How PMS Firms Research Indian Stocks

A professional portfolio management service (PMS) firm in India researches a stock through a disciplined, repeatable process, not a hunch. It defines and screens a universe, reads the primary filings, builds a financial model, tests the story against management and the ground, runs valuation, sizes the position with risk in mind, and then monitors the name continuously for as long as it stays in the book.

Most of that work is unglamorous. The fun part, the judgement, sits at the very end and rests on a large base of data gathering and reconciliation that nobody sees. What follows is the real workflow, stage by stage.

Defining and screening the universe

Before any single stock, a PMS decides what it will even look at. That means a defined universe: a market-cap floor, liquidity thresholds so positions can actually be entered and exited, and often exclusions on sectors or governance profiles the firm will not touch. This is a house policy, set once and revisited, not something reinvented per idea.

Screening then narrows that universe to a shortlist worth deeper work. Filters are usually a mix of quantitative gates (growth, return on capital, leverage, cash conversion, valuation ranges) and qualitative flags (management quality, industry structure, capital allocation history). The screen is deliberately blunt. Its job is not to pick winners but to reject most of the field so analyst time is spent where it can matter.

A good screen is also point-in-time honest. Testing a filter on today’s numbers and pretending you would have known them years ago is the most common way retail-style backtests flatter themselves. Institutional screens are built to ask what was actually knowable on the date of the decision.

Reading the filings

Once a name clears the screen, the analyst goes to the primary documents. In India that means the annual report cover to cover, the last several years of quarterly results, and, critically, the notes to the accounts. The notes are where related-party transactions, contingent liabilities, auditor remarks, accounting-policy changes, and segment detail actually live. Headline profit is easy; the notes are where the real questions are.

The analyst reconciles the story across time. Does the cash flow support the reported profit? Do receivables and inventory grow faster than sales? Has the depreciation policy or revenue-recognition treatment shifted in a way that helps the optics? Reading one year in isolation tells you very little. Reading five or ten years, and forcing the numbers to reconcile, is where the signal is.

This stage is slow and it is supposed to be. A single company can involve thousands of line items across filings, restatements, split adjustments, and changing disclosure formats. Getting a clean, consistent history is often more than half the total effort on a name.

Listening to the calls and tracking guidance

Filings tell you what happened. Earnings calls tell you how management thinks about what happened, and what they expect next. A PMS analyst listens to (or reads transcripts of) several quarters of calls to build a running record of guidance, capex plans, margin targets, and commentary on demand.

The value is in the comparison over time. Management said a plant would commission by a certain quarter: did it? They guided to a margin band: did they hit it, and if not, was the explanation credible or convenient? A pattern of promises met builds trust; a pattern of moved goalposts is itself information. The analyst is tracking not just the guidance but the reliability of the person giving it.

Building the model and forecasting the drivers

Now the analyst builds a financial model. The point is not a single price output. The point is to force explicit assumptions about the drivers of the business: volumes, realisations, mix, cost inflation, capacity, working capital, and capital allocation.

A good model separates the few variables that actually move the outcome from the many that do not. For a cement maker that may be volumes and per-tonne spreads; for a lender, credit growth and asset quality; for a consumer name, volume growth and gross margin. The forecast is then built on those drivers rather than by trending the P&L line by line, so every number can be traced back to an assumption someone can defend.

Point-in-time discipline matters here too. The model should only use information that was available as of the date being modelled. Backfilling later knowledge into an earlier forecast is how research quietly lies to itself.

Channel checks, scuttlebutt, and assessing management

Numbers can be internally consistent and still wrong about the world. So the analyst leaves the desk. Channel checks and scuttlebutt mean talking to distributors, dealers, suppliers, customers, ex-employees, and competitors to test whether the reported story matches what is happening on the ground. Are dealers stocking or destocking? Is a competitor undercutting on price? Is a marquee client quietly leaving?

In parallel, the firm assesses management and governance directly. Where possible that includes meetings with management, but assessment does not depend on access. Board composition, promoter pledging, auditor changes, related-party dealings, remuneration versus performance, and the treatment of minority shareholders are all readable from public disclosure. For a long-term holder, governance is not a soft factor. It is often the factor that determines whether the returns ever reach shareholders at all.

Valuation, position sizing, and monitoring

Only after the business is understood does valuation happen, and it is deliberately last. Valuation frames the range of outcomes across plausible scenarios: what the business is worth if things go well, badly, and roughly as expected, on multiples or cash flows appropriate to the sector. A high-quality business at a demanding price and a mediocre one at a cheap price are both decisions, and the model exists to make that trade-off explicit rather than emotional.

Then the position is sized with risk in mind. Conviction, liquidity, correlation with what the book already owns, and the downside in the bad scenario all feed the weight. Sizing is where research becomes a portfolio decision: a great idea sized carelessly can still sink a book.

Finally, monitoring. A stock in the book is not a settled question. The analyst tracks every new result, call, and disclosure against the original thesis, watching for the specific things that would break it. When the facts change, the position is revisited. This continuous loop, not the initial buy, is where most of the ongoing work actually sits.

Stage	What it mostly is	What it produces
Universe and screening	Rules and filters	A defensible shortlist
Reading filings	Slow reconciliation	A clean, consistent history
Calls and guidance	Listening over time	A management track record
Modelling drivers	Structured assumptions	A driver-based forecast
Channel checks and governance	Legwork and scrutiny	The thesis tested against reality
Valuation, sizing, monitoring	Judgement and discipline	A sized position, watched continuously

Where does the time go? Overwhelmingly into the first three rows. The unglamorous gathering and reconciling of primary data is the bulk of the effort, and the sharp judgement everyone associates with investing is a thin, hard-won layer on top of it. Building and maintaining that data foundation for Indian equities is exactly the kind of research infrastructure Altys works on.

What separates institutional research

The gap between institutional research and a casual retail tip is not access to secret information. It is four habits. Process: a repeatable sequence run the same way every time, so nothing important gets skipped when the story is exciting. Sourcing: conclusions traced back to primary filings and first-hand checks, not to a screenshot or a forwarded message. Point-in-time discipline: an honest account of what was actually knowable on the date of the decision, so hindsight cannot sneak in. And accountability: a written thesis someone owns, with the specific conditions that would prove it wrong stated in advance.

A tip is a conclusion with the work hidden. Institutional research is the work, made visible and owned. That is the difference, and it is why the boring stages are the ones that matter most.