Your privacy choices

Sentimentor uses optional analytics cookies to understand visits and improve the product. Essential cookies remain enabled either way.

Privacy policy

The research library

Investor questions. Connected evidence.

An earnings headline, a filing and a positioning report can point in different directions. These investigations bring the relevant evidence together around a decision—not a definition of a single indicator.

Start with a question investors actually ask

Source-specific research series

Different signals. The same investigation.

Follow a company through each evidence stream, then bring the findings back to its thesis. Published investigations appear below their series. New questions are research briefs until a dated response and its sources are available—not findings or scheduled runs.

Options positioning

Revisit around earnings and material changes in options activity.

Next research brief · response needed

What are options traders positioning for in NVDA?

Inputs to request: Options activity · Earnings · Technicals

Inspect the connected research prompt
What are options traders positioning for in NVDA?

Treat this as one input into a shared NVDA investigation, not a standalone trade signal. Show expiry, strike, trade timestamp, volume, open interest and implied volatility where retrieved. Distinguish trades from positions, new exposure from unknown open/close status, and directional ideas from hedges or multi-leg trades. Do not equate call volume with bullish buying.

Connect the findings with relevant financials, earnings, analyst expectations, options, off-exchange activity, institutional disclosures, news, sentiment and technical context that you can actually retrieve. Do not treat correlated sources as independent confirmation. Return: supporting evidence, contradicting evidence, open questions, and what to monitor. Explain what changed versus a dated baseline, or say that no comparable baseline was retrieved.

Use only retrievable data. State the as-of date/time and timezone, exact observation windows, original source links and coverage limits. Separate reported facts, calculations and interpretation. Keep unavailable or stale inputs explicitly missing; do not use sample numbers. Do not save anything, activate a schedule, place trades or claim a backtest ran. Start with a concise answer, then show the evidence.

See how this input connects · NVDA example →

Off-exchange activity

Revisit when new off-exchange records become available.

Next research brief · response needed

What is dark-pool activity telling us about TSLA?

Inputs to request: Dark pools · Technicals · News

Inspect the connected research prompt
What is dark-pool activity telling us about TSLA?

Treat this as one input into a shared TSLA investigation, not a standalone trade signal. Identify the source feed, venues or reporting facilities, trade conditions, timestamps and corrections. Distinguish off-exchange records from identified dark-pool executions. Compare with the same-window price and volume; do not infer the buyer, institution or trade direction from a print alone.

Connect the findings with relevant financials, earnings, analyst expectations, options, off-exchange activity, institutional disclosures, news, sentiment and technical context that you can actually retrieve. Do not treat correlated sources as independent confirmation. Return: supporting evidence, contradicting evidence, open questions, and what to monitor. Explain what changed versus a dated baseline, or say that no comparable baseline was retrieved.

Use only retrievable data. State the as-of date/time and timezone, exact observation windows, original source links and coverage limits. Separate reported facts, calculations and interpretation. Keep unavailable or stale inputs explicitly missing; do not use sample numbers. Do not save anything, activate a schedule, place trades or claim a backtest ran. Start with a concise answer, then show the evidence.

See how this input connects · NVDA example →

Institutional positioning

Revisit after new holdings disclosures, keeping their as-of dates.

Next research brief · response needed

Are institutions accumulating PLTR—or distributing it?

Inputs to request: Institutional activity · SEC filings · Financials

Inspect the connected research prompt
Are institutions accumulating PLTR—or distributing it?

Treat this as one input into a shared PLTR investigation, not a standalone trade signal. Compare consecutive reported positions using shares and ownership as well as market value. Show holdings dates, filing dates, amendments, manager coverage and reporting lag. Do not infer current flows from lagged holdings or mistake price appreciation for new purchases.

Connect the findings with relevant financials, earnings, analyst expectations, options, off-exchange activity, institutional disclosures, news, sentiment and technical context that you can actually retrieve. Do not treat correlated sources as independent confirmation. Return: supporting evidence, contradicting evidence, open questions, and what to monitor. Explain what changed versus a dated baseline, or say that no comparable baseline was retrieved.

Use only retrievable data. State the as-of date/time and timezone, exact observation windows, original source links and coverage limits. Separate reported facts, calculations and interpretation. Keep unavailable or stale inputs explicitly missing; do not use sample numbers. Do not save anything, activate a schedule, place trades or claim a backtest ran. Start with a concise answer, then show the evidence.

See how this input connects · NVDA example →

Insider activity

Revisit after material insider disclosures or corrections.

Next research brief · response needed

What are insiders actually doing at META?

Inputs to request: SEC filings · Financials · News

Inspect the connected research prompt
What are insiders actually doing at META?

Treat this as one input into a shared META investigation, not a standalone trade signal. Show original Form 4 records, transaction codes, dates, ownership type and reported trading-plan information. Separate open-market purchases and sales from grants, option exercises, gifts and tax withholding; distinguish transaction date from filing date. Do not infer motive or wrongdoing.

Connect the findings with relevant financials, earnings, analyst expectations, options, off-exchange activity, institutional disclosures, news, sentiment and technical context that you can actually retrieve. Do not treat correlated sources as independent confirmation. Return: supporting evidence, contradicting evidence, open questions, and what to monitor. Explain what changed versus a dated baseline, or say that no comparable baseline was retrieved.

