> This is a template.
> The default signal is keyword discovery + LMSYS Chatbot Arena leaderboard monitoring — remix it with AI benchmark APIs (MMLU, HumanEval), tech news RSS feeds, SEC EDGAR filings for IPO signals, or GitHub commit activity as model release early warning.
> The skill handles all the plumbing (market discovery, trade execution, safeguards). Your agent provides the alpha.
AI markets are the fastest-growing category on Polymarket in 2026. AI investment drove >90% of US GDP gains in H1 2025. This skill trades:
Key insight: AI news cycles are fast and retail trades on headlines. Informed traders with benchmark data have significant edge.
AI_TECH_KEYWORDS = [
"AI", "GPT", "Claude", "Gemini", "OpenAI", "Anthropic", "Google",
"model", "benchmark", "AGI", "ChatGPT", "LLM",
"IPO", "valuation", "Stripe", "Databricks", "SpaceX IPO",
"Apple", "Vision Pro", "Tesla FSD", "Optimus", "robot",
"NVIDIA", "GPU", "H100", "datacenter", "quantum",
"EU AI Act", "regulation", "congress"
]
| Parameter | Default | Notes |
|---|---|---|
| ----------- | --------- | ------- |
| Max position size | $40 USDC | High-liquidity markets allow larger size |
| Min market volume | $10,000 | AI markets are deeply liquid |
| Max bid-ask spread | 6% | Tight spreads expected in popular markets |
| Min days to resolution | 5 | News cycles move fast |
| Max open positions | 10 | AI/tech is broad category |
After a major AI announcement (e.g., GPT-5 release), retail traders overweight recency and push other company markets down more than warranted. Fade the overreaction on competing platforms.
Markets often price on last-known model rankings. When a new benchmark publishes at ~midnight UTC, markets take 1–4 hours to fully reprice. Tight monitoring loop captures this window.
clawhub install polymarket-ai-tech-trader
Requires: SIMMER_API_KEY environment variable.
Runs every 5 minutes (/5 *). AI news breaks around the clock; tightest loop of all category skills.
The skill defaults to paper trading (venue="sim"). Real trades only execute when --live is passed explicitly.
| Scenario | Mode | Financial risk |
|---|---|---|
| ---------- | ------ | ---------------- |
python trader.py | Paper (sim) | None |
| Cron / automaton | Paper (sim) | None |
python trader.py --live | Live (polymarket) | Real USDC |
The automaton cron is set to null — it does not run on a schedule until you configure it in the Simmer UI. autostart: false means it won't start automatically on install.
| Variable | Required | Notes |
|---|---|---|
| ---------- | ---------- | ------- |
SIMMER_API_KEY | Yes | Trading authority — keep this credential private. Do not place a live-capable key in any environment where automated code could call --live. |
All risk parameters are declared in clawhub.json as tunables and adjustable from the Simmer UI without code changes. They use SIMMER_-prefixed env vars so apply_skill_config() can load them securely.
| Variable | Default | Purpose |
|---|---|---|
| ---------- | --------- | --------- |
SIMMER_MAX_POSITION | 40 | Max USDC per trade (full conviction) |
SIMMER_MIN_TRADE | 5 | Min USDC per trade (weak conviction floor) |
SIMMER_MIN_VOLUME | 10000 | Min market volume filter (USD) |
SIMMER_MAX_SPREAD | 0.06 | Max bid-ask spread (0.10 = 10%) |
SIMMER_MIN_DAYS | 5 | Min days until market resolves |
SIMMER_MAX_POSITIONS | 10 | Max concurrent open positions |
SIMMER_YES_THRESHOLD | 0.38 | Buy YES if market probability ≤ this value |
SIMMER_NO_THRESHOLD | 0.62 | Sell NO if market probability ≥ this value |
Position size scales automatically with signal strength — no flat bets.
SIMMER_MIN_TRADE, default $5)SIMMER_MAX_POSITION, default $40)Example YES trades:
| Market probability | Conviction | Size |
|---|---|---|
| -------------------- | ------------ | ------ |
| 38% (at threshold) | 0% | $5 (floor) |
| 30% | 21% | $8 |
| 20% | 47% | $19 |
| 5% | 87% | $35 |
| 0% | 100% | $40 |
This means the skill automatically bets more when the edge is larger, and less when the signal is marginal.
共 3 个版本