The 4-Layer Indian Stock Research Stack: Ratios to Concalls

Opening four copies of the same ratio table can feel like thorough research. It leaves a gap if none of those tabs checks the operational assumptions, statutory footnotes or older guidance behind the numbers.

An Indian stock research stack is more useful when each source has a distinct job. Historical accounts tell you what the business earned. Operating disclosures show what is happening inside it. Footnotes expose qualifications hidden by the headline numbers. A guidance history keeps earlier expectations visible when the next call changes the subject.

The last job has a measurable reason to exist. Inve's production snapshot of 4 September 2026 covered 24,328 commitments across 1,648 Indian companies. Of 16,842 commitments with a verdict other than new, 2,271 were classified ghosted — 13.5% of those 16,842 eligible commitments, or 9.3% of all 24,328 tracked commitments. Another 7,486 remained new. Eligible is not the same as finally resolved: an on_track item can still be awaiting its deadline.

These classifications describe the parsed call record, not all company disclosures. The stack needs the other documents precisely because a tracker can be wrong.

Companies and platforms below illustrate a research workflow, not investment recommendations.

Four layers, four different questions

Think of assessing an older factory building. A structural survey, an electrical inspection and a title search answer different questions. Paying for three structural surveys does not settle the title.

These are layers of evidence, not exclusive product categories. Screener.in supports custom ratios and screening; Trendlyne lists a wider collection of screening and research features. A company presentation can serve more than one layer. Buying more subscriptions is not the objective.

The post-screener research workflow concentrates on the hand-off from a shortlist to a guidance review. This article is about assembling the whole document stack without treating one source as a substitute for another.

Layer 1: historical financials need matching scope

Begin with the business economics. Compare returns on capital, borrowing, profit and cash conversion over a period appropriate to the business. A single strong quarter should not erase a weak multi-year record.

Keep consolidated and standalone figures separate. Compare like periods. Check whether a margin excludes other income or exceptional items. A polished table is only as useful as those definitions.

For non-financial businesses, cumulative operating cash flow alongside profit can expose a build-up of receivables or inventory. There is no universal rule that every company must convert exactly 60%–70% of EBITDA into operating cash flow. Taxes, working capital, growth and the business model matter. Banks and NBFCs also need a separate framework; their lending and funding flows cannot be read like a manufacturer's cash cycle.

A shortlist is permission to investigate further. It is not a list of businesses certified as clean. When researching Indian midcaps alongside US stocks, that discipline matters more than whether the financial table comes from a domestic or global terminal.

Layer 2: an order book is evidence, not booked revenue

Interarch Building Solutions illustrates the extra question operating disclosures can answer. Inve's financial rows show revenue of ₹364 crore in Q3 FY25, ₹464 crore in Q4 FY25 and ₹381 crore in Q1 FY26. The sequential dip in the last figure is real, but it does not by itself describe the work still to be executed.

The extracted order book rose from ₹1,305 crore in Q3 FY25 to ₹1,646 crore in Q4 FY25 — about 26.1%. Subsequent disclosures were ₹1,695 crore as of 31 July 2025, ₹1,634 crore in Q2 FY26 and ₹1,685 crore in Q3 FY26 (Inve data, financial rows and operating-KPI extracts). The July figure was reported with Q1 results; it should not be relabelled a June quarter-end balance.

Revenue then reached ₹491 crore in Q2 FY26 and ₹523 crore in Q3 FY26, approximately 52.0% and 43.7% above the corresponding ₹323 crore and ₹364 crore quarters a year earlier. The guidance history moved from 20% FY26 sales growth in Q3 FY25, down to 17.5% in Q4 FY25, back toward 20% in Q2 FY26 and to ₹1,900 crore in Q3 FY26.

That sequence gives you something to investigate beyond a choppy revenue chart. It does not prove that every order will convert on time or at the expected margin. The operating-KPI guide explains why execution dates, cancellations and profitability belong beside the headline order book.

For lenders, distinguish an actual asset-quality measure from a management target. Repco Home Finance reported Stage-2 assets at 9.7% and GNPA at about 3.3% in Q1 FY26. Its guidance was to reach 2.5% GNPA and 7%–8% Stage-2 assets by year-end; 2.5% was not the Q1 actual. The Q4 FY26 ledger and KPI updates record GNPA of 2.55% and Stage-2 assets around 7% (Inve data).

Stage 2 refers to a significant increase in credit risk, not simply a universal 31–90-days-overdue bucket. The IFRS 9 standard uses a rebuttable presumption after payments become more than 30 days overdue; forward-looking credit-risk information also matters. Read the lender's own classification policy before comparing peers.

Layer 3: the annual report can change the meaning of a margin

Centum Electronics makes the case for keeping the annual report open next to the call. On its Q2 FY25 call, held on 14 November 2024, management discussed a consolidated EBITDA-margin objective of 14%–15% over the next two to three years. Its FY25 annual report described an 8.3% consolidated EBITDA margin for the year. One is an aspiration; the other is a reported outcome.

The FY25 annual report also supplies details a margin headline cannot. Form AOC-1, printed page 47, reports the two Canadian subsidiaries' after-tax losses as ₹113 million and ₹10 million, equivalent to ₹11.3 crore and ₹1.0 crore. The table is in rupee millions; its CAD column identifies the subsidiaries' functional currency. Reading those losses as CAD amounts would be a major unit error.

