Data Sources & Research Methodology

Learn how InvestSights sources, processes and presents Indian stock-market data, financial ratios, forensic scores, AI research, portfolio analytics and market insights.

Purpose and regulatory scope

This page explains where InvestSights data comes from, how often it is refreshed, how our derived scores and AI analyses are produced, and what their limitations are. It is intended to let you judge for yourself how much weight any figure on the site deserves.

InvestSights is an analytics platform and is not SEBI-registered as a Research Analyst or Investment Adviser. All figures are educational, descriptive data — not investment advice. Consult a SEBI-registered adviser before investing.

Nothing on InvestSights is a recommendation to buy, sell or hold any security, and no figure should be treated as a price target or as assurance of any outcome. Investing in securities carries risk, including the possible loss of capital.

Market and security coverage

InvestSights covers more than 5,600 NSE and BSE securities. More than 2,800 currently have deep fundamental coverage. The availability of financial statements, ratios, forensic models and historical data varies according to company disclosures, security type and source availability.

"Covered" means a security has a page on InvestSights and appears in our search and listings. It does not mean every metric is populated for it. "Deep fundamental coverage" is the narrower set for which we hold financial statements, computed ratios and the derived scores that depend on them.

Coverage differs by security because it follows what companies actually disclose. Newly listed companies have little history; smaller and SME-platform companies file less and later; some securities are listed on only one exchange; and a company may appear under separate NSE and BSE identifiers.

Update frequency: Updated daily, subject to source availability.

Limitations

A security having a page does not imply complete data for it.
Individual metrics may be blank where the underlying disclosure does not exist or has not yet been published.
Coverage counts change as companies list, delist and file, so they are stated as approximate figures rather than exact totals.

Data-source categories

We combine several categories of source rather than relying on any single one. We describe them by category because our commercial arrangements are confidential.

AMFI publishes mutual-fund net asset values as a public daily file, and we name it because it is a public source rather than a commercial arrangement.

Sources are used in order of authority, listed below from highest to lowest. A company's own filing is treated as the highest authority for its own figures; AI-generated text is never a source in itself and always rests on the layers above it.

1. Exchange filings and regulatory disclosures — highest authority
2. Official issuer documents: annual reports, quarterly results, earnings-call transcripts and investor presentations
3. Public regulatory, macroeconomic and AMFI mutual-fund data
4. Licensed or contracted market-data providers
5. Calculations derived by InvestSights from the above
6. AI-generated summaries and analysis — lowest authority, never a source in itself
Third-party news and information sources, used for context rather than as a basis for figures

Limitations

InvestSights is not affiliated with, endorsed by, or authorised by NSE, BSE, SEBI or any other exchange or regulator, and holds no exchange partnership.
Where sources disagree we follow the company's own filing wherever possible, but discrepancies between sources do occur.
Figures can be restated by the company after publication, and a restatement is reflected only once the corrected source data is available.
Corporate actions such as splits, bonuses and mergers can take time to be reflected across every derived figure.
Data can be missing entirely where a company has not disclosed it. A blank value means the data was unavailable, never that the underlying figure is zero.
No guarantee is made that data is complete, current or free from error. Verify any material figure against the company's own filing or the exchange websites directly.

Data freshness

Stock prices used on public stock pages, in screening and for portfolio valuation are end-of-day observations updated after market close, subject to source availability. InvestSights does not currently provide public real-time stock quotes.

Fundamentals, derived scores and AI analyses refresh on their own daily or source-driven schedules rather than moving with the market.

Where a figure has a date attached, that date is the period the figure belongs to — not the moment we fetched it. When a current value is unavailable we prefer to show the last verified value with its date, or to state that the data is temporarily unavailable, rather than display a placeholder number.

Update frequency: Updated after market close.

Limitations

A delayed or failed source update means the previous value continues to be shown, labelled with its own date.
End-of-day figures will not match an intraday broker terminal, and are not intended for trading decisions.
Update timing depends on when sources publish, which we do not control.

Financial statements and ratios

Annual and quarterly financial statements are taken from company filings. Ratios are then computed from those statements rather than copied from a third party, so a ratio is only as complete as the statement behind it.

Where a company restates a prior period, or reports under a different presentation, our figures follow the filing.

Update frequency: Updated when new source information becomes available.

Limitations

Banks, non-banking financial companies and insurers report under a different structure from other companies, so some line items and ratios are not comparable across sectors.
Smaller and SME-platform companies disclose less and later, so history can be short or incomplete.
Reported figures can be restated by the company after we publish them.

Screener methodology

The screener runs on a pre-computed dataset covering more than 6,000 securities, with 194 filters organised into 18 categories and 24 ready-made screens.

The screener queries a pre-computed snapshot rather than recalculating every company on each request, which is what makes it fast. That snapshot is rebuilt on a daily cycle, so screener results reflect the most recent completed build rather than the current moment.

