contracts are common; and trailing P/E looks cheapest at peak-cycle margin. This is the classic value trap that caught investors in 2007–08 and again in 2018.
- **DSO on trade receivables** — understates the true cycle by 50–100%, because the largest receivable buckets are **contract assets (unbilled revenue)** and **retention money**, which sit outside trade receivables.
- **P/B** — meaningless for asset-light EPC (book value is mostly stranded working capital), but the *primary* anchor for concession assets. Same ratio, opposite validity, inside one consolidated entity.
- **Revenue growth comparability** — for HAM projects the financial-asset model makes construction revenue and "finance income" behave differently from a toll asset's revenue. Headline revenue and EBITDA can fall while economics improve. Growth rates are not comparable across contract types.

## The metrics that actually matter

Ranges are indicative and sub-sector-specific; judge intra-segment and against the company's own history.

| Metric | Definition / how to compute | Indicative healthy range | Why it matters |
| --- | --- | --- | --- |
| **Order book (backlog) and OB/TTM revenue** | Unexecuted value of signed contracts ÷ trailing-twelve-month revenue = years of revenue visibility. Disclosed in DM as backlog or **RUPO / remaining performance obligations** (ASC 606, IFRS 15), often split into ≤12 months and beyond. | Indian EPC 2.5–3.5x TTM revenue (~30–42 months). Below 2.0x = revenue risk inside 12 months. Above 4.5x = execution-bandwidth or working-capital constraint, or a padded book. DM E&C runs leaner at 1.2–2.5x (shorter, services-heavy contracts). Defence order books run 5–10x+ on multi-year platform programmes — do not compare across. | The single best leading indicator; revenue is a lagging, accounting-derived output. But backlog is disclosed at *contract value*, not at margin, so it is only as good as its quality. |
| **Order inflow growth and book-to-bill** | New orders awarded in the period ÷ revenue recognised in the same period. Track on a rolling 4-quarter basis to strip lumpiness. | >1.0x sustained; >1.2x in an upcycle. Two consecutive years below 0.9x means the backlog is being consumed and revenue will roll over. | Book-to-bill turns before revenue, before margin and before the stock. It also exposes who holds pricing discipline in a hot bidding market (inflow dips) versus who buys revenue at bad prices. |
| **Order book quality and mix** | Decompose the backlog: (a) fixed-price LSTK vs item-rate/cost-plus; (b) % carrying price-escalation/price-variation clauses; (c) government vs private vs export client; (d) concentration — top 5 clients and top 5 projects as % of backlog; (e) **slow-moving / non-moving orders** — awarded but no appointed date, no financial closure, no right-of-way, or stayed by litigation; (f) vertical mix (roads, water, T&D, buildings, urban, oil & gas, defence) with their differing margin and working-capital profiles. | Slow-moving <10% of backlog and explicitly disclosed. Single project <10–15% of backlog. Meaningful escalation-clause share in a commodity-inflation environment. Diversification across ≥3 verticals. | Two identical backlogs can have completely different economics. Fixed-price without escalation transfers commodity and wage risk to the contractor; a road order without right-of-way or appointed date is a liability, not an asset. Companies rarely volunteer this — mine concall transcripts and the annual report. |
| **Net working capital days (correctly defined)** | (Trade receivables + contract assets/unbilled + retention money + inventory − trade payables − acceptances − contract liabilities/mobilisation advances) ÷ revenue × 365. **Never use plain debtor days.** | Best-in-class Indian EPC 60–100 days; acceptable 100–130; stress zone >150 and rising. Contractors working for central agencies (NHAI, PGCIL, Railways) sit lower; state-PWD, irrigation and municipal-water exposure runs 180–300 days. Product capital goods often run near zero or negative. | This *is* the sector's return driver: working-capital days × revenue ≈ capital employed, so core ROCE is essentially a function of this number. A 30-day deterioration on a growing top line can consume an entire year's profit in cash. State-government working capital is the single most common cause of Indian infra blow-ups. |
| **Unbilled revenue (contract assets) as % of revenue, and its growth vs revenue growth** | Work recognised under POC but not yet invoiced, as a share of annual revenue; tracked against revenue growth itself. | Typically 15–30% of revenue for Indian EPC. The test is directional: unbilled should grow no faster than revenue. Unbilled growing ~2x revenue growth for two consecutive years is a hard warning. | Unbilled is the standard vehicle for aggressive POC recognition and for parking unapproved variation orders and claims. It is profit the client has not yet agreed to. When it converts to a write-off it hits equity directly. |
