strategy-draft

name	strategy-draft
description	turn a polymarket thesis into a practical strategy draft. use when the user wants entry ideas, exit logic, invalidation, scenario planning, passive versus aggressive execution guidance, or a trade plan that still stops short of sending orders. use opportunity-classifier or deep-market-research first if the market has not been screened or modeled yet.

strategy draft

Use this skill to answer one question:

given a real thesis, what is the best non-executing plan for entering, managing, and exiting this market?

This skill sits between research and action:

• upstream of order-ticket
• downstream of opportunity-classifier, market-memo, deep-market-research, and actuarial-forecasting

It should produce a plan, not a preview and not a live order.

default mode

• Default to plan mode.
• In plan mode, optimize for execution framing, invalidation logic, and timing discipline.
• Do not drift into a fresh research memo or a live order workflow unless the handoff rules require it.

what this skill is for

Use this skill when the user is asking:

• where should I look to enter?
• should this be passive or aggressive?
• what would invalidate the thesis?
• when should I wait instead of trade?
• what is the best exit logic if the thesis is right?
• how should this idea be framed before building an actual order ticket?

Use this skill for:

• one market with a real thesis already in hand
• strategy refinement after deep research
• converting a fair-value view into execution bands and guardrails
• scenario-aware trade planning that still stops short of preview or submission

Do not use this skill to:

• invent fair value from scratch
• classify whether a market deserves research
• submit, preview, or cancel orders
• output exact execution parameters unless the user explicitly asks for a ticket-ready plan

minimum viable input checklist

Do not finalize a strategy draft until you have, or explicitly mark as missing:

• the target market and outcome side
• a thesis or fair-value view
• current market state
• enough tradability context to discuss passive versus aggressive execution
• at least one invalidation condition
• at least one reason to wait, if waiting is plausible

Minimum upstream support should usually be one of:

• opportunity-classifier output
• market-memo
• deep-market-research
• actuarial-forecasting

If the upstream work is only qualitative and has no real probability view, hand off before drafting strategy.

outside-in frame

Before drafting the plan, inspect the trade from these lenses:

• edge lens: do we have a real reason to think the market is mispriced or likely to move?
• microstructure lens: can we express the view passively, or does the book force aggression?
• timing lens: is this an "enter now" market, a "wait for a catalyst" market, or a "monitor only" market?
• failure lens: what would make the thesis wrong before the market settles?
• workflow lens: is strategy the right next step, or should this still be research, triage, or an order ticket?

If the workflow lens says "not ready", stop and route accordingly.

dependency and handoff rules

If the main missing piece is:

• whether the market is even worth working on -> opportunity-classifier
• a compact read on current market state or mechanics -> market-memo
• fair value, evidence, or catalyst mapping -> deep-market-research
• explicit probability modeling, scenario weights, or numerical fair value -> actuarial-forecasting
• exact execution parameters or preview-safe order construction -> order-ticket

Use actuarial-forecasting when the trade idea still lacks:

• a disciplined probability estimate
• a fair-value range
• a scenario-weighted base case

If that skill is not installed locally, use the fallback at ossianravn/actuarial-forecasting.

main workflow

1. Normalize the assignment

Write a 1-2 line intent statement for yourself covering:

• what market and outcome the strategy is about
• what the user wants: entry framing, wait/act decision, exit logic, or full plan
• whether the strategy should stay high-level or be almost ticket-ready

2. Confirm the thesis basis

Start from one of:

• a market memo
• a deep research result
• a classifier output with a credible directional view
• a probability model or fair-value view

If the thesis is still mostly narrative, stop and hand off upstream.

3. Refresh live market state

Before planning, confirm the live setup again with the minimum needed MCP reads:

• get_market_snapshot
• get_orderbook if entry style or slippage may matter
• get_recent_trades if the book may be stale or flow quality matters

Do not assume the market still looks like it did when the thesis was formed.

4. Decide whether the right action is trade, wait, or pass

Make an explicit call between:

• enter now
• wait for better entry
• wait for catalyst
• pass for now

Do not force an entry plan if the right answer is to wait.

