How it works

Where the leads come from

Every morning an AI reads the news, SEC filings, government contract awards, regulation dockets, prediction markets and a few other feeds. It's looking for one thing: something real changed, and nobody has traced it to a company yet.

Almost everything gets thrown out. What survives is written up as a lead and scored out of 100 — the score is how worthwhile it is to spend a research session on. The scoring is here.

What you do with one

1. Open a lead.
2. Hit Copy Deep Research Prompt. That puts the prompt below, plus the whole lead, on your clipboard.
3. Paste it into any model with web access. Deep-research or agent modes are much better.
4. Read what comes back.

The prompt checks the facts, works out what the market already knows, looks for better ways to express the idea than the companies named, and tries to kill the thesis.

The prompt

1853 words, plain text.

I am an opportunity-driven public-market investor.

Below is an early-stage investment "opportunity seed" generated by another research system.

Treat it ONLY as a starting hypothesis.

Do not assume its facts, causal reasoning, claims of underappreciation, suggested companies, suggested questions, or proposed trade expressions are correct.

At the same time, do not ignore useful direction contained in the seed. Its QUESTIONS WORTH ASKING, WHERE TO LOOK, proposed causal chain, and suggested companies are research leads to investigate — not conclusions and not limits on the research universe.

I want you to deeply investigate whether there is actually an attractive public-market trade behind it.

## YOUR JOB

Research this from first principles using current, high-quality sources.

The end goal is not merely to determine whether the story is interesting.

The end goal is to determine:

1. whether the underlying development is real and economically meaningful;
2. whether it creates an investable consequence;
3. whether that consequence is already reflected in market expectations;
4. and whether there is a specific security or instrument worth buying, shorting, or watching at a clearly defined price/event.

If there is no attractive trade, say so clearly.

Do not manufacture one.

---

## 1. VERIFY THE UNDERLYING DEVELOPMENT

Determine:

- what actually happened
- which claims in the seed are correct, incorrect, exaggerated, outdated, or uncertain
- the true magnitude of the development
- the timing and implementation path
- whether the development is temporary, cyclical, structural, or reversible

Prefer:

- primary sources
- company filings
- government data
- industry data
- earnings calls
- reputable trade publications
- strong financial reporting

Distinguish confirmed facts from estimates, forecasts and industry claims.

---

## 2. DETERMINE WHETHER THE CAUSAL MECHANISM IS REAL

Trace the economics step by step.

Ask:

- If this development continues, what actually changes?
- Who receives more revenue?
- Who experiences higher costs?
- Who gains pricing power?
- Who loses volume or market share?
- What becomes scarce?
- What bottlenecks emerge?
- What changes in utilization, margins, capital intensity or returns on capital?
- What second- and third-order consequences are realistic versus merely clever-sounding?

Quantify the effect whenever reasonably possible.

Do not accept a causal link merely because it sounds intuitive.

---

## 3. IDENTIFY AND ANSWER THE DECISION-CRITICAL QUESTIONS

Before reaching an investment conclusion, identify the 3–7 questions whose answers would most change the investment decision.

Use:

- questions already raised in the opportunity seed;
- uncertainties discovered during your research;
- valuation questions;
- company-specific questions;
- timing/catalyst questions.

Then actively research those questions during THIS SAME RUN.

Do not simply move answerable questions into an "unresolved questions" section.

A question should remain unresolved only if:

- the necessary information is not public;
- available evidence genuinely conflicts;
- the answer depends on a future event;
- or resolving it would require specialized work genuinely beyond the scope of this research run.

For every unresolved question, explain why it could not be resolved and what evidence would resolve it.

---

## 4. DETERMINE WHAT THE MARKET ALREADY KNOWS

This is extremely important.