| **Cash conversion: cumulative CFO ÷ cumulative EBITDA over 3–5 years** | Sum CFO over a rolling 3–5 year window ÷ sum of EBITDA over the same window. Single-year figures are noise here. | >60–70% cumulative for a growing contractor; >80% for a mature one. Below 40% over five years while PAT compounds is a strong signal that reported profit is not economic. | The most reliable lie-detector for POC accounting: margin is an estimate, cash is not. Essentially every major Indian infra distress case showed years of PAT growth alongside structurally weak CFO/EBITDA before it broke. |
| **Total leverage including non-fund-based exposure** | Three numbers, not one: (a) standalone/recourse net debt ÷ standalone EBITDA; (b) consolidated net debt ÷ EBITDA with SPV non-recourse debt shown separately; (c) outstanding BGs + LCs + corporate guarantees as a multiple of net worth, and as % of sanctioned non-fund limits utilised. | Recourse net debt/EBITDA <2.0x (<1.5x comfortable, >3.0x stressed). Recourse net debt/net worth <0.5x for a pure EPC. Non-fund-based outstanding at 1.5–3.0x fund-based debt is normal; **utilisation of sanctioned non-fund limits above ~85% is a hidden growth cap** — the company cannot bid new work without a fresh sanction. | The binding constraint on a contractor is usually BG headroom, not balance-sheet debt: every bid needs an EMD and performance BG. Groups have failed with "low" D/E because BGs were invoked. This line also captures acceptances/supply-chain-finance dressing that moves debt into payables. |
| **Core EPC ROCE vs consolidated ROCE, and capital locked in subsidiaries** | EBIT of the contracting business ÷ (net working capital + net fixed assets used in EPC), excluding investments in and loans to concession SPVs. Separately: investments + loans + guarantees to subsidiaries as % of consolidated net worth, plus **pending equity commitment** on won HAM/BOT/transmission projects. | Core EPC ROCE >18–20% for a good contractor; very large diversified players do 15–18% on a much bigger base. Capital locked in concessions ideally <30–40% of net worth for a company marketed as an EPC play. Pending equity commitment should be fundable from 2–3 years of internal accruals without dilution. | This is the fork in the road for the whole sector. EPC is high-return, capital-light and cyclical; concessions are low-return, capital-hungry and bond-like. Blending them destroys the analysis and historically destroyed the companies — the classic failure mode is a good contractor bidding aggressively for BOT work so its own construction arm gets the job, then drowning in SPV equity calls. |
| **Claims, arbitration and variation-order exposure** | Claims filed / awards received but under challenge; amounts already recognised in the books (inside unbilled revenue or "other current assets"), as % of net worth and % of PAT. India: NHAI conciliation-committee status, awards under Arbitration Act s.34/s.37 challenge, and amounts released against BG under the 2016 government scheme (75% of award). DM: unapproved change orders and claims disclosed in the 10-K. | Recognised claim assets <10–15% of net worth. Any year where claim/arbitration settlements contribute >20–25% of PAT should be treated as non-recurring and stripped from the earnings base. | Claim cycles run 5–12 years through arbitration and appeals, with realisation frequently 30–60% of the filed amount. Companies capitalise claims and recognise income long before cash, then quietly write them off. It is also the easiest place to manufacture a good quarter. |
| **Receivables ageing and retention money** | Trade receivables by ageing bucket (>180 days, >1 year, >3 years) from the **Schedule III ageing schedule** (mandatory in India since FY22), plus retention money held by clients as a separate line and as % of revenue. | Receivables >365 days <10% of gross receivables and not growing. Retention typically 5–10% of contract value, released on expiry of the defect-liability period (usually 12–24 months post completion). | The ageing schedule is one of the few pieces of hard, non-estimated data in an EPC balance sheet. A fat >3-year bucket that never moves is a write-off queue. Retention is real money, but only released if the project is actually completed and defects closed — so it doubles as an execution-quality proxy. |
