Wall Street PromptBrand
All briefs
Wall Street Prompt · LinkedIn one-page

The AI Analyst Loop

LinkedIn one-page
AI does not replace analyst judgment. It runs the loop faster: ask, retrieve, verify, challenge, monitor.
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: Allie K. Miller / Nicolas Boucher

Post content

Dense one-page content — sections, tables and frameworks

THE CORE PRINCIPLE

Most people bolt AI onto research and skip the discipline. The output looks finished. The thinking never happened. Treat AI as a fast junior analyst, not an oracle.

  • Human judges. AI builds.
  • AI is fast at retrieval and drafting. Slow at accountability.
  • You still own the thesis, the position sizing, and the loss.
  • A loop beats a one-shot prompt. Research is iterative, not a single answer.
Why it matters: A single prompt gives you a confident paragraph. A loop gives you a defensible view you can put in an IC memo and stand behind in a review.

THE FIVE-STEP LOOP

Run every name through the same five steps. Same order, every time. The order is the control.

StepWhat you doAI roleYour judgment call
1. AskFrame one specific question, not 'is this a good stock'Sharpen the question, list sub-questionsDefine what would change your mind
2. RetrievePull 10-K, 10-Q, earnings call, compsSummarize filings, extract line itemsDecide which sources are primary
3. VerifyTrace every number back to the sourceCite page, table, transcript lineReject any claim with no citation
4. ChallengeArgue the bear case against your own viewGenerate the strongest counter-thesisWeigh which risk is mispriced
5. MonitorSet triggers that break the thesisTrack filings, guidance, KPI driftDecide when the thesis is dead
Why it matters: Skip step 3 and you size a position on a hallucinated margin. Skip step 4 and you only see the bull case you wanted. Skip step 5 and you hold a broken thesis for a year.

MISTAKE -> FAILURE MODE -> FIX

The loop exists because of how AI research fails in practice. Name the mistake, then design the workflow around it.

Common mistakeFailure modeDisciplined workflow
Accept the summary as factHallucinated revenue, wrong segmentRequire a citation for every figure
Ask one vague questionGeneric, unactionable answerDecompose into 5 testable sub-questions
Only ask for the bull caseConfirmation bias, blind spotsForce a written bear case every time
Treat output as finalStale view after next earningsSet monitoring triggers up front
Outsource the decisionNo accountability when wrongHuman signs the IC memo, not AI
Heads up: The dangerous output is the one that is 90 percent right. It reads clean, so you stop checking. That is exactly where verification earns its keep.

VERIFY: THE CITATION GATE

Verification is not optional polish. It is the gate every number passes through before it touches a model or a memo.

Claim typeRequired sourcePass / Reject rule
Reported financials10-K / 10-Q line itemReject if no page or table cited
Forward guidanceEarnings call transcriptReject if paraphrased, not quoted
Peer compsNamed filings per peerReject if peer set is unstated
Operating KPIFiling or investor deckReject if metric is undefined
Tip: Make AI quote the exact sentence and location. If it cannot point to the source, the number does not exist. No citation, no entry into your model.

MONITOR: TRIGGERS THAT KILL A THESIS

A thesis you cannot break is a belief, not analysis. Define the exit conditions before you enter, then let AI watch them.

  • Write 3 to 5 falsifiers the day you open the position.
  • Tie each falsifier to a hard data point: margin, guidance, KPI, balance sheet line.
  • Set AI to flag new filings and call language against those falsifiers.
  • When a falsifier trips, you re-run the loop. You do not rationalize.
Key: The point of monitoring is not to confirm you were right. It is to find out fast when you are wrong, while the cost of being wrong is still small.

WHAT STAYS HUMAN

Draw the line clearly so the tool never drifts into the seat that owns risk.

AI buildsHuman judges
Filing summaries and extractsWhether the thesis is sound
Comp tables and draftsPosition sizing and risk limits
The counter-thesis on demandWhich risk is actually mispriced
Monitoring and alertsWhen to act, hold, or walk away
Why it matters: Speed without judgment is just faster mistakes. The loop gives you the speed. You keep the judgment.

Caption

LinkedIn post copy

AI does not replace analyst judgment. It runs the same loop you already run, just faster. Ask, retrieve, verify, challenge, monitor. Most people skip the verify and challenge steps, then size a position on a number no one traced to a filing. The discipline is the order. Human judges. AI builds. Save this and run your next name through all five steps. Comment LOOP for the verification prompt I use at step three. Educational only. Not investment advice.

Visual design notes

  • Near-black forest-green background (#0B1410 to #0F1A14). ONE green accent (#1FB47A / teal-green) for headline keyword, table header rules, and the subtitle banner. Everything else off-white and muted gray.
  • Heavy condensed sans headline (Anton or Archivo Black) for the title. Emphasize 'Loop' in the green accent. Subtitle sits in a thin teal banner bar directly under the title, full width, left-aligned.
  • All content left-aligned. Generous left margin gutter. Section headings in small-caps condensed with a short green tick mark to the left of each heading.
  • Dense 2-column and 4-column tables with alternating row tint (very subtle, 4 percent white). Thin 1px green top border on each table header row, hairline gray separators between rows. Keep cell text tight, single line where possible.
  • Center visual: a five-node circular loop diagram (Ask -> Retrieve -> Verify -> Challenge -> Monitor) with green directional arrows forming a closed ring, placed beside or above 'The Five-Step Loop' section to anchor the page.
  • Use a 'Why it matters / Heads up / Tip / Key' note style: italic or boxed callout with a thin left green border bar, set apart from body text so the teacher framing reads instantly.
  • Footer locked to bottom: thin divider line, WSP wordmark left, follow line right, small green dot separators between phrases. Maintain 1080x1350 portrait safe margins (min 64px padding all sides).

Production checklist

  • Design the 1080x1350 one-pager in the WSP template: near-black forest background, single green accent, condensed headline with 'Loop' emphasized, teal subtitle banner.
  • Build the four data tables (Five-Step Loop, Mistake->Failure->Fix, Citation Gate, What Stays Human) as dense aligned columns with hairline separators and subtle alternating row tint.
  • Draw the five-node circular loop diagram (Ask, Retrieve, Verify, Challenge, Monitor) with closed-ring green arrows and place it next to the Five-Step Loop section.
  • Style the four teacher callouts (Why it matters, Heads up, Tip, Key) with the green left-border note treatment, consistent across all sections.
  • Add header and footer branding: WSP wordmark, follow line, green dot separators, safe-margin check on all four edges.
  • Proof for compliance: confirm no buy/sell/hold, no price targets, no promised returns, no personalized advice, no em dashes. Educational framing only.
  • Export final PNG at 1080x1350 for the LinkedIn feed and a matching PDF version for DM and lead-magnet delivery.
Follow Dave Wang · Wall Street Prompt · For more AI investing workflows · COMMENT & REPOST

CTA

Save this loop and run your next name through all five steps before you size anything. Comment LOOP and I will share the step-three verification prompt I use.

Wall Street Prompt
Wall Street Prompt — internal