What is the best way to research a stock?
Use nine connected evidence checks: reported results, operating drivers, cash conversion, capital allocation, per-share economics, financial capacity, management guidance, the next official event, and filing changes. For every check, retain the source, observation date, reporting period, currency, calculation, known limitation, and one unanswered question.
Start with SEC filings and issuer material, then add independent reporting, estimates, valuation, and your own interpretation with clear labels. A missing guidance range or unconfirmed event date is not an invitation to improvise. It is an explicit research gap. Finishing the checklist does not produce a buy signal; it produces a better map of what is known, what is calculated, and what still needs proof.
Begin with public evidence you can reopen
The SEC makes public-company filings available through EDGAR, while Investor.gov explains that periodic disclosures give investors information they can use in their own due diligence. FINRA’s due-diligence guide points readers to 10-K and 10-Q filings, company investor-relations material, ownership disclosures, and market context. These sources do not decide whether a stock is attractive. They establish where reported facts, management explanations, financial statements, and ownership records can be verified.
Use the accession, form, filing date, report period, exhibits, and amendments as durable source identity. Search EDGAR ↗
Understand the different jobs of Business, Risk Factors, MD&A, financial statements, controls, legal proceedings, and exhibits. Read the 10-K/10-Q guide ↗
Use official releases, presentations, prepared remarks, calendars, and governance material, while remembering that management statements are not audited facts.
Use filings, company sources, peer context, and market information as separate inputs rather than one blended verdict. Read FINRA’s guide ↗
Give every research question an evidence lane
A broad checklist such as “review financials, valuation, management, and risks” sounds sensible but still leaves a practical problem: which source should you open, which period should you retain, and what would count as an incomplete answer? The nine-lane method makes those decisions explicit. Each lane has one main question, a source route, a recording method, and an interpretation boundary.
The order is deliberate. Reported results establish the current baseline. Operating drivers move below the consolidated total. Cash, capital allocation, per-share economics, and the balance sheet test whether the reported change reaches the financial statements in a coherent way. Guidance and the next official event define forward checkpoints. Filing review closes the loop by showing which primary document should update the research record.
Reported results: What changed in the latest reported period?
Where to look. Start with the latest 10-Q or 10-K and keep the comparative period from the same filing beside it. Record revenue, operating income, net income, margins, and the exact accounting concepts used. A press release can help with navigation, but the filing is the durable source for the reported period and definitions.
What to retain. Capture the issuer, form, accession, filing date, period start and end, currency, unit, current value, comparative value, and arithmetic. Separate a reported total from a percentage or margin that you calculate. If the comparative value comes from a different filing, label that choice and check whether the issuer changed the presentation.
Interpretation boundary. A faster growth rate is not automatically better, durable, or surprising. One quarter does not establish a trend, and a margin change does not reveal its cause. Do not label a result a beat or miss without a licensed, dated consensus source and a clearly defined comparison.
Operating drivers: Which product, segment, geography, volume, or price change explains the total?
Where to look. Read the segment note, revenue disaggregation, MD&A, earnings release schedules, and official presentation. The consolidated total tells you what happened to the company; the driver evidence shows where the change appeared. Preserve the issuer’s current reporting framework rather than forcing every company into the same segment taxonomy.
What to retain. Reconcile disclosed components to the closest reported total. Calculate mix or contribution only when the denominator and component periods match. Keep price, volume, acquisition, currency, and reclassification explanations separate when management discloses them. Write down any part of the bridge that the company did not quantify.
Interpretation boundary. The largest disclosed component is not necessarily the most important economic cause. A contribution calculation is arithmetic, not causal attribution. Reporting-framework changes can make year-over-year comparisons incomplete even when every number is copied correctly.
Cash conversion: How did reported earnings become operating cash and free cash flow?
Where to look. Use the statement of cash flows and its notes. Keep the exact cumulative duration visible because a 90-day quarter, 181-day half year, and 273-day nine-month period are not interchangeable. Record operating cash flow, capital spending, working-capital movements, taxes, stock-based compensation, and other non-cash adjustments that matter to the bridge.
What to retain. If you calculate free cash flow, show the formula and the exact capital-spending concept: operating cash flow minus selected capital expenditures. Compare net income with operating cash only over the same reporting window. Then ask which recurring or timing items explain the difference instead of treating a ratio as a quality score.