| **Bid pipeline, hit ratio and L1 position** | Value of tenders bid; win rate (hit ratio); value where the company is lowest bidder (L1) but the award letter has not been issued; tender-to-award conversion time; and the addressable tendering pipeline of key clients (NHAI/MoRTH awarding targets, PGCIL/TBCB transmission tenders, Jal Jeevan Mission, metro and railway capex, MoD acquisition pipeline). | Hit ratio ~10–20% is normal and healthy. L1 pipeline should be disclosed and convert within 3–6 months. A sharply rising hit ratio usually means cheap bidding, not better capability. | The only forward indicator ahead of order inflow, and the clearest read on pricing discipline. A contractor whose hit ratio jumps from 12% to 35% in a hot market has almost certainly bought backlog that becomes a margin problem 24–36 months later. |
| **Pre-qualification capability and single-project execution size** | The largest single project the company is technically and financially pre-qualified to bid — a function of net worth, turnover, BG limits and completed-project credentials — plus its blacklisting/debarment record with government agencies and multilaterals. | A rising single-project PQ threshold over time *is* the structural growth story. Any active debarment by NHAI, a state PWD, PGCIL, Railways, MoD or the World Bank is disqualifying for the affected segment. | This is the sector's real moat and it is entirely non-financial. Large-ticket work (metros, tunnels, marine, nuclear, HV transmission, refinery LSTK, defence platforms) has far fewer qualified bidders and structurally better margins than commoditised roads and buildings, where anyone with modest net worth can bid. Loss of PQ status kills a business faster than leverage does. |
| **Concession-asset KPIs (BOT / HAM / toll / transmission / airport)** | Asset-level operating and financial metrics: toll traffic growth and WPI-linked rate revision; HAM annuity receipts plus interest at bank rate + 3% on the 60% deferred portion; O&M and major-maintenance cost per lane-km; project DSCR; **residual concession life**; project IRR vs equity IRR; PCOD/COD dates vs schedule. | Minimum DSCR ~1.2x, average 1.3–1.5x. HAM equity IRR contracted around 13–16% pre-refinancing, often 16–20% post-refinancing after COD. Mature-stretch toll traffic growth ~1.0–1.5x real GDP growth. | These are finite-life, contractually defined cash flows and must be monitored as project finance, not as a corporate. Residual concession life is the most important single number — a 25-year asset with 6 years left is a depleting bond, not a growth asset. Post-COD refinancing (200–300bps once construction risk is gone) is the biggest value lever; InvIT monetisation is the standard exit. |
| **Product capital goods: aftermarket share, utilisation, operating leverage** | For equipment OEMs rather than contractors: share of revenue from spares, service, retrofits and AMCs; plant capacity utilisation; incremental/decremental EBITDA margin per unit of revenue change; export share; and the sign of working capital (many run negative on customer advances). | Aftermarket/service 20–40% of revenue for a high-quality franchise; incremental EBITDA margin 20–30% in an upcycle. Negative or near-zero working-capital days is the hallmark of pricing power. | Product capital goods and EPC contracting are routinely lumped into one "sector" and must not be. Product companies have brand, installed-base annuity and pricing power; contractors have a balance sheet and a tender. Aftermarket share is what separates a 40x-multiple franchise from a 12x-multiple contractor. |
| **Defence-specific: order pipeline quality, indigenisation and nomination share** | Split backlog into signed contracts vs AoN/RFP-stage pipeline (India: Acceptance of Necessity is *not* an order). Track indigenous content %, share of revenue from nominated vs competitively tendered orders, export share, offset obligations outstanding, and LD exposure on delayed platform deliveries. | Signed-order backlog is what counts; treat AoN pipeline as optionality only. Rising export share and rising nominated share both improve durability. Any programme slipping >2 years should be re-underwritten. | Defence revenue is politically and budget-driven with multi-year lags between announcement and cash. Nominated orders to PSUs carry lower price risk but come with tight cost-audit scrutiny; competitively tendered private orders carry both price and LD risk. Backlog-to-revenue looks spectacular here purely because programme tenors are long — it is not a quality signal on its own. |
| **Execution throughput and physical progress** | Revenue per employee, revenue per unit of gross block or owned equipment fleet, and — where disclosed — physical progress (lane-km, ckt-km, MW, tonnes fabricated) vs POC percentage claimed. | Directional only; the diagnostic is POC percentage running ahead of physical progress or ahead of billing. | The cleanest non-accounting cross-check available. If accounting completion outruns physical completion, revenue is being pulled forward. Also flags contractors adding backlog faster than they add execution capacity — equipment, people and site management are the real bottlenecks. |