Investigate:

- how widely discussed the underlying development already is
- whether the causal investment thesis itself is already being discussed
- relevant stock-price moves
- analyst commentary
- earnings estimates and revisions
- management commentary
- investor narrative
- positioning where reasonably observable
- current valuation versus history and peers
- what expectations appear embedded in the current price

Do not call something "underappreciated" merely because the causal story sounds interesting.

Also do not assume that publicly available information is automatically correctly priced.

The question is not:

"Does the market know about this?"

The question is:

"What does the market currently appear to EXPECT, and could those expectations be wrong?"

---

## 5. SEARCH BROADLY FOR THE BEST PUBLIC-MARKET EXPRESSIONS

Do NOT restrict yourself to companies mentioned in the seed.

Map the relevant ecosystem and look globally when appropriate.

Consider:

- direct beneficiaries
- second-order beneficiaries
- picks-and-shovels
- suppliers
- bottleneck owners
- public proxies for private assets
- overlooked geographic markets
- companies where the development is unusually material relative to the whole business
- potential losers if the short side is cleaner
- ETFs
- commodities
- options or other liquid instruments when genuinely appropriate

The goal is not to generate a long list.

Search broadly first.

Then narrow aggressively.

---

## 6. NARROW TO THE 1–3 BEST CANDIDATES

After mapping the opportunity, select only the 1–3 securities or instruments with the strongest combination of:

- causal exposure
- economic materiality
- mispricing potential
- valuation
- catalyst
- asymmetry

Spend most of the remaining research effort on these finalists.

Do not dilute the analysis by giving equal attention to six mediocre possibilities.

For each finalist investigate:

- exactly how the company is exposed
- what percentage of revenue/profit/value is actually affected
- operating leverage
- competitive position
- balance sheet
- dilution or financing risk
- capital intensity
- cyclicality
- management commentary
- consensus expectations
- upcoming catalysts
- historical precedent
- current valuation
- current price context

---

## 7. CONVERT THE THESIS INTO AN ACTUAL TRADE DECISION

For each finalist, determine whether there is actually a trade AT THE CURRENT PRICE.

Do not simply call a stock expensive or cheap based on a headline multiple.

Use valuation methods appropriate to the business.

For cyclical companies especially:

- examine normalized or mid-cycle earnings
- historical margins
- historical returns on capital
- prior-cycle valuation
- current-cycle position

Do not mechanically rely on trailing P/E when current earnings are unusually depressed or unusually elevated.

Where reasonably possible, build a simple economic bridge:

development
→ operational impact
→ revenue impact
→ margin / EPS / FCF impact
→ reasonable valuation
→ implied equity value

Where useful, construct bear / base / bull scenarios.

Estimate:

- what needs to happen operationally
- what earnings or cash flow could result
- what consensus currently expects
- where your estimate differs from consensus
- plausible downside
- plausible upside
- expected timing

Try to reverse-engineer what assumptions the current stock price already embeds.

The critical question is:

WHAT DOES THE MARKET HAVE TO BE WRONG ABOUT FOR THIS TRADE TO WORK?

---

## 8. LOOK FOR ASYMMETRY

I am particularly interested in situations where some combination exists:

- downside appears reasonably understandable while upside could be substantial
- an important development has occurred but attention has not fully followed
- a company has de-risked something while price still reflects the older risk
- a foreseeable catalyst could cause investors to suddenly care
- the obvious trade has moved but a second-order beneficiary has not
- a public proxy exists for an attractive private asset/theme
- a political, regulatory or technological regime is changing while assets still reflect the previous regime
- several independent developments reinforce the same investment thesis

Do not force asymmetry if it is not there.

---

## 9. TRY HARD TO KILL THE THESIS

Actively find the strongest evidence AGAINST the opportunity.

Ask:

- What am I overlooking?
- Why might the apparently cheap asset actually deserve to be cheap?
- What base-rate or historical precedent contradicts the thesis?
- Could the timing be much longer than expected?
- Could the catalyst already be priced?
- Could demand overwhelm the proposed supply effect?
- Is there dilution, leverage, capital intensity, cyclicality, regulation, competition, customer concentration or execution risk that destroys the opportunity?
- Is the supposedly direct beneficiary actually only weakly exposed?
- Is there a simpler explanation for the observed price/rate/earnings movement?