Interpretation boundary. Free cash flow is not a single GAAP line and definitions vary. A strong conversion percentage can reflect temporary working-capital inflows, while a weak period can reflect investment or timing. Do not compare unequal durations or claim that cash conversion proves earnings quality by itself.
Capital allocation: Where did the company direct the cash it generated or raised?
Where to look. Trace dividends, share repurchases, acquisitions, capital spending, debt issuance or repayment, and other financing flows through the cash-flow statement, equity note, debt note, and issuer-purchases disclosure. Keep an authorization announcement separate from cash actually spent under a program.
What to retain. Place major cash uses beside operating cash flow and transparent free-cash-flow arithmetic for the same duration. Record remaining authorization only when an official source states it. Ask what reinvestment, liquidity, or balance-sheet commitments compete with distributions and what evidence would show whether the mix changed.
Interpretation boundary. Cash spent on repurchases does not prove that shares were undervalued or that dilution fell. A coverage ratio does not trace the source of every dollar. Capital allocation needs opportunity-cost and per-share evidence; it cannot be reduced to a synthetic score.
Per-share economics: Did the reported business change reach each diluted share?
Where to look. Use the earnings-per-share note and same-filing income-statement facts. Retain basic and diluted weighted-average shares, reported diluted EPS, revenue, operating income, and net income for matching periods. Review compensation, convertibles, acquisitions, and other issuance before linking denominator movement to repurchases.
What to retain. Calculate revenue or operating income per diluted share only with matching numerators and weighted-average denominators. Compare total-company growth with per-share growth and show the percentage-point gap. Keep reported EPS separate because taxes, interest, non-operating items, and the EPS denominator can change the bridge.
Interpretation boundary. A per-share growth gap is arithmetic, not proof that a buyback caused the result. Weighted-average shares are not the same as period-end shares. Per-share improvement says nothing by itself about valuation, reinvestment quality, or future returns.
Financial capacity: What resources, obligations, and financing constraints shape the next decision?
Where to look. Read the balance sheet together with debt, lease, commitment, liquidity, and market-risk notes. Capture cash, restricted cash where relevant, current assets and liabilities, reported debt, maturities, facilities, covenants, and major contractual commitments. Keep point-in-time balances separate from duration flows.
What to retain. Show any liquidity ratio or cash-to-debt comparison as a transparent calculation with the exact included concepts. Read maturities and off-balance-sheet commitments before interpreting a headline balance. Compare the current date with the prior comparable balance only when classification and currency remain consistent.
Interpretation boundary. A current ratio or cash-to-debt percentage is not a credit rating, solvency verdict, or liquidity forecast. Reported debt can exclude leases or other obligations, and cash may be restricted or needed for operations. One balance-sheet date cannot show intraperiod stress.
Management guidance: What forward baseline did management actually state?
Where to look. Use the earnings release, prepared remarks, presentation, webcast materials, or another issuer-hosted source. Record the metric, range or qualitative statement, period, currency, assumptions, publication time, and whether the guidance was initiated, reiterated, raised, lowered, or withdrawn. Preserve the company’s words without rewriting them as a forecast from StockEyes.
What to retain. Place the guidance beside the latest reported baseline and identify which operating evidence would later confirm or challenge it. If a range exists, retain both endpoints rather than collapsing it to a midpoint without explanation. Mark unavailable guidance explicitly when no normalized official baseline has been captured.
Interpretation boundary. Management guidance is a management statement, not a reported fact or promise. It can exclude items, depend on assumptions, or cover a different metric from analyst estimates. Never fill an unavailable company range with consensus, rumor, or a calendar estimate.
Next official event: When is the next source expected, and what should be checked before it arrives?
Where to look. Use the company’s investor-relations calendar, official event release, or another issuer-hosted schedule. Record the event type, confirmed date, time zone when supplied, source-check time, and link. Keep a confirmed event separate from a third-party estimate and keep “not captured” visible when the issuer has not published a date.
What to retain. Turn the event into preparation work: save the current reported baseline, management guidance, operating driver, cash bridge, capital-allocation evidence, thesis claim, and unresolved question. Define which new filing or release would answer each question. The event date organizes research; it does not create urgency or predict a price move.