## How to value companies in this sector

**Sum-of-the-parts (SOTP) is the default, not an optional refinement.** This is the core Indian-market convention and the main departure from a generic checklist. A single consolidated multiple applied to an infra group is almost always wrong, because it prices a capital-light cyclical services business and a finite-life annuity asset with the same number.

Standard SOTP build-up:

1. **Core EPC business** — EV/EBITDA of roughly 7–12x, or P/E of 12–20x on standalone/core earnings, using **mid-cycle rather than peak margin**. The multiple is justified by order-book visibility, working-capital discipline and segment mix, not by growth alone. Names with >2.5x book-to-revenue, sub-100-day working capital and clean cash conversion earn the top of the band; state-government-exposed contractors on 200-day cycles trade at 6–10x P/E and deserve to.
2. **Concession / BOT / HAM / transmission SPVs** — DCF or NPV of free cash flow to equity over the **remaining** concession life, discounted at cost of equity (roughly 12–15% in India). Terminal value must be zero or the residual transfer value only. A common street shortcut is P/BV on invested equity: 0.8–1.2x for operational HAM/annuity assets, 1.0–1.5x for de-risked toll assets with proven traffic, and below 1.0x (often 0.3–0.6x) for under-construction or traffic-disappointing assets. **Never apply a P/E to a concession** — the asset amortises to zero by design.
3. **Real estate / land bank, listed investments, InvIT and subsidiary stakes** at market or NAV, with a holding-company discount of 20–30% where relevant.
4. **Less net *recourse* debt at the parent.** SPV non-recourse debt is already embedded inside the SPV equity value — deducting it again double-counts.

**Why EV/EBITDA for assets and earnings multiples for services.** Operating infrastructure assets are capital-intensive with heavy D&A and different leverage, so EV/EBITDA neutralises capital structure and depreciation policy. The EPC arm is asset-light and its "capital" is working capital, so earnings-based multiples and core ROCE are more informative there.

**Developed-market conventions differ.**

- **E&C services firms** (AECOM, Jacobs, WSP, Stantec, Fluor, Quanta) are valued on **EV/EBITA** — not EBITDA, because acquisition intangible amortisation is deliberately included for serial acquirers — at roughly 8–14x, and on P/E on adjusted EPS at 15–25x. Margins and multiples are computed on **Net Service Revenue**, excluding pass-through subcontractor cost.
- **Backlog multiples** (EV / backlog, typically 0.2–0.6x) are a cross-check only, never primary — backlog is at contract value, not gross profit.
- **Regulated infrastructure** (transmission, water, regulated-till airports) is valued on **Regulated Asset Base**: EV/RAB, where 1.0x means the market pays exactly the allowed asset base and any premium is a bet on beating the allowed return or on RAB growth. Applies to UK/EU utilities, Indian TBCB transmission and regulated airports.
- **Toll roads and yield assets** globally trade on EV/EBITDA (10–16x for concession operators) plus DCF on remaining life.
- **Defence primes** are valued on P/E and EV/EBITDA with a visibility premium (global primes historically ~15–22x earnings; Indian defence PSUs have re-rated far above that on order-book and indigenisation narratives). Free cash flow conversion and the signed-vs-pipeline split matter more than backlog headline.

