Test Trading

AION Market Divergence Trader

This skill is a publishable trading template for AION Market. 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 process. The skill handles trade discipline, operator output, and AION trade tagging.

When To Use

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

Skill Folder

This skill ships with three files:

Defaults

Required Inputs

The skill expects:

Optional inputs:

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 or flip-flop risk. 5. Return HOLD when the edge is smaller than the configured threshold.

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:

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

Live trade:

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

Expected Operator Output

The script prints an operator-style summary:

Skill: aionmarket-trading
Venue: polymarket
Risk alerts:
none
Decision:
MARKET_ID: TRADE YES size=5.0 reason=edge +7.0%

Remix Ideas

print(f"Wallet: {wallet}") print(f"CLOB API Key: {creds.api_key}")

2. Register with AION Market

Step 4 — Configure Risk Limits

Step 5 — Kalshi Prerequisites (Optional, for Kalshi venue)
Ensure Solana wallet is available (SOLANA_PRIVATE_KEY / userPublicKey)
Use kalshi_quote() to request unsigned tx, locally sign, then call kalshi_submit()
For BUY flows, provide userPublicKey so KYC/geo checks can be evaluated upstream
Workflow: Market Search & Context
Understanding Market Data Structure
> Critical: get_markets() returns events, each containing a nested markets[]
> array of sub-markets. Trading fields like conditionId, clobTokenIds,
> outcomePrices, negRisk, and orderPriceMinTickSize live on the sub-market, NOT the event.

Search returns events

for event in events: print(f"Event: {event['title']}") for sub in event.get("markets", []): # Parse JSON strings prices = ast.literal_eval(sub.get("outcomePrices", '["0.5","0.5"]')) token_ids = ast.literal_eval(sub.get("clobTokenIds", '[]')) print(f" Sub-market: {sub['question']}") print(f" conditionId: {sub['conditionId']}") print(f" YES price: {prices[0]}, NO price: {prices[1]}") print(f" YES token: {token_ids[0] if token_ids else 'N/A'}") print(f" negRisk: {sub.get('negRisk', False)}") print(f" tickSize: {sub.get('orderPriceMinTickSize', 0.01)}") print(f" minSize: {sub.get('orderMinSize', 5)}")

Get Market Context (REQUIRED before trading)

print(f"Name: {context.get('name')}") print(f"Position: {context.get('myPosition')}") print(f"Risk limit: {context.get('riskLimit')}") print(f"Warnings: {context.get('warnings')}")

if context.get("warnings"): print("⚠️ Review warnings before proceeding!")

Automatic Pre-Trade Readiness Checks
Before signing or submitting any order, the agent should automatically verify:
wallet credentials are already registered, otherwise register them
collateral balance is sufficient for the intended spend
Polygon gas balance is sufficient if an onchain approval may be needed
the required spender allowance already exists for the intended order
If allowance is insufficient and gas is available, the agent should send the approval transaction automatically instead of asking the user to do it manually.
Workflow: Trade Execution
Building a Signed Order (REQUIRED)
The trade() endpoint requires a complete EIP712-signed order from py-clob-client.
Passing "order": {} will cause a credential lookup failure — the server reads
order.signatureType to find your wallet credentials.

1. Initialize CLOB client with creds (derived earlier in bootstrap)

2. Extract trading params from sub-market (NOT event)

3. Build signed order — MUST use CreateOrderOptions for proper signing

4. Convert dataclass to dict for JSON serialization

Place a Trade

Default Execution Policy
For general-purpose execution, use this default policy unless the user overrides it:
1. prefer market mode
2. use 2 USDC as default buy size
3. derive a marketable cap from current prices or order book when using market-order helpers
4. ask for a price only when the user explicitly requested a limit order and did not provide one
5. report the market snapshot and final order parameters without blocking on extra confirmation
> Common pitfalls:
>
> - "order": {} → server cannot find signatureType, returns credential error
> - Using event numeric ID as marketConditionId → use sub-market hex conditionId
> - Missing walletAddress → "walletAddress 不能为空" error
> - Not setting neg_risk=True for neg-risk markets → invalid signature, Polymarket rejects
> - orderSize below orderMinSize → order rejected
> - Insufficient USDC balance → server returns generic INTERNAL_ERROR
> - Insufficient allowance → downstream exchange cannot spend collateral
> - SDK result is empty or generic → verify recent trades and orders before reporting failure
Order types: GTC (default), FOK, GTD, FAK
Sides: BUY, SELL
Outcomes: YES, NO
Workflow: Kalshi Quote -> Sign -> Submit
Use this flow for Kalshi venue execution.

Kalshi currently does not expose a dedicated cancel API in this SDK. Use a SELL flow (kalshi_quote -> sign -> kalshi_submit) to reduce/exit exposure.
Workflow: Order Management

Workflow: Heartbeat / Briefing
Call periodically (every 30s–15min depending on strategy frequency).

Workflow: Redemption
When a market settles, redeem winning positions:

Error Handling

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