5. Draft the execution shape

Translate the thesis into:

• preferred side
• preferred entry zone or band
• passive-versus-aggressive preference
• conditions that justify acting now
• conditions that justify waiting

Prefer passive entries when spread, depth, and catalyst timing make patience sensible. Prefer aggressive execution only when:

• the edge is time-sensitive
• catalyst risk is imminent
• the book can support the size without unacceptable slippage

6. Define invalidation and exit logic

At minimum specify:

• what disproves the thesis
• what would block entry immediately
• what would force a fast exit
• what would count as first take-profit
• what would count as final exit
• whether there is a time-based exit or review point

The exit plan should be linked to the thesis, not just to a generic profit target.

7. Add sizing discipline without turning into an order ticket

Keep sizing qualitative unless the user explicitly asks for exact size.

Use sizing notes to discuss:

• conviction versus uncertainty
• concentration concerns
• catalyst proximity
• whether the setup deserves only starter size
• whether existing thesis exposure should limit new entry

Point to configs/risk-limits.yaml when relevant, but do not perform order preview here.

hard rules

• Do not submit, preview, or cancel orders in this skill.
• Do not invent fair value when the edge has not been modeled.
• Do not output exact execution parameters unless the user explicitly wants a ticket-ready plan.
• Do not recommend aggression just because the thesis is strong; aggression must be justified by time sensitivity and book conditions.
• Do not confuse "interesting market" with "good setup right now".
• Do not omit invalidation. A strategy without a failure condition is incomplete.
• Do not ignore waiting as a valid recommendation.

output contract

Return exactly these sections in this order.

[strategy title]

intent

State in 1-2 sentences:

• what the strategy is trying to do
• whether the current recommendation is enter, wait, or pass

thesis basis

Summarize:

• the directional view
• the fair-value or research basis supporting it
• the single biggest assumption

market setup

List:

• current implied probability
• best bid / ask or spread context
• liquidity or flow note
• whether the setup favors passive or aggressive execution

entry plan

List:

• preferred side
• preferred entry band
• what must be true to enter now
• what should make you wait
• whether passive limits or aggressive execution are preferred

risk and invalidation

List:

• what disproves the thesis
• what blocks entry now
• what would trigger a fast exit
• the most important unresolved uncertainty

exit plan

List:

• first take-profit zone
• final exit condition
• time-based review or expiry condition if relevant

sizing notes

Keep this qualitative by default:

• starter versus full-size logic
• concentration warnings
• catalyst-driven size adjustments
• references to local risk policy when relevant

next catalyst

List:

• what to watch
• when the setup should be revisited
• what evidence would most improve or damage the trade

next handoff

Choose one:

• remain in strategy-draft
• hand off to order-ticket
• hand off upstream to opportunity-classifier, market-memo, deep-market-research, or actuarial-forecasting

Explain why that is the cheapest useful next step.

completion criteria

The strategy draft is complete only when:

• the thesis basis is explicit
• the current recommendation is clearly enter, wait, or pass
• entry logic is distinct from exit logic
• invalidation is concrete
• passive versus aggressive preference is justified from market conditions
• the user could decide whether to advance to order-ticket
• no live order action is taken

common failure modes

• rewriting the memo instead of drafting strategy
• pretending a thesis exists when the edge was never modeled
• giving a generic "buy dips, sell rips" plan with no invalidation
• skipping the possibility that the right move is to wait
• turning strategy-draft into order-ticket prematurely
• giving sizing language that ignores existing exposure or uncertainty

activation tests

Should trigger

• "turn this thesis into a trade plan."
• "where would you want to enter this market?"
• "should this be passive or aggressive?"
• "what would invalidate the trade?"
• "draft an exit plan for this idea."
• "give me a strategy, but do not place any orders."

Should not trigger

• "is this market worth researching?"
• "estimate the fair odds."
• "preview a 25c bid."
• "submit this order now."
• "summarize the current market state."

resource map

• deep-market-research — use when fair-value work, evidence gathering, or catalyst mapping is still missing
• market-memo — use when current state and mechanics need a compact summary first
• opportunity-classifier — use when triage or prioritization is still the main problem
• order-ticket — use when the strategy is ready to become a guarded execution plan