Miles Free AI Library
← Back to resources
PDFMiles FinanceJul 27, 2026

Claude as Your Personal Wall Street Analyst

A 10-minute Claude setup that captures how you invest, builds your research files, and creates reusable analysis templates.

Get the full PDF

Claude Stock Research System

Free
Download PDF

Claude Stock Research System

One-File Setup Guide & Templates by Miles Deutscher · Full walkthrough on the MDF YouTube channel · Last updated: July 2026

What this is: Drop this file into Claude and say "set this up". Claude will build your entire investing context system - folder, files, and a personal interview to fill them in. Takes ~10 minutes, most of which is you answering questions about how you invest.

How to Use It (Human Instructions)

  1. Upload this file to Claude (claude.ai, Claude Desktop, or Cowork)
  2. Say: "Set this up for me"
  3. Answer the interview honestly - the files are only as good as your answers
  4. When it's done, follow the connection guide at the bottom to plug in live data

Instructions for Claude (Execute When User Asks to Set This Up)

Prompt
You are setting up a personal investing context system for this user. Follow these steps exactly:

Step 1 - Detect your environment, then create the folder.

  • If you have access to the user's file system (Claude Desktop / Cowork / Claude Code): create a folder called investing-context in their working directory or another location they approve.
  • If you're in a sandboxed environment (claude.ai with code execution): create the folder there, and at the end zip it and give the user a download link.
  • If you have no code execution at all: create each file's content in your response so the user can copy them, and tell them the ideal home is a Claude Project.

Step 2 - Create these 5 files using the templates below. Copy the templates exactly, then fill them via the

interview.

Step 3 - Interview the user to fill investor-profile.md and strategy.md. Rules:

  • ONE question at a time. Never present a form or a list of questions.
  • Conversational, direct, no corporate tone.
  • If an answer is vague ("medium risk I guess"), push back once: ask for a concrete example, e.g. "If a position dropped 30% in a month, what would you actually do?"
  • 5-7 questions for the profile, 4-6 for strategy. Don't over-interview.
  • If the user wants to stop mid-interview, save what you have, mark unanswered fields as TODO, and tell them they can resume anytime by re-uploading the folder and saying "finish my setup".
  • Write their answers into the files as clean, structured markdown - third person is fine - dense enough that a future AI conversation can use it as context without the user re-explaining anything.

Step 4 - Portfolio and watchlist. Ask the user if they want to paste their current holdings and watchlist now.

If yes, populate portfolio.csv and watchlist.md. If no, leave the templates with example rows and remind them to fill these in - the Morning Loop prompt depends on them.

Step 5 - Verify and finish cleanly.

  • If the user has the FMP connector available, run a live test now: pull the current price and P/E of one of their holdings and confirm the data came from the connector. If FMP isn't connected, point them to the connection guide below before anything else.
  • Tell the user where the folder lives (or give the zip download).
  • Recommended setup: create a Claude Project called "Stock Research" (or, if Projects aren't available on their plan, keep one dedicated chat), upload all 5 files to it, and paste the contents of investor- profile.md + strategy.md into the Project instructions.
  • Their first message tomorrow morning: the Morning Loop prompt from prompts.md.
  • Last tip for them: every quarter, open the Project and say "review my profile and strategy with me" - a stale profile quietly degrades every answer.

Do NOT skip the interview. Do NOT fill the profile with placeholder guesses. Do NOT provide investment advice during setup - you are building context, not making recommendations.

File Templates

1. investor-profile.md

Copy
# Investor Profile
## Background
- Experience level:
- Years investing:
- Relevant background (finance, tech, none):
## Goals
- Primary goal (wealth building / income / speculation / learning):
- Time horizon:
- Target outcome:
## Risk
- Risk tolerance (with a concrete example of behaviour in a drawdown):
- Max acceptable position drawdown before acting:
- Portfolio size bracket (small / mid / large - no exact figures needed):
## Constraints
- Time available for research per week:
- Tax situation notes (optional):
- Anything off-limits:

2. strategy.md

Copy
# Investing Strategy
## Style
- Approach (long-term compounding / swing / narrative-driven / income / hybrid):
- Typical holding period:
## Position Rules
- Position sizing rule (e.g. max % per position):
- Max positions held at once:
- When I add to a position:
- When I trim or exit:
## Focus
- Sectors / assets I focus on:
- What I deliberately avoid:
- Edge I believe I have (be honest):
## Discipline
- Profit-taking rule:
- Stop-loss / drawdown rule:
- Mistakes I've made before that Claude should call out if I repeat them:

3. portfolio.csv

Copy
ticker,name,type,weight_pct,avg_entry,notes
NVDA,NVIDIA,equity,10,450.00,example row - replace with your holdings

4. watchlist.md

Copy
# Watchlist
| Ticker | Why I'm watching | What would make me buy |
|-|-|-|
| EXAMPLE | replace me | replace me |

5. prompts.md

Copy
# Research Prompt Library
## 1 - The Snapshot
Give me a full snapshot of $TICKER: current price and how it's moved this month, valuation
vs its 5-year averages, last quarter's results vs analyst estimates, current analyst price
targets, and any red flags in recent filings.
## 2 - The Filings Digest
Pull the latest earnings report and recent SEC filings for $TICKER. Summarise what
actually changed this quarter - revenue drivers, margins, guidance - and flag anything
management seems to be downplaying or burying.
## 3 - The Head-to-Head
Compare $A vs $B on valuation, revenue growth, margins, debt, and analyst sentiment. Build
me a comparison table, then tell me which one the data favors and make the strongest case
for it.
## 4 - The Finder
Screen the market for small caps under $2B with revenue growth above 25%, positive free
cash flow, and expanding margins. Cross-check against analyst coverage and flag the 3-5
trading furthest below price targets - give me the bull case and the biggest risk for
each.
(Note: starting points for your own diligence, not picks - small caps are illiquid.)
## 5 - The Morning Loop (run inside your Stock Research Project - it needs your profile,
portfolio, and watchlist as context)
Run my morning loop: overnight moves in the major indexes, pre-market gainers and losers,
today's earnings and economic calendar, latest news on my watchlist - then check it all
against my portfolio and investor profile: flag anything that hits my holdings, upcoming
earnings dates I need to watch, and the one thing that most needs my attention today.
## Trust rule (applies to every prompt)
If Claude gives numbers without calling the FMP connector, they may be from memory. Ask:
"Cite where each figure came from." If it can't, don't trust it.

Connection Guide (Do this once, after setup)

FMP - The Live Market Data

  1. Claude → Settings → Connectors → search FMP (Financial Modeling Prep) → Connect
  2. FMP is a paid service - pricing scales with usage and data depth. Start on the lowest tier (~$22/mo) and only upgrade when a prompt actually hits a data wall. Check current plans at their pricing page.
  3. Test it: "What's the current price and P/E of $AAPL?" - numbers back = you're live

Google Sheets - Your Portfolio, Live (Optional but recommended)

  1. Move your portfolio into a Google Sheet (same columns as portfolio.csv )
  2. Claude → Settings → Connectors → connect Google Drive
  3. Test it: "Read my portfolio sheet and list my holdings"
  4. Now the Morning Loop reads your real positions - never attach a file again

Research tool, not financial advice - Claude informs the decision, you make it. Free from the Telegram - full video walkthrough on the Miles Deutscher Finance YouTube channel.

Continue exploring

What to watch next