Polymarket Divergence Trader

Trades a Simmer-indexed market when your estimated probability diverges from the live market price, with dry-run on sim by default, context checks, reasoning tags, and optional live execution on Polymarket through ClawHub.

This skill is a publishable Simmer trading template for ClawHub. It compares your probability estimate with the live market price, checks market context, and only places a trade when the edge clears a configurable threshold.

This is a template. The default signal is probability divergence: if your model thinks the event is more likely than the market implies, the skill buys YES; if less likely, it buys NO. Remix the signal source with your own model, API, or research. The skill handles trade discipline, operator output, and Simmer trade tagging.

When To Use

Use this skill when an agent already has a probability estimate for a specific Simmer-indexed market and needs a safe execution wrapper around that signal.

Skill Folder

This skill ships with three files:

• SKILL.md
• clawhub.json
• divergence_trader.py

Defaults

• Dry-run venue: sim
• Live venue: polymarket
• Trade size: 5.0
• Minimum edge: 0.05
• Execution mode: dry-run unless --live is passed
• Trade source: sdk:polymarket-divergence-trader
• Skill slug: polymarket-divergence-trader

Required Inputs

The skill expects:

• SIMMER_API_KEY
• MARKET_ID for the target market
• MY_PROBABILITY as a decimal between 0 and 1

Optional inputs:

• TRADE_SIZE
• MIN_EDGE
• REASONING_PREFIX
• LIVE_VENUE to override the live venue if you intentionally want something other than polymarket

Safety Rules

1. Always fetch market context before deciding.
2. Always default to dry-run. Live trading requires explicit --live.
3. Always attach source, skill_slug, and human-readable reasoning.
4. Skip trading when context reports warnings, severe flip-flop risk, or excessive slippage.
5. Return HOLD when the edge is smaller than the configured threshold.
6. Pass the venue explicitly by choosing the correct client venue for dry-run versus live execution.

Decision Logic

1. Read the live market probability from market context.
2. Compute edge = my_probability - market_probability.
3. If edge >= min_edge, buy YES.
4. If edge <= -min_edge, buy NO.
5. Otherwise hold.

Example Commands

Dry-run:

bash

python divergence_trader.py --market-id 12345 --my-probability 0.62

Live trade:

bash

python divergence_trader.py --market-id 12345 --my-probability 0.62 --amount 5 --live

Expected Operator Output

The script prints an operator-style summary:

text

Skill: polymarket-divergence-trader
Venue: sim
Risk alerts:
- none

Decision:
- 12345: TRADE YES size=5.00 mode=dry-run edge=+7.0%

Remix Ideas

• Replace MY_PROBABILITY with a forecast model output.
• Drive MARKET_ID from a market scanner.
• Add bankroll-aware sizing before the trade call.
• Extend the loop to poll on a cron schedule and auto-redeem winning positions.