How to Read an Equity Research Report: What to Read First and What to Question

What an equity research report is

An equity research report is a structured argument about one company. It states a thesis (why the stock is worth owning, avoiding, or watching), presents evidence (the business, the financials, the competitive position), attaches a valuation (what the evidence implies the shares are worth), and lists risks (where the argument could fail). Everything else in the document supports one of those four jobs.

Reports come from three broad sources, and the source shapes how to read them:

• Sell-side research is produced by analysts at investment banks and brokerages and distributed to clients. It is often the deepest work available — sell-side analysts cover a handful of companies for years, talk to management, and build detailed models. But the institutions that employ them also pursue banking relationships and trading business with the companies covered, and industry-wide, ratings skew heavily toward “buy.” That does not make the research wrong; it means the rating deserves less weight than the evidence underneath it.
• Independent research comes from firms whose revenue is the research itself — subscriptions rather than banking fees. The conflict profile is cleaner, though independents have their own incentives: subscription businesses reward bold, memorable calls, and short-focused shops profit when a stock falls.
• AI-generated research is produced by software that reads public sources and assembles a report. Its incentive profile is whatever its operator’s is; its distinctive risk is different — fluent output whose claims may not trace to real sources. How that plays out in practice is covered on AI stock research.

Knowing who wrote a report and what they are paid for is not cynicism. It is the first input to reading it well, because it tells you which sections to trust readily and which to interrogate.

The anatomy of a report

Formats vary, but almost every equity research report contains five components. Each exists to answer one question — knowing the question lets you find what you need without reading cover to cover.

• Thesis and rating. What is the argument, in one place? The summary, the rating (buy/hold/sell or equivalent), and usually a price target. This is the report’s claim; the rest is supposed to be its proof.
• Business overview. What does this company actually do, and how does it make money? Segments, customers, competitive position. In long-running coverage this section can be boilerplate carried forward from prior reports; in a report on an unfamiliar company it is where a reader should slow down.
• Financial analysis. Is the business healthy, and which direction is it moving? Revenue, margins, cash flow, balance sheet. The useful part is rarely the numbers themselves — those are public — but which numbers the analyst treats as the ones that matter.
• Valuation. What is the evidence worth in price terms, and under what assumptions? The model, the multiples, the target. This section deserves more scrutiny per page than any other, covered below.
• Risks. Where does the argument fail? The quality of this section is the fastest single indicator of the quality of the whole report — also covered below.

Treat this anatomy as a map, not an itinerary. The sections are listed here in the order reports usually present them, which is not the order worth reading them in.

What to read first

Nobody should read a research report cover to cover on the first pass. Reports are arguments, and the efficient way to evaluate an argument is to start where it can fail. A fifteen-minute first pass looks like this:

1. Read the thesis — and the date. One paragraph, usually on page one: what is the claim? Note when the report was written and what share price it was written at. An undated argument, or one built on a price that has since moved substantially, has already changed under you.
2. Go straight to the valuation assumptions. Skip the model’s output and find its inputs: what revenue growth, what margins, what multiple must hold for the target to be right? This is where optimism hides, and it is checkable.
3. Read the risks. Are they specific to this company, or could they be pasted under any ticker? Does the report say what would prove the thesis wrong, or only that “risks exist”?

After those three stops you know whether the argument is coherent, whether its assumptions are demanding or conservative, and whether the author has genuinely considered being wrong. If it passes, the business overview and financial sections are worth the full read. If it fails, you have spent fifteen minutes instead of two hours.

How to read ratings and price targets

A rating is a compressed opinion, and compression loses the most important information. “Buy” does not say how confident the analyst is, over what horizon, or what has to go right. Two useful facts recalibrate how much a rating should mean to a reader:

Ratings skew positive across the industry. At most sell-side firms, “buy” and “hold” ratings vastly outnumber “sell” ratings — a long-documented pattern driven by the relationships analysts and their employers maintain with covered companies. In that distribution, “hold” often functions as polite discouragement. Read ratings relative to the rater’s own scale, not the dictionary.

A single price target hides its assumptions and its odds. “$240” sounds precise, but precision is not confidence. The target is the output of a model whose inputs — growth, margins, multiple — the analyst chose, and it typically comes with no probability attached. It is one scenario presented as the scenario. And a target without a downside case is half an analysis: it tells you what the analyst thinks happens if the thesis works, and nothing about what happens if it doesn’t.

The stronger format is a range of valuation scenarios with probabilities — an upside case, a base case, a downside case, each with its own assumptions and likelihood. Where a report offers only a point target, a reader can partially reconstruct the missing half by asking: what would this analyst’s bear case look like, and why isn’t it here?

Reading the valuation section

The valuation section is where a report converts argument into a number, and the number is the least informative part. What matters is what must be true for the number to hold.

