I want to analyze a market or stock.

How to brief an AI to analyze a market or a stock

Use this form to have an AI look at a company, a sector or the economy. This site composes prompts and gives no financial advice. The prompt asks the AI which way the evidence leans, and tells it not to tailor the answer to your finances.

We found no prompting guidance from an AI maker written specifically for markets, so the form follows published tests of AI research. No prompt supplies an edge by itself. In a study by JPMorganChase researchers of four AI research tools, 11% to 34% of claims were not in the sources cited for them, depending on the tool, and analysts’ reports were rated higher.

Why the recommended questions matter

• Subject. Name the exact company, ticker or market and the period. Perplexity’s prompt guide warns that made-up answers are most likely when search returns a near miss, such as a different year or a related company.
• Question. Perplexity’s help center says a focused question leads to a focused report.
• Time frame. Anthropic’s sector-overview skill asks for report dates and flags data that may be stale.
• Sources and as-of date. In the FinanceBench study, GPT-4-Turbo with a document search system answered 81% of questions about company filings wrongly or not at all. A date can steer what the AI looks up, not what it remembers: in one study, telling a model its knowledge ended in 2010 did not stop it recalling what came after.
• Your view. Anthropic’s research on sycophancy found that AI assistants tend to agree with the user, so the form asks the AI to test your view.

A worked example

An investment club member prepares for a meeting. With “A company or stock” chosen, the form composes:

I want an analysis of: Costco (NASDAQ: COST), fiscal years 2023 to 2026.

<context>
Time frame: Back 5 years; ahead 12 to 24 months.
Sources and as-of date: The latest 10-K and Q3 call transcript, attached; data as of November 7, 2026.
My current view: I think it's priced for perfection; a drop in renewal rates would change my mind.
What this analysis is for: My own learning before an investment club meeting.
Treat my view as something to test. Give the strongest case against it.
</context>

<task>
What I want to understand: Is its membership-fee growth sustainable, and what could slow it?
Work out what the current share price already assumes (growth, margins, returns), where the evidence disagrees with that, and which dated events or numbers would settle it.
Check each forecast you make against a base rate: say how often that outcome has happened in comparable cases, and name them.
Form your own view from the primary numbers before you weigh analyst ratings or commentary; report those as what others think, not as evidence.
</task>

<constraints>
Cite a source for each factual claim. Take figures from what I supply, filings, official releases and market data, never from memory, and prefer those to news or commentary. Give each figure its date, period and units; where you cannot find a figure, write "not found".
Do calculations with code if you can, and show the formula.
Include the case for and the case against, and make the downside as detailed as the upside. Give ranges and rough probabilities, not single-point forecasts. Keep reported facts, your estimates and your opinions visibly apart.
Say which way the evidence leans and how confident you are. Do not tailor the answer to my finances or tell me what to do with my money.
</constraints>

<output_format>
Start with the answer to my question in three sentences. Then: what is priced in and where you differ; the two or three drivers that matter; bull, base and bear cases with rough probabilities; dated signposts that would prove your conclusion wrong; what you could not check. Use headings that state the finding. Leave out background I did not ask for. End with sources and their dates.
</output_format>

Today's date is November 10, 2026. If you find only a near match (a different year, a related company), say so. If the information here is not enough to answer well, say so and tell me what you would need. If something missing would materially change the result, ask me up to 3 short questions first. Otherwise proceed and state your assumptions.

What the form adds for you

• What is priced in. The CFA Institute’s guide to research reports asks an analyst to say why a security is mispriced and what the market is not discounting. The form asks the AI the same.
• A base rate. In a test of 38 prompt wordings, few improved forecasts; asking for base rates was one, slightly, and in only one of the study’s two tests.
• Its own view first. One study found that an analyst’s rating placed in front of a model pulled its answer toward the rating, even when the rating was fake.
• Figures. In a test of financial calculations, most models scored higher writing code than reasoning in prose. The form also asks for each figure’s source, date, period and units.
• Ranges and the downside. In the JPMorganChase study, most AI tools’ price targets ran more optimistic than the analysts’, and came as single figures.
• Facts apart from opinions. The CFA Institute’s Standard V(B) asks analysts to separate fact from opinion.
• A layout. If you leave “Output” empty, the AI is asked to answer your question first, and to say what would prove it wrong and what it could not check.