context-create-news-to-market

name	context-create-news-to-market
description	Turn a news headline or event into a well-formed prediction market

News to Market

Transform a news article, headline, or event into a properly structured prediction market with clear resolution criteria.

When to Use

The user has a news event or topic and wants to create a prediction market around it.

Steps

1. Evaluate the input — is the event resolvable, binary, timely, and interesting?

   • Resolvable: can the outcome be definitively determined?
   • Binary: is it clearly yes/no?
   • Timely: will the outcome be known within a reasonable timeframe?
   • Interesting: would people trade on this?

2. Identify the claim type — event-by-deadline, threshold, period-gated, durational, or none/never. This determines how the oracle can resolve.

3. Draft the question — start with "Will...", include a specific outcome and timeframe. Keep under 150 characters. Create a shorter shortQuestion variant.

4. Write resolution criteria — the most important part:

   This market resolves YES if [specific condition].
   This market resolves NO if [the market end time passes without the condition being met].
   
   Evidence sources: [@handle1, @handle2] on X/Twitter.
   [OR] Evidence sources: Official announcements from [specific outlets].
   
   Clarifications:
   - [What counts/doesn't count]
   - [Edge cases]
   

5. Choose evidence mode:

   • social_only if X/Twitter accounts will cover the event
   • web_enabled if official data sources (SEC filings, sports scores, government sites) are needed

6. List sources — for social_only: specific X handles. For web_enabled: types of authoritative sources.

7. Set end time — give buffer after the expected event for evidence to appear. If the event happens at 3pm, set end time to midnight or later.

8. Submit:

   • MCP: context_agent_submit_market({ formattedQuestion, shortQuestion, marketType: "OBJECTIVE", evidenceMode, resolutionCriteria, endTime: "2026-06-15 23:59:59", timezone: "America/New_York", sources: ["@handle"], explanation: "Brief reason" })
   • SDK: ctx.questions.agentSubmitAndWait({ market: { formattedQuestion, shortQuestion, marketType, evidenceMode, resolutionCriteria, endTime, timezone, sources } })
   • Then create on-chain: ctx.markets.create(submission.questions[0].id)
   • CLI: context questions agent-submit-and-wait --formatted-question "..." --short-question "..." --market-type OBJECTIVE --evidence-mode web_enabled --resolution-criteria "..." --end-time "2026-06-15 23:59:59" --timezone "America/New_York" --sources "@Apple,@tim_cook"

9. Verify — call context_get_market with the returned market ID to confirm it was created.

Gotchas

• End time too tight. If the event happens at 3pm, don't set end time to 3:01pm. Evidence needs hours to appear on social media or news. Add at least 4–6 hours of buffer.
• No sources specified. Without sources, the oracle has no eligible evidence. Always specify X handles for social_only, or describe authoritative sources for web_enabled.
• Ambiguous criteria. "If it goes well" — what does "well" mean? Define every subjective term. Write criteria as if drafting a contract.
• Combining conditions. "Will X happen AND Y happen?" — make two separate markets. The oracle resolves one question at a time.
• Period-gated confusion. Don't write criteria that imply early resolution for end-of-period questions. "Will X close higher on Friday?" cannot resolve before Friday's close.
• Wrong evidence mode. Using social_only for an earnings report that will be on the SEC website but not tweeted by anyone specific. Use web_enabled for official data.
• Submission takes 30–90 seconds. The oracle processes the draft, validates criteria, and may reject if the market doesn't meet quality standards.

Verification

• context_get_market returns the created market with status "active".
• Resolution criteria in the created market match what you submitted.
• Evidence sources and evidence mode are correctly configured.

See Also

• Resolution Criteria Guide — Deep dive on writing criteria
• Agent Submit API — Full field reference and types