Learn

Research Records and Reviews

Keep source documents, evidence logs and follow-up questions usable as the facts change.

96 guides

• Bound a slow AI stock research task without hiding missing work

  Limit the stock universe, source attempts and elapsed research budget, then publish completed calculations separately from timed-out or unstarted work.

• Build a stock research workflow with explicit dependencies

  Connect a dated stock price, target-year EPS and peer inputs into a reproducible valuation task that blocks dependent calculations when inputs fail.

• Update stock valuation rows only when broker EPS changes

  Deduplicate broker reports, compare like-for-like forecast observations and recalculate affected valuation rows without mistaking rating changes for EPS revisions.

• Block errors in automated stock research with structured checks

  Join extracted valuation fields with an independent formula branch by company, fiscal year and source row, then inject arithmetic and year errors to prove publication is blocked.

• Batch stock research while preserving partial failures

  Run one dated forward-PE research task across a stock list, return a consistent row for every input, and preserve missing-year, source, numeric and schema failures.

• Keep useful stock research when consensus earnings data is missing

  Separate available broker reports from missing consensus earnings, calculate only supported valuation rows and block comparisons that require a verified market consensus.

• ADR and ordinary shares: reconcile issuer exposure

  Combine ADR and ordinary-share holdings using depositary ratios, share-class rights and a common valuation basis while keeping separate custody records.

• Holding changes: separate cash trades from noncash events

  Reconcile holding quantities and account cash separately to distinguish purchases, sales, in-kind transfers, distributions and corporate actions in a research record.

• Cash labels: distinguish bank sweeps and fund holdings

  Classify portfolio cash by legal claim, bank sweep, money fund units, settlement status and withdrawal conditions instead of relying on a single cash label.

• Common customers: test cross-holding demand dependence

  Map verified common customer groups across holdings while preserving anonymous concentration disclosures and separating revenue dependence from customer identity guesses.

• Currency exposure: map revenue, costs and debt together

  Map portfolio companies by revenue currency, cost currency, monetary balances and debt, separating operating cash exposure from translation and stock quotation.

• Holdings disclosure lag: quantify what remains unknown

  Quantify portfolio look-through coverage and disclosure age, distinguishing quarter-end 13F snapshots from fund holdings and preserving unresolved exposure.

• Dividend records: reconcile entitlement, tax and net cash

  Reconcile stock dividend entitlements with gross amounts, withholding, depositary fees, conversion and cash received, without treating an assumed tax rate as law.

• Earnings calendar: align the periods behind holding reports

  Align portfolio earnings checkpoints by actual operating periods, fiscal calendars and disclosure dates so shared business signals are compared on compatible windows.

• Fund overlap: calculate indirect issuer exposure

  Calculate direct and fund-held issuer overlap using holdings weights, disclosure dates and unresolved coverage, avoiding duplicate counting of wrappers.

• Dividend reinvestment: reconcile cash and new share lots

  Reconcile dividend reinvestment as a distribution plus a separate purchase, preserving fractional shares, fees and residual cash instead of recording a new external deposit.

• Halted holdings: track ownership when prices stop updating

  Track halted or delisted holdings with separate share ownership, quotation age, trading status and valuation uncertainty instead of deleting them or assuming zero value.

• Portfolio guarantees: map obligations across held issuers

  Map portfolio companies as borrowers, guarantors and beneficiaries using contract limits, triggers and shared obligations without equating guarantees with expected losses.

• Issuer refinancing: compare shared funding windows

  Compare portfolio issuers by debt maturity month, funding currency and shared financing channels, retaining each company cash boundary and refinancing conditions.

• Holding value changes: separate local price and FX

  Separate holding-value changes into local-price, FX and interaction effects under fixed quantities, then isolate trades and cash flows before interpreting performance.

• Operating events: prioritize checks across affected holdings

  Prioritize operating-event checks across holdings using verified dependency paths, inventory buffers and the next observable disruption point rather than headline volume.

• Revenue geography: look beyond the listing country

  Build a holding-weighted revenue geography map using external-customer revenue and explicit unknown regions, keeping listing location separate from business activity.

• Securities lending: reconcile loans and entitlements

  Reconcile fully paid stock lending through shares on loan, contract rights, substitute distributions, collateral records and loan-income receipts.

