The worst AI investing prompt is also the most common. Reframe AI from stock-picker to research-builder, with a prompt swap table and a five-step workflow that keeps judgment with you.
All briefs
Wall Street Prompt · LinkedIn one-page
Stop Asking ChatGPT What Stock To Buy
LinkedIn one-page
The worst AI investing prompt is also the most common. Make AI structure the research, not pick the name.
LinkedIn1080 × 1350
References & validationExact post missing
Reference creator resolved, but the exact source post was not captured. Capture the specific post (not just the profile) before treating as validated.
Reference creator: Ticker Symbol: YOU / AI stock picking blogs
Post content
Dense one-page content — sections, tables and frameworks
The Prompt Everyone Types First
One question dominates AI investing. It is the one that breaks the process before it starts.
- The prompt: "What stock should I buy?"
- The model answers with confidence it has not earned. No 10-K read, no comp set, no position size.
- You outsourced the one step you can never outsource: judgment.
Why it matters: A confident answer to a bad question feels like research. It is the opposite. You skipped the work and kept the risk.
Why The Common Prompt Fails
Name the mistake, then trace where it breaks. Every shortcut has a failure mode.
| Common Mistake | Failure Mode |
|---|---|
| "What stock should I buy?" | Answer with no thesis, no horizon, no risk frame. Unusable. |
| "Is TICKER a good buy?" | Model role-plays an analyst. You inherit a guess as a fact. |
| "Give me 5 stocks for 2026" | A list with no comps, no margins, no balance sheet context. |
| "Will this stock go up?" | Prediction theater. No model can price the future. |
| "Should I sell at a loss?" | Personalized advice the model is not equipped to give. |
Heads up: The pattern is the same in every row. You asked for a conclusion and skipped the process that earns it.
Reframe: AI Builds The Research, You Decide
Same tool, different job. Stop asking for the answer. Ask AI to assemble the inputs you use to reach your own.
| Weak Prompt (asks for the pick) | Strong Prompt (structures the work) |
|---|---|
| "Should I buy TICKER?" | "Summarize TICKER's last 10-K: revenue mix, margin trend, debt maturities." |
| "Is this a good company?" | "Build a comp table: TICKER vs 4 peers on growth, margin, and multiple." |
| "What's the price target?" | "List the 3 assumptions a bull case depends on. Flag the weakest." |
| "Will earnings beat?" | "Extract guidance and management tone from the last 2 earnings calls." |
| "Pick me a winner" | "Draft an IC memo skeleton: thesis, risks, what would prove me wrong." |
Key: Human judges. AI builds. The left column wants a verdict. The right column wants raw material you can audit.
The Disciplined Workflow
A repeatable five-step loop. AI does the assembly. You hold every decision gate.
| Step | AI Builds | You Judge |
|---|---|---|
| 1. Frame | Pulls business model, segments, key drivers | Is this in your circle of competence? |
| 2. Read | Summarizes 10-K, 10-Q, earnings calls | Do the numbers match the story? |
| 3. Compare | Builds comp set on growth, margin, multiple | Is the valuation gap real or a trap? |
| 4. Stress | Lists bear case and disconfirming evidence | What breaks the thesis? |
| 5. Size | Lays out scenarios at different weights | How much can you lose and still sleep? |
Why it matters: The decision lives in the right column. AI never crosses into it. That boundary is the entire discipline.
Guardrails Before You Prompt
Treat model output as a junior analyst's draft. Useful, fast, and unverified.
- Cite or discard: every claim traces to a filing or transcript, or it does not exist.
- Recency check: models hallucinate stale numbers. Confirm the fiscal period.
- No verdicts: if the prompt asks should I, rewrite it to summarize, compare, or list.
- Disconfirm on purpose: always ask for the bear case in the same session.
Tip: The goal is not a faster answer. It is a more honest process that you can defend line by line.
Caption
LinkedIn post copy
Visual design notes
- Near-black forest-green background (#0B1410 to #0E1A14). One teal-green accent (#1Fae7e) for the emphasized word, table header rules, and step numbers only. Everything else off-white on dark.
- Header: heavy condensed sans (Anton or Druk Wide) all-caps headline, left-aligned. 'STOP' in the green accent, rest in off-white. Teal subtitle banner bar directly under the headline, full content width, single line.
- Tables are the spine of the page: dense 2 and 3-column layouts, thin 1px hairline row dividers in muted green, generous left padding, no vertical borders. Left column slightly dimmed (weak prompt) vs accent-tinted right column (strong prompt) in the reframe table.
- Section headings in condensed caps with a short green tick mark to the left. 'Why it matters / Key / Tip / Heads up' notes in a subtle bordered callout box with the label in green bold.
- Workflow section can use a vertical 5-step rail with green numbered nodes on the left edge connecting each row of the table, reinforcing the loop.
- All content left-aligned, single column, no center text. Tight vertical rhythm so all 5 sections fit cleanly within 1080x1350 without crowding. Footer hairline rule above the footer line.
Production checklist
- ☐Lay out the 1080x1350 one-pager in the WSP template: near-black forest bg, single green accent, heavy condensed headline with 'STOP' emphasized, teal subtitle banner.
- ☐Build the four tables: Mistake/Failure Mode (2-col), Weak/Strong Prompt (2-col), 5-step Frame-Read-Compare-Stress-Size workflow (3-col), and style the bullet + callout sections for the opener and guardrails.
- ☐Add header and footer branding: WSP mark, Dave Wang attribution, footer line with COMMENT & REPOST, and the green tick marks on section headings.
- ☐Proof for compliance: no buy/sell/hold, no price targets, no promised returns, no personalized advice, no em dashes. Confirm 'Educational only' framing holds throughout.
- ☐Export final asset as PNG at 1080x1350 plus a PDF version sized for DM delivery, and verify table text is legible at LinkedIn feed scale.
Follow Dave Wang · Wall Street Prompt · For more AI investing workflows · COMMENT & REPOST
CTA
Save this prompt swap table. Comment WORKFLOW and I'll send the full IC-memo prompt structure.