◆ ClaudeThings
PromptsBlogUse casesFree toolsGet ClaudeThings
Home/Blog/Investing
Applied prompts

7 Claude Prompts for Smarter Investing Research

July 20268 min readEducational — not financial advice

Before anything else: Claude is a research assistant, not a financial advisor — and neither are we. Nothing here is a recommendation to buy or sell anything. These prompts help you read faster, question harder, and document your own reasoning. Decisions, and their consequences, stay yours. Verify every number against primary sources; see our full disclaimer.

The honest case for using Claude in investing research is narrow but real: most retail investing mistakes aren't information problems, they're discipline problems — unread filings, unexamined theses, decisions made in a mood and reviewed never. A model that reads 200 pages without fatigue and argues against you without ego attacks exactly that weak spot.

What Claude can't do: know today's prices (verify anything time-sensitive), predict returns, or want your money to grow. Use it where it's strong — comprehension, skepticism, and process — and keep it away from "what should I buy."

1. The annual report distiller

A 10-K is 150+ pages designed to be legally complete, not readable. Claude reads all of it. Upload the filing (or paste sections) with this:

prompt 01 — annual report distiller
Read this annual report (10-K) and distill it for a careful non-professional investor.

[attach or paste the filing]

1. How does this company actually make money? Break revenue down by segment with rough percentages.
2. What changed vs. last year — in the business itself, not just the numbers?
3. Risk factors: which 3 of the listed risks are SPECIFIC to this company (skip the boilerplate every filing has)?
4. Read the footnotes and MD&A: any changes in accounting treatment, one-time items dressed as recurring, or growing gaps between net income and cash flow?
5. What questions would a skeptical analyst ask management after reading this?

Quote the filing directly for anything important. If something is unclear or missing, say so rather than smoothing over it.
Why it works: The footnote/MD&A instruction points Claude at where filings hide things. The 'skip the boilerplate' filter on risk factors saves you from the generic litigation-and-competition list.

2. The thesis red team

Write down why you own (or want to own) something, then make Claude attack it. This is the highest-value prompt on the page.

prompt 02 — thesis red team
Here is my investment thesis:
[write it out — the company, why you think it wins, your time horizon, what you paid or would pay]

Red-team it without mercy:
1. Restate my thesis in its weakest defensible form — what am I actually betting on?
2. What has to go RIGHT, in sequence, for this to work? Which link is most fragile?
3. Argue the bear case as a smart short-seller would — not strawmen, their best material.
4. What evidence, if I saw it in the next two quarters, should make me exit? Be specific enough that I can't rationalize past it.
5. What does the current price already assume? Am I paying for the optimistic scenario?

Do not reassure me. If the thesis is mostly vibes, say so.
Why it works: Pre-committing to exit evidence (step 4) is the discipline most investors lack — deciding what would change your mind BEFORE you're emotionally invested in being right.

3. The earnings call decoder

prompt 03 — earnings call decoder
Analyze this earnings call transcript:
[paste transcript]

1. What did management emphasize — and what did they conspicuously not mention that they discussed last quarter?
2. Decode the hedged language: list phrases like "largely on track" or "some softness" and what they likely mean.
3. Analyst Q&A: which questions did management actually answer, and which did they deflect? Deflections are data.
4. Guidance: what exactly was promised, in numbers? How does it compare to what they promised last time vs. what they delivered?
5. One-line verdict: did this call raise or lower the quality of information available to shareholders?
Why it works: Comparing what disappeared between quarters, and treating deflections as signal, is how professional listeners read calls — most retail summaries capture neither.

4. The portfolio concentration X-ray

prompt 04 — concentration x-ray
Here are my holdings and rough weights:
[list them — tickers/funds and % of portfolio]

Without giving buy/sell advice, X-ray the structure:
1. What are my REAL exposures once you look through the funds — by sector, geography, factor, and single-company overlap? (e.g. how many of my funds are secretly the same 7 mega-caps?)
2. What single scenario would hurt the most positions at once?
3. Where am I taking risks I'm probably not being paid for (overlap, correlation), vs. deliberate concentrated bets?
4. What questions should I be asking myself about this structure? Frame them as questions, not recommendations.
Why it works: Look-through overlap is the analysis retail portfolios most need and least get — three diversified funds are often one concentrated bet wearing three hats. The questions-not-recommendations frame keeps this as research, where it belongs.