**India-specific instruments.** InvITs (transmission and highway vehicles, including NHAI's sponsored trust) are valued on **distribution yield versus the 10-year G-Sec spread** (typically 250–450bps), on NDCF (net distributable cash flow) per unit, DPU growth, and P/NAV — the same logic as FFO/AFFO and cap rates for REITs. This also gives you the right way to value a developer's mature assets: **at the price a yield vehicle would pay, not at book.**

**Multiples to avoid.** P/B on an asset-light EPC (book is mostly stranded working capital); trailing P/E at cycle peak; EV/Sales (revenue is a POC estimate and includes pass-through); and any consolidated DCF that assumes perpetual-growth terminal value while a large share of assets have contractually finite lives.

**Cross-checks worth running.** Replacement-cost and per-unit metrics (EV per lane-km, per ckt-km, per MW, per MTPA of installed capacity) for asset owners; and **market cap ÷ order book** as a sanity screen — a contractor trading at 1.5x its order book on 6% margins is pricing in a decade of flawless execution.

**Product capital goods are valued completely differently.** Quality Indian capital-goods franchises with MNC parentage and high aftermarket share have long traded at 40–70x earnings against 8–25x for global peers, driven by scarcity value, negative working capital, cash-rich balance sheets and domestic-capex-cycle positioning. Do not benchmark these against EPC contractors, and do not assume the Indian premium is available on a developed-market listing of the same parent.

## Peer set construction

The commonest analytical error in this sector is a peer table that mixes business models. Build the peer set on **contract structure and client type**, not on the exchange's sector label.

Splits that must not be mixed:

- **Contractor vs concession owner vs product OEM.** Three different return profiles, capital intensities and valuation frameworks. A company that is all three requires three separate peer sets, one per segment, reassembled in SOTP.
- **Vertical.** Roads, water/irrigation, T&D and transmission, railways/metro, buildings and factories, urban infra, oil & gas and refinery LSTK, marine and dredging, tunnelling, defence. Margins, working-capital cycles, competitive intensity and PQ barriers differ by a factor of two or more across these.
- **Client type.** Central agencies (NHAI, PGCIL, Railways, NTPC, MoD) vs state departments (PWD, irrigation, municipal bodies) vs private capex vs export/EPC-abroad. This single variable often explains more of the working-capital and ROCE difference between two "road contractors" than anything about their operations.
- **Contract structure.** LSTK/fixed-price vs item-rate vs cost-plus vs EPC-with-escalation. Also the pass-through content, which mechanically sets the margin optics.
- **Scale and PQ tier.** A contractor qualified for a single project of a few hundred crore competes in a fundamentally different, more commoditised market than one qualified for multi-thousand-crore tunnels or HV transmission. Do not put them in one table.
- **Defence PSU vs private defence.** Nominated-order PSUs with cost-audit pricing are not comparable to private primes winning competitive tenders, and neither is comparable to a defence-adjacent components supplier.

Normalisation steps before any cross-company comparison:

1. **Rebuild margin on a comparable revenue base.** Where disclosed, strip pass-through/free-issue material to get an NSR-equivalent. Where not disclosed, do not compare margin at all — compare EBITDA per unit of core capital employed and cash conversion instead.
2. **Use standalone/core numbers, not consolidated**, for the contracting comparison; handle SPVs separately.
3. **Match the point in the order-execution cycle.** A company one year into a large new backlog and one in the final year of an old one show different margins for reasons unrelated to quality.
4. **Compare working-capital days on the full definition** (contract assets + retention included), or the comparison is worthless.
5. **Global comparisons:** compare Indian EPC to global E&C services only on cash conversion, book-to-bill and ROCE — never on margin, since NSR-based DM margins are structurally higher on a smaller revenue base.

## Sector-specific red flags

Ordered roughly by how reliably each precedes trouble.