Interpretation boundary. A scheduled event is not observed evidence. Dates can change, and an unavailable date must not be guessed. The proximity of an event does not prove materiality, volatility, or an expected outcome.
Filing and disclosure changes: Which primary document should be read next, and what changed in it?
Where to look. Use SEC EDGAR for the filing identity, form, report period, filing date, accession, exhibits, and amendments. Read the relevant filing sections instead of relying on a document count. For 10-K and 10-Q reviews, business, risk factors, MD&A, financial statements, controls, legal proceedings, and exhibits answer different questions.
What to retain. Compare like sections across consecutive filings and retain both accessions. Separate numeric changes, added or removed wording, accounting-concept changes, reclassifications, and true source corrections. Route a document change into a research question that names the claim, alternative explanation, and next proof required.
Interpretation boundary. Text similarity, sentence counts, or a new filing do not establish materiality. A document can change because of routine updates, taxonomy, or presentation. Filing review does not end with finding a difference; it ends with understanding what evidence the difference changes and what remains uncertain.
Six labels prevent evidence from becoming one confident story
A research note becomes difficult to audit when a reported number, management explanation, derived ratio, outside estimate, personal view, and unavailable source appear in the same paragraph without labels. Use the following claim types. A single company can have all six at once, and they can disagree without one automatically invalidating the others.
A value, date, filing identity, or disclosed condition in a primary source. Preserve the exact period, unit, and source. Do not extend the fact beyond what was reported.
An issuer explanation, expectation, guidance range, or strategic claim. Retain who said it, when, the covered period, and the assumptions or exclusions.
Arithmetic built from cited inputs, such as a margin, growth rate, free-cash-flow measure, contribution, or per-share bridge. Show the formula and limitations.
Consensus, independent reporting, industry data, or another third-party claim. Record the publisher, timestamp, methodology, and whether redistribution rights permit display.
Your thesis, scenario, question, or judgment. Keep it separate from the public evidence and define what future source would support or challenge it.
A source, metric, date, or comparison that has not been captured or cannot be reconciled. State the gap directly rather than fabricating a substitute.
research conclusion = labelled evidence + visible calculation + explicit uncertainty + next verification question
Two company briefs show why coverage and conclusion are different
The examples below come from the validated StockEyes Company Research Brief snapshot dated Jul 19, 2026. Each company uses the same nine lanes, but its fiscal calendar, available sources, reporting framework, and missing evidence remain distinct. These are demonstrations of research organization, not current recommendations or claims that nine lanes capture every fact an investor may need.
Microsoft Corp.
9 of 9 lanes available. Unresolved: No unavailable lane in this dated nine-lane snapshot. Coverage is a boundary, not a company score.
- Q3 reported results: Revenue increased 18.3% year over year
- Cash conversion: StockEyes free cash flow was $47.3B
- Per-share economics: Revenue per diluted share changed +18.6%
Next question: Which operating changes, EPS numerator adjustments, and denominator movements explain the total-versus-per-share gap?
Open MSFT evidence dashboard →NVIDIA Corp.
9 of 9 lanes available. Unresolved: No unavailable lane in this dated nine-lane snapshot. Coverage is a boundary, not a company score.
- Q1 reported results: Revenue increased 85.2% year over year
- Cash conversion: StockEyes free cash flow was $48.6B
- Per-share economics: Revenue per diluted share changed +86.9%
Next question: Which operating changes, EPS numerator adjustments, and denominator movements explain the total-versus-per-share gap?
Open NVDA evidence dashboard →Open the full Company Research Briefs to inspect all 18 lane records, official source links, observation dates, calculation boundaries, and unresolved questions. The public JSON contract is also available for readers who want to audit the structure directly.
