Please enable JavaScript to continue using this application.

Stress-Test a Short Thesis: AI Prompt for Claude & ChatGPT

Copy this Claude prompt to stress-test a short thesis for stock research. Works with Claude, ChatGPT, Gemini, and Grok.

This AI prompt for stock analysis steelmans a bear case, then lists falsifiable checks. Paste the evidence stack into Claude, ChatGPT, Gemini, or Grok; the template does not recommend shorting any security.

Recommended tools

ChatGPT

Claude

Gemini

Grok

The Stress-Test a Short Thesis prompt (Claude, ChatGPT, Gemini, Grok)

Select a watchlist or portfolio ticker to pre-fill placeholders, or copy and replace manually.

You are a research analyst running a disciplined red-team exercise on [COMPANY_NAME] (ticker: [TICKER]). Steelman and then stress-test the bearish case.

## Context
Use SEC filings and the bear arguments provided in [BEAR_ARGUMENTS] (paste a short report, thread, or your own bear notes). If none are provided, construct the strongest bear case yourself. This is a balanced, evidence-weighing exercise - not a recommendation and not a price target. Keep language neutral (bullish/bearish, not buy/sell).

## Required Analysis

### Part 1 - Restate the Bull Case
Summarize the prevailing bullish view in 3-5 points so the bear case is tested against a fair baseline.

### Part 2 - Enumerate & Steelman the Bear Arguments
List each distinct bearish argument. For each, write the strongest possible version before critiquing it. Categorize each as: accounting, competitive, demand, balance-sheet, valuation, or governance.

### Part 4 - Quantify the Downside
If the credible bear arguments prove correct, estimate the impact on revenue growth, margins, and an estimated fair-value range. Show the assumptions. Keep this framed as a scenario, not a forecast.

### Part 5 - Falsifiable Checks
List 3-5 specific, checkable data points that would confirm or refute the bear case over the next 1-4 quarters (e.g. "gross margin below X% for two consecutive quarters").

### Part 6 - Balanced Verdict
Summarize which bear arguments are strongest, which are weak or already priced in, and where the bull and bear cases genuinely disagree on facts vs. interpretation.

### Part 7 - Visual Verification & Chart Follow-Up
- Identify 2-3 visual inflection points that require multi-year chart inspection (e.g. multi-year margin degradation, FCF burn rate, and debt-to-EBITDA trajectory).
- Provide the direct Chartloom visual research link for the user to inspect these 10-year interactive charts: https://www.chartloom.com/c/ticker/[TICKER]

## Output Format
Headers, tables, bullets. Neutral and balanced - present both sides, avoid buy/sell/hold language, and give no price target.

**Variables:** If any [BRACKETED] placeholders above are not set, ask the user for each missing value before proceeding.

Variables

Replace these placeholders in the prompt with your own values.

Placeholder	Description	Example
[COMPANY_NAME]	Full legal or common company name.	Example Corp.
[TICKER]	Stock ticker symbol.	EXMP
[BEAR_ARGUMENTS]	The bearish arguments to stress-test. If empty, the AI builds the bear case itself.	Paste short-report bullet points here

How to use this prompt

Paste the bear report or your own bear notes into [BEAR_ARGUMENTS], plus the relevant filing excerpts. Replace the bracketed variables. Force the model to steelman each argument before critiquing it so you are not just confirming your prior view.

Tips

Steelmanning the bear case first is the whole point - a rebuttal is only credible after you have stated the strongest version.
Separate arguments that are verifiable from the filing from those that rest on interpretation or forecast.
The most valuable output is the list of falsifiable checks - they turn a debate into a monitorable research plan.
Ask which arguments are already priced in; a true fact can still be a weak short if the market knows it.
Keep the downside framed as a scenario with explicit assumptions, not a prediction.

When to use this prompt

Pressure-testing a long thesis before committing to deeper research

Evaluating a published short report objectively

Building a pre-mortem style risk section for a research memo

Preparing balanced questions for an earnings call or management meeting

Updated 2026-07-08

These prompts are research tools, not financial advice. Outputs should be independently verified. They are not a recommendation to buy, sell, or hold any security. See our disclaimer for more information.

Back to all prompts