1. **Unbilled revenue / contract assets growing materially faster than revenue for two or more consecutive years.** The single most common vehicle for aggressive POC recognition and for capitalising unapproved variation orders and claims.
2. **Cumulative CFO below 40–50% of cumulative EBITDA over 3–5 years while reported PAT compounds.** The gap is where the accounting is.
3. **Margin improvement driven by a "change in estimate" of total contract cost-to-complete**, with no corresponding improvement in collections or working-capital days. Look for the change-in-estimate disclosure, and for POC percentage jumping ahead of physical progress or billing.
4. **PAT propped by claims, arbitration settlements, interest on delayed payments, land sales, asset-monetisation gains or other income.** Strip these out and check whether core EPC EBIT is actually growing.
5. **Order book padded with LOIs, MOUs, framework agreements, in-principle approvals, AoN-stage defence pipeline, or L1 positions not yet awarded**; failure to disclose slow-moving orders (no appointed date, no financial closure, no right-of-way, stayed by court); silent cancellations appearing only as an unexplained backlog reduction.
6. **Hit ratio spiking while new-order margins fall** — buying backlog to sustain a growth narrative. The damage surfaces 24–36 months later as onerous-contract provisions, by which time the stock has usually re-rated *upward* on order-inflow headlines.
7. **Acceptances, channel finance, vendor finance or supply-chain finance parked in trade payables rather than borrowings.** Detect via a sudden jump in payable days combined with an implied interest rate (finance cost ÷ average reported debt) far above the company's actual borrowing rate.
8. **BG and LC outstanding rising much faster than revenue; non-fund-based limit utilisation above ~85%; corporate guarantees to SPVs growing** — or, most seriously, an actual **BG invocation or LD levy** by a client.
9. **Investments, loans and advances to subsidiaries and "other current assets" ballooning without explanation**; parent EPC margins booked on construction of the group's own captive BOT/HAM SPVs and not properly eliminated on consolidation — profit manufactured against the group's own balance sheet.
10. **Related-party subcontracting to promoter-owned or promoter-connected entities**, and unusual related-party purchases of materials or equipment hire — the classic siphoning route in Indian infra.
11. **Promoter share pledge rising, promoter stake falling, repeated dilutive QIPs or preferential issues used to fund working capital rather than growth capex**, and reliance on short-term debt to finance long-cycle project working capital (asset-liability mismatch).
12. **Receivables >365 days and >3 years buckets growing and never converting** in the Schedule III ageing schedule; large retention outstanding on projects completed years ago (defects never closed, or the client disputes completion).
13. **Auditor resignation, auditor change without credible reason, qualified opinion, or an Emphasis of Matter on recoverability** of receivables, claims, unbilled revenue or SPV investments. Repeated CFO turnover belongs in the same bucket.
14. **Working-capital days improving sharply only at 31 March and reversing in Q1** — year-end window dressing via bill discounting, factoring or short-dated collection pushes. Compare the March-quarter balance sheet with the September half-year balance sheet.
15. **Migration from asset-light EPC into owning concessions, real estate or unrelated verticals**, especially bidding aggressively for BOT projects primarily to feed the in-house construction arm. This exact pattern preceded the destruction of an entire generation of Indian infra names in the 2008–2018 cycle.
16. **Heavy state-government, municipal or irrigation exposure without a matching working-capital buffer.** Payment cycles routinely run 12–24 months and are budget-dependent, unlike central agencies.
17. **Aggressive capitalisation** — interest during construction, pre-operative expenses and even O&M costs capitalised into concession intangibles; delaying declaration of COD to keep capitalising and defer depreciation.
18. **Fixed-price LSTK backlog with no escalation clause entering a commodity or wage inflation cycle**, with no disclosed hedging or back-to-back supplier lock-in.
19. **Debarment or blacklisting** by NHAI, a state PWD, Railways, PGCIL, MoD, a multilateral lender, or (in DM) suspension/debarment from federal contracting. This removes pre-qualification and can be existential even for a profitable company.
20. **Contingent liabilities section growing faster than the balance sheet** — disputed tax and duty demands, counter-claims by clients, guarantees. In this sector the contingent-liability note is often more informative than the balance sheet itself.

## Cycle and structural context

**Where you are in the cycle changes which metrics to trust.** The sector's cycle is: government/private capex announcement → tendering → order inflow → execution ramp → margin realisation → working-capital strain → payment/claims settlement. Order inflow leads reported revenue by roughly 12–24 months and reported margin by 24–36 months. At the top of the cycle, margin is peak, trailing P/E is lowest, competition for orders is fiercest and bid discipline is weakest — so cheap trailing multiples at peak margin are the sector's characteristic value trap. At the bottom, backlog is thin, balance sheets are repaired and the surviving contractors emerge with better pricing power because weaker bidders have been eliminated.