The Evidence-to-Question Stock Research Worksheet
Use one row per lane. The worksheet is intentionally not a scorecard. “Complete” means you preserved a source and boundary, not that the evidence is favorable. If a source is unavailable, write that state and the next source to check. Print this page to isolate the worksheet, or reproduce the same columns in your own research notes.
| Lane | Claim to test | Source identity and date | Period, unit, currency | Reported fact or statement | Calculation and assumptions | Alternative explanation | Next proof and date |
|---|---|---|---|---|---|---|---|
| 01 · Reported results | What changed in the latest reported period? | Form, accession, issuer page, publisher, publication time | Exact start/end or instant date; units and currency as reported | Copy the bounded claim without interpretation | Show inputs, formula, exclusions, and whether the result is non-GAAP | Name another mechanism that could explain the same observation | Specify the filing, release, event, note, or dataset that would resolve the question |
| 02 · Operating drivers | Which product, segment, geography, volume, or price change explains the total? | Form, accession, issuer page, publisher, publication time | Exact start/end or instant date; units and currency as reported | Copy the bounded claim without interpretation | Show inputs, formula, exclusions, and whether the result is non-GAAP | Name another mechanism that could explain the same observation | Specify the filing, release, event, note, or dataset that would resolve the question |
| 03 · Cash conversion | How did reported earnings become operating cash and free cash flow? | Form, accession, issuer page, publisher, publication time | Exact start/end or instant date; units and currency as reported | Copy the bounded claim without interpretation | Show inputs, formula, exclusions, and whether the result is non-GAAP | Name another mechanism that could explain the same observation | Specify the filing, release, event, note, or dataset that would resolve the question |
| 04 · Capital allocation | Where did the company direct the cash it generated or raised? | Form, accession, issuer page, publisher, publication time | Exact start/end or instant date; units and currency as reported | Copy the bounded claim without interpretation | Show inputs, formula, exclusions, and whether the result is non-GAAP | Name another mechanism that could explain the same observation | Specify the filing, release, event, note, or dataset that would resolve the question |
| 05 · Per-share economics | Did the reported business change reach each diluted share? | Form, accession, issuer page, publisher, publication time | Exact start/end or instant date; units and currency as reported | Copy the bounded claim without interpretation | Show inputs, formula, exclusions, and whether the result is non-GAAP | Name another mechanism that could explain the same observation | Specify the filing, release, event, note, or dataset that would resolve the question |
| 06 · Financial capacity | What resources, obligations, and financing constraints shape the next decision? | Form, accession, issuer page, publisher, publication time | Exact start/end or instant date; units and currency as reported | Copy the bounded claim without interpretation | Show inputs, formula, exclusions, and whether the result is non-GAAP | Name another mechanism that could explain the same observation | Specify the filing, release, event, note, or dataset that would resolve the question |
| 07 · Management guidance | What forward baseline did management actually state? | Form, accession, issuer page, publisher, publication time | Exact start/end or instant date; units and currency as reported | Copy the bounded claim without interpretation | Show inputs, formula, exclusions, and whether the result is non-GAAP | Name another mechanism that could explain the same observation | Specify the filing, release, event, note, or dataset that would resolve the question |
| 08 · Next official event | When is the next source expected, and what should be checked before it arrives? | Form, accession, issuer page, publisher, publication time | Exact start/end or instant date; units and currency as reported | Copy the bounded claim without interpretation | Show inputs, formula, exclusions, and whether the result is non-GAAP | Name another mechanism that could explain the same observation | Specify the filing, release, event, note, or dataset that would resolve the question |
| 09 · Filing and disclosure changes | Which primary document should be read next, and what changed in it? | Form, accession, issuer page, publisher, publication time | Exact start/end or instant date; units and currency as reported | Copy the bounded claim without interpretation | Show inputs, formula, exclusions, and whether the result is non-GAAP | Name another mechanism that could explain the same observation | Specify the filing, release, event, note, or dataset that would resolve the question |
Completion rule: every row needs either a verifiable source or an explicit unavailable state, plus one question that a future source could answer. A blank row is not neutral; it is an unexamined part of the research boundary.
A 90-minute first evidence pass
This time box is for triage, not completion. Its purpose is to decide whether the idea deserves deeper work and to leave an auditable list of gaps. Pause when a source cannot be reconciled. A clean unresolved question is more useful than a fast answer built on mismatched periods.
- Minutes 0–5: write the claim. State why the company deserves attention in one sentence. Label the trigger: reported change, valuation observation, product development, portfolio exposure, or outside claim. Add what would make the idea irrelevant before opening more tabs.
- Minutes 5–15: establish source identity. Confirm ticker, legal issuer, exchange, CIK, latest 10-K, latest 10-Q, recent material 8-K exhibits, and investor-relations home. Record accession numbers and report periods so similarly named securities or stale documents cannot enter the note unnoticed.
