A Stock Research Workflow for Your Grok Bot: What to Ask Once It Has Market Data
Three checks to run with a Grok Bot that already has market data: a morning read on analyst moves, sentiment and the stories behind them; an after-earnings read that separates the print from the outlook; and a filings read on insider and 13F activity. Real answers from October 10, 2026.
Once a Grok Bot can look up market data, the questions worth asking are the ones that cross signals: what analysts changed, where sentiment sits, who filed with the SEC, and which stories explain the move. This post is a three-check routine you can hand the bot as saved prompts. Every answer below came back on October 10, 2026.
The one prerequisite is a bot with the SentiSense skills installed and a free API key; that takes about two minutes and is covered in How to Add Live Market Data to Your Grok Bot. This post assumes it is done.
What can the bot look up?
For the most-watched US stocks: analyst price targets and rating actions by firm, the SentiSense Score and news and social sentiment, the daily SentiSense Rating and its dimensions, insider Form 4 trades, institutional 13F holders, curated news stories with source links, and the earnings calendar. Read-only: it looks things up and nothing more.
A free key returns the analyst target band in full, the newest three rating actions, the top five rows of insider and 13F lists, twenty full story reads a month, and the snapshot and Rating without limits. PRO returns the full lists. The skills tell the bot to label a truncated answer as a preview.
1. The morning check
Give me a morning read on NVDA: any analyst rating changes in the last 30 days, the consensus target against the price, where sentiment sits, and the three stories from this week most likely to move it. Cite the sources.
What came back on October 10:
- Analysts: four actions in 30 days, all reiterations (UBS Buy on Oct 9, BNP Paribas Outperform on Oct 5, Cantor Fitzgerald Overweight on Oct 1, Rosenblatt Buy on Sep 29). No upgrade or downgrade. Consensus Strong Buy, mean target $328.72 across 59 analysts against a $229.28 price.
- Sentiment: strongly bullish on a 30-day basis, and one of the three most-discussed names on our board.
- Stories: a report that NVIDIA halted RTX 5090 production to move GB202 chips to workstation cards (negative, 3 sources, unconfirmed by the company); NVIDIA-backed Firmus withdrawing a US$5 billion Australian IPO (negative, 10 sources); Microsoft's Surface Laptop Ultra launching on NVIDIA's RTX Spark chip (positive, 19 sources).
The read: nothing changed on the Street this month, and the week's news is split between ecosystem demand and a supply-side report the company has not confirmed. A useful morning, because the bot told you what did not happen.
2. The after-earnings check
PEP reported this week. Separate the print from the outlook: what beat or missed, what guidance changed, how the stock reacted, and whether any firm moved its rating before or after the report. Link the sources.
What came back on October 10:
- The print (Oct 8): core EPS of $2.34 against a $2.29 estimate; net revenue of $25.27 billion, up 5.6%; organic growth of 3.1%, with both North American segments flat.
- The outlook: full-year core EPS growth cut to 2.5% to 3.5% from the low end of 5% to 7%, while the revenue growth outlook rose to about 6%. Sources: CNBC and Seeking Alpha, linked in the story.
- The reaction: a measured close-to-close move of +3.73% on the report, per the ranked earnings board.
- Analysts: fifteen actions in 14 days, twelve of them reiterations. The two downgrades came before the print, not after: JP Morgan to Neutral on Sep 29 and Deutsche Bank to Hold on Sep 28. Consensus Hold, mean target $143.18 against a $125.97 price.
- Attention: a mention spike about four standard deviations above the stock's own baseline, with sentiment near neutral.
The read: a beat on both lines paired with a profit-outlook cut, and the Street had already de-rated it the week before. The prompt asks for the reaction as the stock's own measured move rather than a direction inferred from headlines, and it treats a mention spike as attention, not a verdict: spikes fire the same way on bad news as on good.
3. The filings check
Before I form a view on NVDA, what have insiders done in the last 90 days, separating open-market sales from tax withholding and 10b5-1 plan sales, and are institutions adding or trimming in the latest 13F quarter? Then show me which Rating dimensions disagree with each other.
What came back on October 10:
- Insiders: 22 sales totalling $765.6 million from three filers, nine under 10b5-1 plans, no open-market purchases. One director accounted for $744.6 million. Five rows worth $34.0 million were shares withheld for tax, listed separately because they are not sales.
- Institutions: the September 30 quarter is still filing, so the bot flagged early filers only: 328 holders so far; among the twenty largest by shares, twelve increased and eight decreased.
- Divergence: the Rating breakdown put crowd sentiment and fundamentals at the top of the day's cross section and options positioning near the bottom, with an unusual options flow flag attached as an observation.
The read: a concentrated seller, routine plan sales, an incomplete 13F quarter, and a crowd more bullish than the options market. None of that tells you what to do; it tells you what to look at next.
Make it a routine
Save the three prompts and give them to the bot on a schedule: the morning check daily on your watchlist, the after-earnings check whenever a holding reports, the filings check before you change a position. Because Grok Bot agents message each other, the bot that runs these can hand its findings to a writer bot for your notes.
Data is provided for research and education, not investment advice. If the bot is not connected yet, the two-minute setup is in How to Add Live Market Data to Your Grok Bot.