Use only retrievable data. State the as-of date/time and timezone, exact observation windows, original source links and coverage limits. Separate reported facts, calculations and interpretation. Keep unavailable or stale inputs explicitly missing; do not use sample numbers. Do not save anything, activate a schedule, place trades or claim a backtest ran. Start with a concise answer, then show the evidence.

After earnings

Revisit each earnings release and subsequent filing.

Published investigations

Next research brief · response needed

What changed in AMZN’s thesis after earnings?

Inputs to request: Earnings · Financials · Analyst estimates

Inspect the connected research prompt
What changed in AMZN’s thesis after earnings?

Treat this as one input into a shared AMZN investigation, not a standalone trade signal. Compare matching-period revenue, margins, earnings, cash flow and guidance with the previous report and dated pre-release expectations if available. Separate recurring performance from one-off items. Preserve missing transcripts or consensus snapshots as gaps.

Connect the findings with relevant financials, earnings, analyst expectations, options, off-exchange activity, institutional disclosures, news, sentiment and technical context that you can actually retrieve. Do not treat correlated sources as independent confirmation. Return: supporting evidence, contradicting evidence, open questions, and what to monitor. Explain what changed versus a dated baseline, or say that no comparable baseline was retrieved.

Use only retrievable data. State the as-of date/time and timezone, exact observation windows, original source links and coverage limits. Separate reported facts, calculations and interpretation. Keep unavailable or stale inputs explicitly missing; do not use sample numbers. Do not save anything, activate a schedule, place trades or claim a backtest ran. Start with a concise answer, then show the evidence.

See how this input connects · NVDA example →

Expectations versus evidence

Revisit when a dated before-and-after estimate set is available.

Next research brief · response needed

Did Wall Street change its expectations for MSFT—and does the evidence support it?

Inputs to request: Analyst estimates · Earnings · Financials

Inspect the connected research prompt
Did Wall Street change its expectations for MSFT—and does the evidence support it?

Treat this as one input into a shared MSFT investigation, not a standalone trade signal. First establish whether expectations changed. Compare dated consensus snapshots for the same fiscal period, including analyst counts and dispersion if available. Separate target-price, rating and earnings-estimate changes. Do not manufacture revisions from one current snapshot.

Connect the findings with relevant financials, earnings, analyst expectations, options, off-exchange activity, institutional disclosures, news, sentiment and technical context that you can actually retrieve. Do not treat correlated sources as independent confirmation. Return: supporting evidence, contradicting evidence, open questions, and what to monitor. Explain what changed versus a dated baseline, or say that no comparable baseline was retrieved.

Use only retrievable data. State the as-of date/time and timezone, exact observation windows, original source links and coverage limits. Separate reported facts, calculations and interpretation. Keep unavailable or stale inputs explicitly missing; do not use sample numbers. Do not save anything, activate a schedule, place trades or claim a backtest ran. Start with a concise answer, then show the evidence.

See how this input connects · NVDA example →

Attention versus conviction

Revisit comparable observation windows, preserving sampling limits.

Next research brief · response needed

Is sentiment around NVDA actually improving—or is there just more attention?

Inputs to request: Sentiment · News · Financials

Inspect the connected research prompt
Is sentiment around NVDA actually improving—or is there just more attention?

Treat this as one input into a shared NVDA investigation, not a standalone trade signal. Separate attention from sentiment, compare matched platform-specific windows, and report sample sizes, ticker ambiguity, duplicate posts and bot or coverage limits. Cite representative sources. Do not present a retrieved sample as the whole market or sentiment as proof of business performance.

Connect the findings with relevant financials, earnings, analyst expectations, options, off-exchange activity, institutional disclosures, news, sentiment and technical context that you can actually retrieve. Do not treat correlated sources as independent confirmation. Return: supporting evidence, contradicting evidence, open questions, and what to monitor. Explain what changed versus a dated baseline, or say that no comparable baseline was retrieved.

Use only retrievable data. State the as-of date/time and timezone, exact observation windows, original source links and coverage limits. Separate reported facts, calculations and interpretation. Keep unavailable or stale inputs explicitly missing; do not use sample numbers. Do not save anything, activate a schedule, place trades or claim a backtest ran. Start with a concise answer, then show the evidence.

See how this input connects · NVDA example →

See the inputs, not just the conclusion.

Each page identifies the period examined, the documents used, and the gaps that remain. Reported facts sit beside the calculations and interpretations they support. A checked category means evidence is present in that investigation; an unavailable source stays unavailable rather than becoming a sample number.

The examples are historical, not real-time recommendations. NVIDIA’s cash conversion raises a different question from Apple’s diluted share count. Gold’s ETF flows cannot substitute for the settlement history of a specific futures contract. Matching the instrument, date and definition is part of the work.

Open the complete supplied response to inspect the original question and limitations. Use the source links to check material claims independently, then investigate a stock you follow. Reading a report does not save research, schedule a workflow, execute a backtest or place a trade.

Investigate a stock you follow.

Bring your question to Augur in Sentimentor. Start with one free guest investigation, then use your account to keep the research, work with charts, test explicit rules with PineNode, and schedule checks through workflows. Keep reviewing what new evidence means for your thesis.

One guest preview, subject to rate limits. Account features and scheduled runs use trial or plan allowances. Nothing is saved, scheduled, or traded by reading this page.