- Minutes 15–25: capture reported results. Retain current and comparative revenue, operating income, net income, margins, and diluted EPS from one filing. Show the arithmetic for growth and margins. Mark concept or presentation differences before calling the series comparable.
- Minutes 25–35: find the operating bridge. Read the segment or product disclosure and MD&A. Identify the largest disclosed component, but also record what was not quantified. Note price, volume, mix, acquisition, divestiture, and currency effects only when the issuer supports them.
- Minutes 35–45: reconcile earnings to cash. Match the cash-flow duration to the income period. Record working capital, taxes, capital spending, compensation, and other material reconciling items. Calculate free cash flow only with a named definition.
- Minutes 45–55: trace capital and shares. Place repurchases, dividends, debt changes, acquisitions, and investment beside cash generation. Then inspect weighted-average shares and the equity note. Keep authorization, execution, compensation, issuance, and denominator movement separate.
- Minutes 55–65: review financial capacity. Capture cash, current assets and liabilities, debt, maturities, leases, and commitments. Note whether currency, restriction, classification, or an acquisition makes the point-in-time comparison incomplete.
- Minutes 65–75: save the forward baseline. Record official guidance and the next confirmed issuer event. If either is unavailable, write “not captured,” the source checked, and the check time. Do not substitute a third-party calendar without a separate label.
- Minutes 75–85: read the change surfaces. Scan the latest filing’s MD&A, risk factors, controls, legal proceedings, and exhibits against the prior comparable document. Separate changed wording from changed economics and write which difference deserves a full section review.
- Minutes 85–90: close with questions. Write one supporting observation, one challenging observation, one alternative explanation, the most important unavailable source, and the next document or event that could change the view. Decide only whether to stop, monitor, or continue research - not whether the checklist “passed.”
What should happen after the first pass?
- Can the issuer and security be identified unambiguously?If not, stop. A wrong ticker, entity, share class, or stale filing invalidates every later step.
- Do the current and comparative periods reconcile?If not, preserve both sources and resolve the calendar, concept, acquisition, currency, or reclassification issue before calculating a trend.
- Is the attractive claim supported by more than one lane?If only one headline or ratio supports it, keep the idea in research. Look for the cash, driver, balance-sheet, per-share, and filing evidence that could disagree.
- Is important evidence unavailable?Record the gap and next source. Do not reduce a coverage problem to a favorable or unfavorable score.
- Can you state a plausible alternative explanation?If not, the research is probably still anchored to the original story. Add a competing mechanism and the proof that would distinguish it.
- Is there a dated next checkpoint?Save the expected filing, official event, source refresh, or thesis test. Research becomes useful when it can be updated rather than rewritten from memory.
Eight ways a stock checklist creates false confidence
A completed row proves that a question was reviewed, not that the answer is favorable or complete. Coverage must not become a quality score.
Annual, quarterly, half-year, and nine-month values can look compatible while describing different durations. Keep every start and end date visible.
Guidance and explanations are valuable evidence, but they remain statements with assumptions. Place them beside reported results and later verification.
Free cash flow, margins, contribution, and per-share bridges need inputs and formulas. A precise decimal does not make an undocumented method reliable.
A headline published near a price move does not prove it caused the move. Verify the claim, timing, mechanism, and relevant company evidence separately.
Filling a missing guidance range, event date, segment bridge, or source passage with an estimate can make a brief look complete while weakening it.
A thesis should retain its original claim, supporting and challenging evidence, unresolved questions, and next checkpoint so later facts can be compared against it directly.
A model can be internally correct and still rest on weak margins, reinvestment, cash, or terminal assumptions. Trace each important assumption back to an evidence lane.
How to identify, verify, and track unusual trading volume
Unusual trading volume occurs when a stock's daily volume exceeds its historical average (typically measured over 20, 50, or 90 days) by a significant multiplier. While volume spikes are frequently cited as signs of institutional interest or "smart money" accumulation, they can also reflect technical events or administrative changes. To track and analyze volume spikes responsibly, follow a structured verification process:
Do not rely on absolute volume figures alone. Calculate the Relative Volume (RVOL) by dividing the current day's volume by the average daily volume (ADV) over a representative historical window. An RVOL greater than 2.0x indicates unusual volume, while an RVOL exceeding 5.0x represents a significant liquidity event that warrants immediate source verification.