• Share classes: separate economic ownership and votes

  Reconcile economic shares and voting power across share classes, conversion conditions and depositary procedures, preserving the difference between ownership and control.

• Shared suppliers: trace a common operating bottleneck

  Trace shared supplier bottlenecks across holdings using specific products, qualification lead times, inventory buffers and verified dependencies rather than industry labels.

• Stock splits: preserve a continuous holding history

  Maintain stock-split history by applying verified ratios to share quantities and per-share historical cost while isolating fractional-share cash and actual trades.

• Spinoffs: reconcile new holdings and remaining exposure

  Reconcile a spinoff distribution through entitlement ratios, parent and new-company holdings, fractional cash and separate post-separation operating evidence.

• Value-chain sales: avoid counting the same demand twice

  Trace sales through a portfolio value chain to distinguish supplier revenue, downstream revenue and franchise system sales from final customer demand.

• Import duties: test origin, scope and the importer obligation

  Map US antidumping and countervailing duty exposure by product scope and entry cohort, separating cash deposits, final assessment and contractual recovery.

• Merger remedies: test whether the divested business can operate

  Map divested-business assets, personnel, licenses and transition services to operational milestones, testing independence beyond approval and closing.

• Accelerated approvals: track indication-level obligations

  Track US oncology accelerated approvals by indication and requirement, separating study progress, report submission, fulfilled obligations and FDA approval changes.

• Airport slots: reconcile allocated, scheduled and used capacity

  Reconcile airport slot allocations, schedules, actual flights and qualifying historical waiver returns, preserving FAA airport, season and evidence boundaries.

• Card disputes: trace reversals, evidence and final recoveries

  Build card-dispute cohorts linking original sales, debits, evidence, provisional credits and final outcomes, with separate cash and loss bridges under the stated Visa guide.

• Channel demand: reconcile shipments with end-customer sales

  Reconcile distributor receipts, end-customer sales, returns and stock coverage to distinguish channel loading or destocking from final demand.

• Clinical readouts: match endpoints with the tested population

  Check clinical readouts against prespecified endpoints, analysis populations and cutoffs, distinguishing response arithmetic from survival and approval claims.

• Cloud SLAs: recompute uptime and eligible service credits

  Recalculate contractual cloud availability and a conditional credit using matched resource scope, outage intervals and eligible charges rather than total cloud spending.

• Container charges: reconstruct demurrage and detention clocks

  Reconstruct container demurrage and detention from free time, terminal access, equipment return and daily rates, separating invoiced amounts, disputes and actual cash outcomes.

• Indexed contracts: test the lag between costs and repricing

  Rebuild indexed contract repricing through weights, reset lags, caps and index vintages, calculating the gap between input costs and actual billing.

• Cyber incidents: verify restoration beyond a systems update

  Verify cyber-incident recovery through orders, shipments, backlog, invoicing and matched collections rather than equating system availability with normal operations.

• Dam records: separate hazard potential and condition ratings

  Read dam hazard potential and condition as separate evidence axes, match assessment dates and company dependencies, and avoid inventing failure probabilities from labels.

• Weather normalization: separate degree days and base demand

  Recalculate weather-adjusted demand using degree days, a separate base load and service-area weights, while preserving model residuals and parameter uncertainty.

• Discharge reports: rebuild pollutant loads from monitoring data

  Rebuild illustrative discharge loads from matched flow, concentration and days, then reconcile EPA monitoring-period estimates, non-detect treatment and permit-limit boundaries.

• Drug shortages: reconcile product presentations and status

  Reconcile US drug-shortage status with manufacturer presentations and dated availability evidence, without turning a supplier update into national resolution or sales.

• Oil wells: reconcile drilling, completions and first production

  Reconcile drilled-but-uncompleted well counts with completions and dated first production, separating inventory revisions, connection milestones and producing-day volumes.

• Emissions intensity: reconcile boundaries before comparing

  Reconcile emissions-intensity changes across disposals, operating reductions and methods, keeping absolute emissions and activity denominators on matching boundaries.

• Executive incentives: test what the payout formula rewards

  Reconstruct incentive weights, payout curves, gates and discretion to test what executive bonus formulas reward without confusing awards with accounting cost.

• Executive succession: verify authority through the transition

  Map executive succession through decision approvals, signing rights and effective mandates, distinguishing titles from evidence of operating authority.