**Order-inflow is politically and fiscally driven in India.** Central capex allocations, election cycles (award activity typically slows into a general election and accelerates after), state fiscal stress, and specific programme funding (highway awarding targets, Jal Jeevan Mission, metro sanctions, TBCB transmission tendering, defence capital acquisition budget) all move the addressable pipeline more than any company-level factor. Track the client's awarding pipeline, not just the company's order book.

**Structural shifts to weigh.**

- **Model migration in roads:** BOT-toll → annuity → HAM → back toward BOT-toll for select stretches, plus asset monetisation into InvITs. Each shift changes how much equity a contractor must sink per rupee of construction revenue. A contractor that avoided BOT in 2010–14 survived; the model risk is real, not theoretical.
- **Commodity and wage inflation** is the fastest way to destroy a fixed-price backlog. Check escalation-clause coverage before every inflation cycle, and check the *lag* structure of the escalation index versus actual input costs.
- **Energy transition and grid capex** is a durable multi-year driver for T&D, transmission EPC, cabling, switchgear and renewables-linked balance-of-plant, and simultaneously a terminal threat to thermal-power BTG and coal-linked equipment franchises.
- **Import substitution / indigenisation and PLI** shift value toward domestic equipment makers and defence suppliers, but the earnings arrive with long qualification lags and often depend on policy that can be reversed.
- **Labour availability, land acquisition and environmental clearance** are the recurring physical bottlenecks in India; right-of-way delay is the most common cause of a good order becoming a bad one.
- **Regulation and arbitration regime.** In India, changes to the Arbitration and Conciliation Act, the conciliation-committee mechanism and the government scheme releasing 75% of awards against BG materially change claim-realisation timelines and therefore the value of claim assets on the balance sheet. In DM, watch federal infrastructure funding cycles, Buy-America-type content rules and the suspension/debarment regime.
- **Consolidation and PQ inflation.** Larger tender sizes and stricter pre-qualification steadily push work toward bigger balance sheets. This is a structural tailwind for scale players and a structural squeeze on the mid-tier — and it is invisible in any financial ratio.

## India vs global notes

| Dimension | India (NSE/BSE, Ind AS) | US / global (10-K, GAAP/IFRS) |
| --- | --- | --- |
| Revenue standard | Ind AS 115, over-time recognition, usually cost-to-cost. Contract assets/liabilities presented separately. | ASC 606 / IFRS 15, same principle. **RUPO / remaining performance obligations** disclosed with a time-band split — a genuinely useful disclosure Indian filers rarely match. |
| Pass-through disclosure | Rarely disclosed; margins reported on gross revenue including free-issue and subcontract pass-through. | **Net Service Revenue** widely reported by E&C services firms; margins and multiples computed on NSR. |
| Units and scale | Crore and lakh. Order book usually quoted in Rs crore; convert consistently before any cross-border comparison. | Millions/billions of USD or EUR. |
| Concession structures | BOT-toll, annuity, **HAM** (40% construction grant + 60% deferred annuity with interest at bank rate + 3%), TOT, TBCB transmission, airport concessions. NHAI/MoRTH/PGCIL are the dominant counterparties. | PPP/P3, availability payments, DBFOM, regulated-asset-base frameworks in UK/EU; US toll concessions are comparatively rare. |
| Ownership and governance | **Promoter holding** and **promoter pledge** are first-order signals; check quarterly shareholding pattern. Related-party subcontracting is a known siphoning route. | Dispersed ownership; the analogue signals are insider transactions, related-party disclosures in the proxy, and auditor changes via 8-K. |
| Audit disclosures | **CARO** annexure (fixed-asset verification, loans to related parties, statutory dues, fund diversion, whistle-blower matters), Ind AS ageing schedules under Schedule III, and Key Audit Matters — POC estimation and claim recoverability are almost always a KAM. | Critical Audit Matters in the audit report; ICFR attestation; SEC comment letters on revenue-recognition estimates. |
| Management communication | **Quarterly concall transcripts** are indispensable — order book quality, slow-moving orders, L1 pipeline, BG utilisation and claim status are usually disclosed only verbally there. Investor presentations often carry backlog splits absent from the annual report. | Earnings calls plus much richer written segment disclosure; backlog, book-to-bill and NSR margins are typically in the press release itself. |
| Contingent liabilities | Extensive disputed tax/duty demands and client counter-claims in the notes; often larger than net worth. Read this note every year. | Legal proceedings and commitments disclosed in 10-K Item 3 and the notes; generally tighter and more quantified. |
| Yield vehicles | **InvITs** (SEBI-regulated) are the standard monetisation route; valued on distribution yield vs 10-year G-Sec, NDCF/unit, DPU growth, P/NAV. | REIT/YieldCo analogues; infrastructure funds and direct pension/sovereign-fund buyers set the private-market clearing price. |
| Defence | MoD acquisition process (AoN → RFP → trials → contract) with very long lags; defence PSUs get nominated orders; indigenisation content rules and offset obligations. | DoD programme-of-record structure, multi-year procurement, FAR/DFARS cost accounting, and the federal suspension/debarment regime. |
| Multiples | Product capital goods carry a large domestic scarcity premium; contractors trade at low single-to-mid-teens earnings multiples. | Product OEMs and E&C services trade in tighter, more globally arbitraged bands; no equivalent scarcity premium. |