Before assuming a volume spike reflects fundamental sentiment, cross-reference the date with known technical events. Large volume spikes are common during quarterly option expirations ("triple witching"), major index rebalancings (such as the Russell 2000 reconstitution), or block trades executed at the close. These are mechanical liquidity events, not discretionary buying or selling.
Verify whether the volume spike coincides with a corporate catalyst. Open SEC EDGAR and look for Form 8-K filings (earnings announcements, material contracts, management changes), prospectus supplements (indicating a share offering or dilution), or Form 4 insider cluster buys. Volume without a matching public disclosure or news catalyst is highly speculative and requires close monitoring. As a research-checklist step, InsiderTrace ranks the largest recent insider buys by reported open-market purchase value over the last 12 months, which shows which insiders are putting the most money behind their own stock.
Evaluate the volume in relation to price action. A high-volume spike accompanied by a narrow price range (known as "churning") often signals intense distribution or accumulation where buyers and sellers are matched. A high-volume breakout with a wide bid-ask spread suggests a strong, directional momentum shift. State both price change and volume ratios in your research notes.
Use official investor education to understand the source, then read the company
Investor.gov describes investment research as part of due diligence and directs readers to public disclosures. Its 10-K and 10-Q guide explains the roles of Business, Risk Factors, MD&A, financial statements, controls, legal proceedings, and exhibits. The SEC’s beginner guide explains how the income statement, balance sheet, cash-flow statement, and statement of shareholders’ equity fit together. FINRA adds company websites, peer and industry context, ownership disclosures, and market information to the source map.
These guides are orientation, not substitutes for current issuer evidence. A company can change its segments, fiscal calendar, accounting concepts, guidance policy, capital structure, or risks after an educational page is published. Reopen the current filing, preserve the accession, and date every source check.
Investor.gov: Research Before You Invest ↗ · SEC: Beginner’s Guide to Financial Statements ↗ · FINRA: Stock Up on Information Before Buying Stock ↗
Questions about stock research and due diligence
What is the best way to research a stock?
Use a repeatable evidence process rather than one ratio or headline. Review reported results, operating drivers, cash conversion, capital allocation, per-share economics, financial capacity, management guidance, the next official event, and relevant filings. Keep the source, period, calculation, limitation, and next question beside every conclusion.
Which documents should I read when researching a public company?
Start with the latest 10-K and 10-Q filings, then use current 8-K exhibits, proxy materials, insider ownership filings, earnings releases, presentations, and the company investor-relations calendar as the research question requires. SEC EDGAR is the durable source for U.S. filing identity and accessions.
How long should stock research take?
A first evidence pass can be time-boxed, but complete research has no universal duration. The 90-minute StockEyes pass is designed to expose missing evidence and define follow-up work, not to declare a stock researched or ready for action. Complex businesses and important decisions require deeper review.
Should valuation be the first step in stock research?
Valuation is important, but a model is only as useful as the evidence behind its assumptions. Understand the business, financial statements, cash, capital allocation, risks, and forward baseline before treating a precise valuation output as decision-ready.
What if a company does not provide guidance or an earnings date?
Record the lane as unavailable and keep the official source check visible. Do not replace a missing issuer statement or date with an unsourced estimate. The gap itself becomes a research boundary and a prompt to check the next appropriate primary source.
Does completing a stock research checklist mean the stock is attractive?
No. Checklist completion records coverage, not quality, valuation, suitability, or expected return. The evidence can remain mixed, conflicting, stale, or incomplete after every row has been reviewed.
How do I track and interpret unusual stock volume?
First, calculate the Relative Volume (RVOL) against a 20- or 50-day average. Then, rule out technical index rebalancings or options expirations, and match the volume spike with corporate filings (such as 8-Ks, share dilution, or insider Form 4s). Always treat volume spikes as research leads, not buy or sell triggers.
StockEyes provides informational research organization and transparent calculations, not personalized investment advice, security recommendations, valuation conclusions, trade instructions, or predictions. The MSFT and NVDA examples are a bounded official-source pilot dated Jul 19, 2026. Review each original source, current filing, methodology, and limitation before relying on a statement or calculation.