• Export controls: match products, destinations and license scope

  Map US export-control evidence by product version, transaction parties, destination and authorization scope, using a dated source cutoff and explicit unknowns.

• Electric reliability: separate outage frequency and duration

  Rebuild customer interruption counts and minutes to compare SAIFI, SAIDI and CAIDI, then reconcile staged restoration, repeat customers and major-event exclusions.

• Export sales: reconcile commitments, shipments and carryover

  Reconcile US grain export sales, cancellations, shipments, destination transfers and marketing-year carryover, keeping accumulated exports distinct from outstanding commitments.

• FDA 483 observations: trace responses and agency follow-up

  Build a dated evidence chain for FDA 483 observations, company responses and agency follow-up, distinguishing implementation evidence from final regulatory conclusions.

• Grid interconnection: test queue milestones and upgrade costs

  Track generation interconnection study versions and shared network upgrades, showing how a project withdrawal can change the remaining cost allocation and evidence milestones.

• Government awards: separate contract ceilings and funded orders

  Trace US government contract ceilings through supplier minimums, task orders, funding and cash, separating shared potential value from observed obligations.

• Guidance scorecards: compare results on the original basis

  Compare actual results with original guidance on matching definitions, bridging revisions, non-GAAP exclusions and business scope before labeling a beat.

• Exploited vulnerabilities: verify exposure and remediation

  Match KEV entries to a defined deployed asset cohort, separate mitigation from verified repair, and preserve unknown coverage without inferring a company breach.

• Market share claims: reconstruct the addressable denominator

  Rebuild market-share denominators across products, geography, imports and sales stages, distinguishing shipment-value share from consumer and unit measures.

• Metal stocks: separate warrant cancellation and load-out

  Rebuild live and cancelled metal-warrant balances alongside actual deliveries, preserving location, reporting delays and off-warrant coverage before inferring consumption.

• Operating licenses: verify entity, site and permitted activity

  Match operating permits to legal holders, sites, activities and conditions, testing expansion claims against specific authorization rather than group headlines.

• Patent assignments: trace transfers without assuming title

  Trace US patent documents by asset, parties, execution and recording dates, distinguishing transfers, security interests and releases without declaring legal title.

• Physical hazards: test facility exposure beyond the head office

  Match physical hazards to actual facilities, elevations and operating dependencies, separating modeled exposure from observed interruption and financial loss.

• Product recalls: trace affected units and remedy completion

  Trace product recalls through affected-unit populations, consumer response and completed remedies, separating quarantined stock from corrected customer products.

• Renewable electricity claims: inspect certificate matching

  Reconcile U.S. renewable-electricity claims from consumption to eligible retired RECs, checking exclusive attributes without implying hourly physical supply.

• TAM claims: separate need from serviceable demand

  Rebuild serviceable demand from customer qualification, budgets and channel reach, separating annualized market potential from deliverable sales scenarios.

• Labor contracts: phase wage changes and lump sums correctly

  Phase labor-contract wages by effective dates and covered hours, compound base-rate increases, and separate lump sums from recurring payroll costs.

• Software end of support: map versions and migration exposure

  Map software versions to their actual support entitlement and migration wave, distinguishing end of ordinary support, ESU coverage and verified retirement.

• Union campaigns: distinguish petitions, votes and certification

  Trace a US NLRB representation case from petition through ballot challenges and certification, keeping bargaining-unit scope distinct from a signed labor contract.

• WARN notices: separate planned layoffs and realized job losses

  Link original and revised WARN notices to the same event, then reconcile announced positions with actual departures without adding duplicate versions.

• Warranty exposure: connect failure cohorts and repair capacity

  Connect warranty sales cohorts, age-specific first failures and repeat repairs to service capacity, separating reliability claims from the repair backlog.

• Water risk: map consumption to the actual basin

  Reconcile facility withdrawals and compatible returns, map actual source basins, and separate group water intensity from site-specific seasonal dependencies.

• Test whether an old research archive can recover its evidence

  Test an archive by retrieving a historical source, matching its version to an old note, and reopening the evidence from a fresh location with limitations recorded.

• Build a company research dossier that remains searchable

  Organize a company dossier around stable identity, untouched source files, working notes, and a manifest so later researchers can retrieve the right evidence.