Update frequency: Updated daily, subject to source availability.

Limitations

A filter can only match securities where the underlying field is populated, so a sparsely disclosed field narrows the result set in ways that are not always obvious.
Results are a point-in-time snapshot, not an intraday screen.
Ready-made screens are illustrative starting points, not recommendations.

Forensic Red-Flag Analysis

Forensic Red-Flag Analysis combines published academic accounting models with checks derived from company filings to surface patterns that are worth a closer look.

Forensic models highlight patterns that may warrant further investigation. They do not prove fraud, misconduct or financial safety.

Each model is applied only where it is valid. Several widely used models were developed for non-financial companies and are reported as not applicable for banks, non-banking financial companies and insurers rather than being computed and shown misleadingly.

Red-flag alerts come from a daily rule-based scan, not from an AI model. A stock is flagged when it is on one of the exchange's surveillance lists; when both the Altman Z-Score (below 1.81) and the Piotroski F-Score (2 or less) are outside their usual range for a company that is still trading; or when a smaller company's price rises sharply (50% in 20 sessions, or doubles in 60) on a run of large daily gains or unusually high volume. That pattern alone is a Medium alert; a second, independent signal — an exchange surveillance listing, a reported bulk or block sale, or a fall in promoter holding — raises it to High.

Surveillance and accounting alerts are marked resolved the day their rule stops firing. A price-volume alert stays open for two weeks after the pattern was last seen. Resolved alerts remain visible.

Update frequency: Generated on demand.

Limitations

A red flag is an indicator for further research, never a finding of wrongdoing.
Models are sensitive to missing inputs: incomplete disclosure can move a score without anything having changed at the company.
These are third-party academic models with their own published error rates; they are not audits and not opinions on any company's integrity.
There are no real-time forensic alerts.

AI-generated research

Some written analysis on InvestSights is generated by large language models working from the filings, transcripts and figures already held in our own database, rather than from open-ended web browsing.

AI output is machine-generated. It can be incomplete or wrong, it is checked automatically before it is displayed, and it is not reviewed by a human analyst before publication. It must not be relied on as advice.

We do not publish which model produced a given piece of analysis.

Update frequency: Generated on demand.

Limitations

We make no accuracy guarantee for AI-generated text.
Because the analysis is grounded in our own data, any gap or error in that data can carry through into the narrative.
Automated quality control reduces unsupported statements but does not eliminate them.

Automated AI quality control

AI-generated content is passed through a separate automated AI quality-control agent before it is published. This is an automated check. It is not a human editorial review, and nothing on InvestSights is described as human-reviewed, editor-reviewed, expert-reviewed or analyst-approved, because no such routine step exists.

When the quality-control agent rejects an output, the content is regenerated and checked again. If content continues to fail after more than three attempts it is escalated to an administrator for human verification, who approves or rejects it. Quality-control approval and rejection records are retained.

This process reduces the rate of low-quality or internally inconsistent output. It does not certify that an AI-generated statement is correct, and it is not a substitute for checking a material figure against the company's own filing.

AI-generated pros and cons
Earnings-call (concall) analysis
Management consistency analysis
Annual-report extraction
Quarterly-report extraction

Update frequency: Updated when new source information becomes available.

Limitations

Quality control checks an output for internal consistency and for support in the underlying data. It cannot detect an error that is present in the source itself.
Content types not listed above do not pass through this agent.
Approval by the quality-control agent is not a statement that a human has read the content.

Management Credibility

Management Credibility analysis is produced from available earnings-call transcripts and management commentary. Coverage varies by company and reporting history, and each result reflects the source information available when the analysis was generated.

This is descriptive analysis of what was said against what was subsequently reported. It is not an integrity verdict, and it does not prove fraud or misconduct.

Every result carries the date it was generated and the periods it draws on, so you can see how current it is.

Limitations

It may be unavailable where a company has too little transcript history to make a meaningful comparison.
It reflects only statements present in the source material we hold.
An individual result is a snapshot from its generation date and is not continuously updated, so check that date before relying on it.

Market Psychology

The market-mood measure is a composite built from five views of the broad market: how many securities are advancing versus declining, a relative strength measure, price momentum, and where prices sit relative to their 50-day and 200-day averages.

It is a descriptive summary of prevailing market conditions. It is not a signal to buy or sell, and it does not forecast returns.

Update frequency: Updated after market close.

Limitations

It is computed from our own end-of-day dataset, so it inherits any coverage gap in that dataset.
A composite index is a construct chosen by us, not an observed market quantity, and other reasonable constructions would give different readings.

Index data

Index levels are end-of-day closing values for the major Indian indices.

When no current level is available we show the last verified close together with the date it belongs to. If even that is too old to describe the current market, we state that index data is temporarily unavailable rather than showing a stale value as though it were current, and never a placeholder number.

Update frequency: Updated after market close.