## Checklist

- Split the company into contracting, concessions and products before computing a single ratio; never analyse the consolidated blend.
- Pull order book, order inflow, book-to-bill and OB/TTM revenue for at least 8 quarters; check whether backlog growth is real or definitional.
- Decompose backlog by vertical, client type, contract structure, escalation-clause coverage and top-5 project concentration; quantify slow-moving/non-moving orders.
- Recompute working-capital days on the full definition (receivables + contract assets + retention + inventory − payables − acceptances − contract liabilities). Ignore plain debtor days.
- Chart unbilled revenue growth against revenue growth over 5 years; flag two consecutive years of unbilled outrunning revenue.
- Compute cumulative CFO ÷ cumulative EBITDA over 3 and 5 years. Below 40–50% while PAT compounds = stop and do forensic work.
- Build the three-line leverage picture: recourse net debt/EBITDA, consolidated with SPV debt shown separately, and BG+LC+corporate guarantees vs net worth. Check non-fund-limit utilisation for a hidden growth cap.
- Compute core EPC ROCE excluding SPV investments and loans; separately quantify capital locked in subsidiaries and pending equity commitments on won projects.
- Quantify recognised claim and arbitration assets as % of net worth and % of PAT; strip settlement income out of the earnings base.
- Read the Schedule III receivables ageing schedule; check whether the >1 year and >3 year buckets are converting or accumulating. (India)
- Check retention money outstanding against projects completed more than two years ago.
- Read the last four concall transcripts specifically for BG utilisation, slow-moving orders, L1 pipeline, hit ratio and claim status.
- Compare the March-quarter balance sheet with the September half-year balance sheet for year-end working-capital window dressing.
- Check promoter pledge, promoter stake trend, and the last three equity raises — were they for growth capex or to plug working capital? (India)
- Scan CARO, Key Audit Matters, auditor changes and the contingent-liability note; check for Emphasis of Matter on receivables, claims or SPV recoverability.
- Search for related-party subcontracting, equipment hire and material purchases from promoter-connected entities.
- Check debarment/blacklisting status with every major client agency and multilateral lender.
- For each concession asset: residual concession life, DSCR, COD status, project vs equity IRR, and refinancing/InvIT monetisation potential.
- For product capital-goods names: aftermarket revenue share, working-capital sign, incremental EBITDA margin and capacity utilisation — and value them on a completely separate framework from contractors.
- For defence names: separate signed backlog from AoN/RFP pipeline; check programme slippage, LD exposure and nominated-vs-tendered mix.
- Value via SOTP: mid-cycle EPC multiple + concession DCF over remaining life + investments at NAV − net recourse debt. Never a single consolidated multiple.
- Sanity-check with market cap ÷ order book and per-unit replacement metrics before concluding anything on valuation.
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