• Tag research documents by role without inventing certainty

  Create a small document taxonomy that separates source role, research topic, and inspection status, making archives searchable without implying unsupported trust labels.

• Keep an evidence log that points to inspected source objects

  Log source inspections with document identity, exact locator, action performed, and unresolved limits so a saved URL is not mistaken for verified evidence.

• Deduplicate research events without deleting distinct sources

  Separate event identity from document identity, group repeated disclosures carefully, and preserve translations, amendments, and independent observations as distinct sources.

• Export a research evidence packet another reader can reopen

  Assemble an exportable research packet with a manifest, source references, working notes, and a readme, then verify it outside the original folder.

• Check research file integrity without confusing it with truth

  Record file origin and checksums, preserve untouched originals, and test restoration while distinguishing byte integrity from authenticity and factual accuracy.

• Freeze a research information cutoff without rewriting history

  Create a dated information snapshot that distinguishes publicly available sources, inspected evidence, and later updates so retrospective notes do not acquire hindsight.

• Preserve quotation context without copying whole reports

  Store concise source excerpts with their qualifiers and locators, distinguish direct quotation from paraphrase, and check that a short note preserves the passage's meaning.

• Build a filing reading queue around questions and dependencies

  Prioritize a filing reading queue by the questions it can answer, source dependencies, and completion criteria so unread downloads do not masquerade as finished research.

• Hand off a research question with evidence and a clear next action

  Prepare a research handoff that states the exact question, checked evidence, unresolved limits, and acceptance criteria so a reviewer can continue without reconstructing the chat.

• Define evidence review triggers that someone can actually check

  Turn vague reminders into observable evidence triggers with a source route, a defined review action, and a retirement rule for obsolete follow-ups.

• Close a research session with a resume-ready record

  Finish a research session by preserving inspected sources, unfinished questions, and the next concrete action so the next session starts without rebuilding context.

• Resolve conflicting source statements by comparing their scopes

  Compare apparently conflicting statements by entity, date, definition, and source role, then document genuine unresolved contradictions without forcing a verdict.

• Create source locators that survive a PDF-to-HTML switch

  Combine document identity, section heading, search phrase, and format-specific page references so another reader can locate the exact evidence without repeating your search.

• Build a source map that separates origin from repetition

  Map sources by their roles and relationships so repeated issuer messages, translated copies, and independent observations are not treated as equivalent evidence.

• Replace a broken or superseded source without erasing its trail

  Replace inaccessible or superseded research sources by recording the reason, new identity, content relationship, and affected notes while preserving the original reference.

• Turn missing evidence into an actionable research question list

  Convert vague research gaps into questions with a source target, a next action, and an explicit closure condition so unanswered items remain visible and actionable.

• Compare source versions with a change register

  Pair the correct source versions, classify textual changes, and inspect their meaning before updating the notes that depend on the changed material.

• Define research watch states without turning them into ratings

  Use evidence-based workflow states such as collecting, checking, waiting, and archived to show what research needs next without implying investment ratings.

• Workforce productivity: preserve the outsourcing boundary

  Test workforce productivity after outsourcing using compatible internal and external activity hours, separating employee ratios from output efficiency and cost.

• Workplace injuries: reconcile case severity and exposure hours

  Recalculate workplace injury rates from cases and actual exposure hours, retain DART and severity measures, and distinguish US survey estimates from company evidence.

• Resume stock research from verified data steps

  Resume a stock valuation from a serialized checkpoint, reuse only compatible unexpired inputs, retry failed dependencies within a budget, and preserve distinct completed versions.

• Schedule a stock screen with data refresh and validation

  Design a scheduled stock screen that refreshes and validates dated data before selection, preserves prior results on failure, and suppresses duplicate logical slots.

• Dry-run a scheduled stock screen and test its failure paths

  Replay scheduled-screen triggers and failures with real step traces and dated output counts, then inject defects to verify that the local test suite detects them.

• Recompute affected stock research after a financial correction

  Invalidate the stock screens and valuation outputs that depend on a corrected financial input, recompute the affected graph, and preserve unaffected results.

• Rerun affected stock screens after an earnings release

  Use verified earnings-release events and a rule dependency map to update one company's affected screening conditions, with duplicate suppression and transactional local checks.