The same report lists disputed income taxes of ₹92.06 million versus ₹46.44 million a year earlier, and disputed indirect taxes of ₹116.53 million. These are contingent exposures, not amounts that can simply be deducted as certain future cash payments. The guide to contingent liabilities and notes explains how to read that distinction.

Now add the call sequence. Inve's historical record shows the long-term margin target unmentioned in Q3 FY25, recut to 11%–12% in Q4 FY25, restated at 13%–15% in Q1 FY26, then unmentioned in Q2 and Q3 FY26. That is a history of changing guidance, not a claim that the target disappeared forever.

There is a later development that must travel with the example. The Q4 FY26 transcript reports the Canadian and European businesses under discontinued operations and gives a FY26 consolidated EBITDA margin of 14.22% for the continuing business. Inve's extracted history instead records 14.16% and labels the target achieved. The primary transcript controls here. Removing loss-making operations from the reported perimeter means the new margin is not an unchanged-scope test of the earlier group target.

This is why the concall and annual report belong together. The call gives the forward expectation; the statutory notes and later accounting explanation tell you what the outcome actually measures.

Ownership belongs in this layer too. Inspect promoter shareholding disclosures, changes in control and any relevant pledges alongside related-party transactions. A high promoter stake is neither a quality certificate nor, by itself, evidence of abuse.

Layer 4: keep separate targets separate

Goodluck India shows how a favourable segment target can coexist with a less favourable group target. Its defence-margin history begins at 15%–25% in Q1 FY25, moves above 25% in Q3 FY25, to 25%–35% in Q1 FY26 and to 30%–35% in Q2 FY26, with the higher range reiterated in Q3 FY26 (Inve data, commitment 15615).

The separate FY26 revenue history starts at about ₹4,800 crore, moves to roughly ₹4,650 crore in Q3 FY25, and is discussed around ₹4,500 crore in Q2 FY26. The Q3 FY26 update retains a 15%–20% growth expectation. Inve's stored financial rows put quarterly revenue at ₹983 crore, ₹991 crore and ₹1,032 crore through Q1–Q3 FY26.

A higher segment margin is not evidence that consolidated revenue guidance improved. The two targets have different denominators and different economic effects. A guidance review should preserve both rather than selecting whichever makes the latest narrative look strongest.

For an example of a target exceeded in the parsed record, Apar Industries guided conductor EBITDA per tonne at ₹28,500 on an annual basis in Q3 FY25. Its history records a revision to ₹30,000 plus tailwinds in Q4 FY25, with that quarter's actual at ₹41,430. The 9M FY26 weighted figure was ₹42,311 and the FY26 figure ₹43,012 (Inve data, commitment 654). Those are a quarterly actual, a nine-month weighted actual and an annual actual, respectively; do not average them together.

Both cases teach the same habit: retain the metric, business scope, period and original target before interpreting the status label.

How Inve fits into a four-layer routine

Use an initial time budget to organise the reading, not to certify that a stock has passed diligence. The order can change depending on the question:

1. Financials: form a shortlist and identify the cash-flow or return assumption most likely to fail.
2. Operations: find the company-disclosed KPI that bears on that assumption and compare periods on the same basis.
3. Guidance: use Concall AI for the current call and the Promise Tracker for earlier targets, updates and repeated analyst questions. Detailed Indian ledgers are Pro-gated.
4. Footnotes and valuation: resolve discrepancies in the filings, then test the assumptions with the DCF calculator where suitable.

The guidance-verification workflow provides the detailed checklist for the third step. Inve's benefit is organising the source trail; the Centum example shows why the trail must remain reviewable. As the platform describing its own workflow, Inve has an incentive to emphasise this layer. The evidence should earn the emphasis.

Where the stack can mislead you

More documents do not guarantee better returns. A company can meet conservative guidance during a favourable cycle while remaining expensive. Another can miss an ambitious target and still allocate capital sensibly. A research stack supports judgement; it does not issue position sizes.

The roughly two-year window in these examples is also shallow. The September 2026 production snapshot averaged 4.18 distinct commitment birth quarters per company, not eight complete quarterly observations for every issuer. Companies with limited call disclosure need more attention to annual reports and other filings, not an automatic penalty for an empty ledger.

The corpus was extracted with two model generations without a common holdout comparison. Status labels can misread periods, scopes and omissions. The Repco and Centum examples are practical reasons to verify a row before acting on it.

Frequently asked questions

What is the difference between a research stack and several screeners?

A stack assigns different questions to different evidence. Several screeners may help compare data, but repeating historical ratios does not answer what happened to an earlier management target or what sits in subsidiary notes.

Do I need multiple paid subscriptions?

No fixed subscription bundle is required. Company filings and your own notes can perform the work. Inve's detailed Indian guidance ledgers require Pro; they should not be described as unrestricted free-tier access.

Where should operating KPIs come from?

Start with company presentations, results releases and calls. Tools can organise those disclosures, but check whether a figure is an actual or guidance, its date, and its business scope before comparing companies.

Does a clean guidance history establish management quality?

It is one piece of evidence over a limited window. Easy targets, a favourable cycle, incomplete coverage and extraction errors can all produce a reassuring-looking history. Valuation and capital allocation still need independent work.

A useful stack ends with a clearer unresolved question, not merely fewer open tabs. Which assumption in the investment case is still supported only by management's words?