I want falsification, not confirmation.

---

# FINAL OUTPUT

## Trade decision

START THE REPORT WITH ONE OF THESE:

### BEST TRADE NOW: [company / ticker / instrument]

or

### BEST SHORT NOW: [company / ticker / instrument]

or

### NO TRADE NOW

Do not bury this conclusion.

If there is a trade, immediately state:

- current price context
- why this is the best expression
- what you believe the market is getting wrong
- base-case upside/downside
- bull-case upside
- primary catalyst
- approximate time horizon
- thesis-breaking condition

If there is no attractive trade, say NO TRADE NOW clearly and explain why.

Do not default to "worth watching" simply because uncertainty exists.

Investment decisions always involve uncertainty.

Make the best probabilistic judgment the evidence supports.

---

## What is actually happening

Concise factual explanation of the underlying development and its economic importance.

---

## What the original seed got right / wrong

Separate:

- verified facts
- exaggerated claims
- incorrect claims
- unsupported assumptions
- genuinely insightful parts of the original hypothesis

---

## The causal chain that survives scrutiny

Show only the links that remain defensible after research.

Quantify important links where possible.

---

## Decision-critical questions investigated

List the major questions you identified and what the research found.

These should be ANSWERS, not merely questions.

---

## Best investment expressions

Rank only the strongest candidates you actually investigated.

For each give:

- company / asset and ticker
- current price context
- why it is exposed
- materiality
- valuation
- what consensus appears to expect
- what your view requires the market to be wrong about
- bear / base / bull outcome where reasonably possible
- catalyst
- biggest risk
- conclusion:

BUY NOW
SHORT NOW
WATCH AT [specific price / valuation / event]
PASS

"Worth watching" without a specific condition is not an acceptable conclusion.

---

## The most interesting non-obvious angle

If research uncovered a better second- or third-order trade than the seed proposed, explain it.

This is especially important.

Do not force one if none exists.

---

## What would make me act

If the conclusion is not BUY NOW or SHORT NOW, state the exact:

- price
- valuation
- earnings estimate
- data point
- regulatory event
- filing
- contract
- guidance change
- or other confirmation

that would change the decision.

Make this concrete.

---

## What would invalidate it

Give clear thesis-breaking evidence.

---

## Research questions still unresolved

Only include genuinely important questions that could NOT be answered during this research run.

For each explain:

- why it remains unresolved
- whether it matters enough to prevent a trade
- what future evidence would answer it

Do not use this section as an escape hatch for research that could have been done above.

---

## Overall assessment

Rate the underlying OPPORTUNITY:

- Exceptional / actionable now
- Strong / actionable with attractive asymmetry
- Interesting / specific trigger or price required
- Weak / little investable edge
- Dead end

Explain why.

Then end with one final sentence:

**TRADE: [specific action]**

Examples:

TRADE: Buy XYZ around current levels.
TRADE: Buy XYZ only below $42.
TRADE: Short XYZ if Q4 guidance confirms X.
TRADE: No trade — thesis is real but already reflected in valuation.

Keep the analysis investment-oriented and specific.

Do not give generic portfolio-management advice.

Do not confuse caution with rigor.

Do not force a trade, but do not avoid making a judgment merely because the outcome is uncertain.

---

## OPPORTUNITY SEED TO INVESTIGATE

Notes

• Give it web access. Without it, it will make things up.
• The prompt ends at ## OPPORTUNITY SEED TO INVESTIGATE — everything after that heading is the lead you're feeding it.
• The prompt is built to disagree with the lead. "Mostly already priced" is a valid result.
• The most interesting non-obvious angle section is usually where a second- or third-order idea turns up.