5. The moat interrogator

prompt 05 — moat interrogator
Company: [name]
What I believe their competitive advantage is: [your words]

Interrogate the moat:
1. Classify the claimed advantage: network effects, switching costs, scale economics, brand, regulation, or none of the above. What's the evidence FOR the classification — pricing power, retention, margins vs. peers?
2. Who has attacked this moat in the past 5 years and what happened?
3. What technology or behavior shift would make this moat irrelevant (not weaker — irrelevant)?
4. Rate the moat: durable / eroding / narrative. Justify with observable facts, and tell me what to watch that would signal erosion early.
Why it works: Forcing the moat into a named category with observable evidence separates real structural advantages from stories. 'Narrative' as a rating option keeps the exercise honest.

6. The fee and behavior auditor

prompt 06 — fee & behavior audit
Audit the costs and behaviors in my investing, not my picks.

My setup: [account types, funds and their expense ratios, how often you trade, tax situation basics]
My last 5 buy/sell decisions and roughly why: [honest list]

1. Total up what I'm paying annually in expense ratios and estimated trading/tax friction. Show the 20-year compounding cost of that drag on a [portfolio size] portfolio.
2. From my 5 decisions: what behavioral patterns do you see? (chasing recent performance, selling on drawdowns, thesis drift, position sizing by conviction or by mood?)
3. Which ONE change — in cost or behavior — would statistically matter most, based on what actually drives retail underperformance?
4. Design a simple rule I could adopt to interrupt my worst pattern before it executes.
Why it works: Costs and behavior are where the research consensus says retail returns are actually won and lost. Claude computing your personal 20-year fee drag makes the abstract visceral.

7. The decision journal

prompt 07 — decision journal
I'm about to make this investment decision: [what and why, in your own words]

Interview me for a decision journal entry, one question at a time:
1. What's my specific expectation — over what time frame, with what confidence?
2. What information am I acting on, and what's its source and age?
3. What would make this a mistake even if the price goes my way (bad process, right outcome)?
4. Am I deciding differently than I would have a month ago? What changed — the facts or my mood?

Then write the entry: date, decision, reasoning, expected outcome, exit conditions, review date. Keep it under 150 words so I'll actually do this every time.
Why it works: The process-vs-outcome question (3) is the core of good decision hygiene. A journal Claude maintains with you turns every trade into training data for your own judgment.

The pattern behind all seven

None of these prompts ask Claude what to buy — that's not a question it can answer well, and anyone selling prompts that claim otherwise is selling something else. What they do is put a tireless, well-read skeptic between your impulses and your money: reading the filings you'd skim, voicing the bear case your friends won't, and writing down the reasoning you'd otherwise reconstruct favorably in hindsight.

For the general research techniques underneath these — hypothesis-first analysis, sanity-checking claims, executive summaries — see the data analysis prompt collection.

FAQ

Can Claude tell me what stocks to buy? +
It will decline personalized investment advice, and that's correct behavior — it doesn't know current prices, your finances, or the future. Its genuine edge is comprehension and structured skepticism over documents you give it, which is what these prompts use.
How current is Claude's market knowledge? +
Claude's training has a cutoff date, and it doesn't see live prices unless you provide data or enable web search. Best practice: paste or attach the current documents you want analyzed, and independently verify any figure before acting on it.
Is it safe to share my portfolio with Claude? +
Anthropic's consumer and commercial products have differing data policies — review the one for your plan. Sharing tickers and weights (without account numbers or personal identifiers) is the sensible middle ground the prompts here assume.

Keep reading

10 Claude Prompts for Data Analysis

The profiling, hypothesis, and sanity-check prompts behind good research.

Read →

10 Prompting Techniques for Claude

The technique layer under every prompt on this page.

Read →

10 Claude Prompts for Business

Strategy red-teams, pricing briefs, and decision journals.

Read →
◆ ClaudeThings

Your AI engineering & marketing team for Claude Code.

Product
What's insideKitsPricingFAQ
Resources
Claude promptsBlogUse casesComparisonsFree tools
Legal
LegalTermsPrivacyRefundsDisclaimer
Connect
epictools.io@gmail.comclaudethings.com
Unofficial & independent. ClaudeThings is not affiliated with, endorsed by, or sponsored by Anthropic. "Claude," "Claude Code," and "Anthropic" are trademarks of Anthropic. Content on this site is for educational purposes — see our disclaimer.

© 2026 ClaudeThings. All rights reserved.