Every valuation — discounted cash flow, multiples-based, sum-of-the-parts — reduces to a small set of load-bearing assumptions: how fast revenue grows, where margins settle, and what multiple the market pays for the result. A reader who extracts those three and asks “is that demanding or conservative, given what this company has actually done?” cannot be dazzled by a target, because the target is now just arithmetic on assumptions the reader has judged independently.

A concrete illustration from Monsaic’s example analysis of NVIDIA Corporation (NVDA): the report’s one-line thesis calls the company a system-level AI factory supplier whose stock remains earnings-revision driven — but flags that the valuation is sensitive to hyperscaler capital spending and to gross-margin normalization. That single sentence names the load-bearing assumptions. A reader now knows what to check in the valuation math: what capex trajectory and what margin level the targets assume. If the report had said only “price target: $X,” all of that structure would be invisible — present in the model, absent from the page.

Two further checks: the valuation should state the price it was built on and when (“price as of”), because a target’s implied upside changes every trading day; and the terminal assumptions — what the model assumes about the distant years that usually carry most of the value — should be stated, not buried.

Reading the risk section

The risk section separates reports that engaged with the company from reports that engaged with a template. The test is specificity. “Market conditions may vary,” “competitive pressures may intensify,” “regulatory changes could affect results” — these fit every public company on earth, which means they say nothing about this one. A real risk section names concrete, checkable exposures: a customer concentration with a percentage, a margin dependency on a specific input, a pending proceeding with a docket.

The strongest reports go one step further: they state what would invalidate the thesis. These invalidation conditions — kill criteria, or thesis-breakers — turn a story into a falsifiable claim. From the same NVDA example analysis, the kill criteria are: hyperscaler and AI-cloud capex plans rolling over for two consecutive quarters; gross margin structurally falling below 68% without a clear mix-transition explanation; and customer silicon or competitor accelerators taking enough share to flatten data center growth. Each is observable, has a threshold, and could not be pasted under a different ticker.

Kill criteria do two things a generic risk list cannot. They discipline the author, because a thesis that must specify its own failure conditions cannot be vague. And they serve the reader after the report is published: the criteria are a watchlist. When one triggers, the thesis is broken by the report’s own stated terms — no reinterpretation required. A report whose thesis nothing could invalidate is not a strong thesis; it is a story engineered to survive any outcome.

Questions to ask of any report

A compact interrogation list, applicable to any report from any source:

• Where does each claim come from? Pick two or three numbers the thesis leans on and trace them. Do they cite a filing, a dated disclosure, a transcript — or nothing?
• When was this written, and at what price? An analysis is a snapshot. Without an as-of date and price context, you cannot know what the report knew.
• Where is the bear case? If the report argues only one side, you are reading half a picture — whether from agenda or blind spot, the effect on you is the same.
• What must be true for the valuation? Extract the growth, margin, and multiple assumptions and judge them against the company’s record.
• What would prove it wrong? Look for kill criteria. If the report never says what failure looks like, it has not really made a claim.
• Who benefits from my agreeing? Banking relationships, subscription renewals, a short position — every producer has incentives; the question is whether the report’s structure lets you check the work despite them.

For AI-generated reports specifically, these questions extend into a fuller structural checklist — sourcing, dating, falsifiability, scenario framing — laid out in how to evaluate AI-generated stock reports.

How a Monsaic report is organized

Monsaic is an AI stock research platform, and its reports apply the discipline described above as fixed structure. Every report contains fifteen sections, grouped by theme: the argument (executive summary, fundamental analysis, balance sheet analysis, catalyst analysis), the price (valuation and scenarios, comparative precedents, asymmetric risk profile), the failure modes (risk assessment, technical analysis), and the forensic layer (insider and institutional signals, news sentiment, management and governance, narrative versus substance, dilution and shareholder value, revenue quality).

Valuation is expressed as four valuation scenarios — bull, base, bear, and tail — each with a price target, a probability, what must be true, and what breaks it, rather than a single target. Kill criteria are defined before the grade, so the verdict is falsifiable by construction. Forensic grades (A+ through F, each with a written summary) cover management and governance, legal and regulatory exposure, dilution, and revenue quality, alongside a 1–10 narrative-versus-substance score; the report carries an overall verdict with a conviction level. Every report ships with its source list as claim-level citations, an analysis date, and price-as-of context. The reasoning behind this structure is on the methodology page.

The same honest limits apply to Monsaic as to any report on this page: research can be incomplete or outdated, it depends on the quality of its sources, it reflects its analysis date rather than this moment, and it is not personalized to any reader.

FAQ

Read a live one, not a diagram

The fastest way to learn the anatomy is on a real example. A covered stock’s excerpt shows the verdict, thesis, and key risk in place — and lists every section the full report walks through.

Browse covered stocksSee pricing

Keep reading

• AI stock research, done rigorously — the Monsaic pillar
• The Monsaic methodology
• How to evaluate AI-generated stock reports
• What is a valuation scenario?
• Generate your first Monsaic