Limitations

End-of-day only; these are not exchange feeds and not intraday quotes.
A value shown with an earlier date is that date's close, not today's.

Commodity data

Every commodity instrument is shown with its own unit, currency and date. Read the label on the instrument rather than assuming one basis applies to all of them.

The headline commodity prices on our commodities page are quoted in US dollars per the unit stated against each one — for example per ounce, per barrel, per MMBtu or per pound — alongside the US dollar to Indian rupee rate. These are international benchmark prices rather than Indian rupee exchange contracts, so they will not match a rupee-denominated commodity contract quote.

Other pages reference additional raw-material instruments for sector context. Each is labelled separately, and its unit and basis may differ from the instruments above.

Instruments are dated independently, because a currency rate and a commodity benchmark do not necessarily publish on the same days.

Update frequency: Updated after market close.

Limitations

Check the stated unit and currency on each instrument before comparing it with a quote from elsewhere.
End-of-day values only; these are not intraday commodity feeds.
An instrument may be unavailable for a given day if its source has not published.

Portfolio analytics

InvestSights does not connect to your broker, demat account or bank account. You can add holdings manually or upload a CSV/XLSX file that you provide. Uploaded holdings are used for portfolio analytics within InvestSights.

Valuations use end-of-day prices, so a portfolio value will not match an intraday broker statement.

Update frequency: Updated after market close.

Limitations

Accuracy depends entirely on the holdings you provide, including quantities and cost basis.
We do not reconcile or verify your holdings against any external record.
Corporate actions are not automatically applied to holdings you have entered or uploaded.
Analytics are descriptive and are not portfolio advice.

Mutual funds

Mutual-fund coverage extends to more than 14,000 schemes.

Net asset values come from AMFI's daily published NAV file and are refreshed each evening after AMFI publishes. Returns shown on the site are computed from that NAV history rather than supplied by the fund.

Past performance does not indicate or guarantee future results. Mutual fund investments are subject to market risk; read all scheme-related documents carefully.

Update frequency: Updated daily, subject to source availability.

Limitations

NAV is an end-of-day value by definition; there is no intraday NAV.
Scheme mergers, renames and closures can interrupt a scheme's history.
Returns are derived by us from NAV history and may differ from a fund's own published figures owing to methodology or rounding.

IPO and Grey Market Premium

IPO schedules, price bands, issue sizes and subscription figures are assembled from exchange and registrar disclosures together with third-party IPO information sources.

Grey Market Premium is unofficial, unregulated market information collected from available public sources. It is not issued or verified by NSE, BSE or SEBI. InvestSights typically refreshes available GMP information as part of its daily IPO-data collection, but coverage is partial and values may be delayed, unavailable or inconsistent across sources. GMP should not be used as the sole basis for an investment decision.

Where we hold a source or update date for a GMP value we show it, so you can see how current the figure is.

Update frequency: Updated daily, subject to source availability.

Limitations

GMP is not an exchange price and has no official standing.
GMP is available for only some IPOs, and its absence does not mean zero.
GMP values from different sources frequently disagree.
We do not maintain a continuous intraday GMP history.

Earnings-call analysis and quality control

More than 4,300 earnings-call analyses covering more than 1,300 companies have passed our automated quality checks and are available on the site.

Analyses are generated from company transcripts and related filings, then put through an automated quality check that compares each statement against the source document. Analyses that fail this check are withheld rather than displayed, which is why the number published is lower than the number generated.

Update frequency: Updated when new source information becomes available.

Limitations

Coverage depends on whether a company publishes a usable transcript; some companies do not.
Documents filed under a transcript heading are sometimes only covering letters, which cannot be analysed.
Automated quality control reduces unsupported statements but is itself imperfect in both directions.

Data limitations

Some limitations apply across the whole platform and are worth stating plainly rather than burying in each section.

The pages and analytics you use are end-of-day, and are not suitable for trading execution. Public real-time stock quotes are not offered.
Coverage follows disclosure. Where a company has not published something, we do not have it, and we prefer a blank to a guess.
Derived scores and AI analyses are only as good as the filings beneath them.
Sources sometimes disagree, restate or correct earlier figures.
We make no guarantee of accuracy, completeness or timeliness, and no figure on the site is investment advice.

Report a data issue

Report a data issue: If you notice a potentially incorrect or outdated figure, email connect@investsights.in with the page URL, security name, affected field and supporting source if available.

We investigate reported figures against the underlying disclosure. We are not able to commit to a response time, and reporting an issue does not guarantee that a figure will be changed.

Disclaimers

Securities investing involves risk, including possible loss of principal. Past performance does not indicate or guarantee future results. Nothing on InvestSights constitutes investment, legal, accounting or tax advice, or an offer or solicitation to buy or sell any security.

InvestSights is not affiliated with, endorsed by, or authorised by NSE, BSE, SEBI, AMFI or any exchange or regulator. Company and index names are used for identification only.

Report a data issue