simmer-sdk 0.2.0

Python SDK for Simmer prediction markets

Simmer SDK

Python client for trading on Simmer prediction markets.

Trading Modes

Simmer SDK supports two trading modes for different use cases:

Training Mode (Sandbox)

Import markets as isolated sandboxes for RL training and development:

# Import a Polymarket market as sandbox (training mode)
result = client.import_market("https://polymarket.com/event/btc-updown-15m-...")

# Trade in isolation - no other agents, no impact on production
client.trade(market_id=result['market_id'], side="yes", amount=10)

Best for:

• RL training with thousands of exploration trades
• Strategy backtesting without affecting real markets
• Development and debugging
• Ultra-short-term markets (15-min crypto predictions)

Production Mode (Shared Markets)

Trade on existing Simmer markets alongside AI agents and other users:

# Get active markets where Simmer's AI agents are trading
markets = client.get_markets(status="active", import_source="polymarket")

# Trade alongside GPT-4o, Claude, Llama and other agents
client.trade(market_id=markets[0].id, side="yes", amount=10)

Best for:

• Benchmarking your bot against Simmer's AI agents
• Real multi-agent price discovery
• Production deployment after training

Workflow

1. Train: Import markets as sandbox, run RL training loops
2. Evaluate: Deploy trained model on shared production markets
3. Benchmark: Compare your bot's P&L against Simmer's native agents

Installation

pip install -e sdk/

Quick Start

from simmer_sdk import SimmerClient

# Initialize client
client = SimmerClient(
    api_key="sk_live_...",
    base_url="http://localhost:8000"  # or https://api.simmer.markets
)

# List available markets
markets = client.get_markets(import_source="polymarket", limit=10)
for m in markets:
    print(f"{m.question}: {m.current_probability:.1%}")

# Execute a trade
result = client.trade(
    market_id=markets[0].id,
    side="yes",
    amount=10.0  # $10
)
print(f"Bought {result.shares_bought:.2f} shares for ${result.cost:.2f}")

# Check positions
positions = client.get_positions()
for p in positions:
    print(f"{p.question[:50]}: P&L ${p.pnl:.2f}")

# Get total P&L
total_pnl = client.get_total_pnl()
print(f"Total P&L: ${total_pnl:.2f}")

API Reference

SimmerClient

__init__(api_key, base_url)

• api_key: Your SDK API key (starts with sk_live_)
• base_url: API URL (default: https://api.simmer.markets)

get_markets(status, import_source, limit)

List available markets.

• status: Filter by status (active, resolved)
• import_source: Filter by source (polymarket, kalshi, or None for all)
• Returns: List of Market objects

trade(market_id, side, amount)

Execute a trade.

• market_id: Market to trade on
• side: yes or no
• amount: Dollar amount to spend
• Returns: TradeResult with execution details

get_positions()

Get all positions with P&L.

• Returns: List of Position objects

get_total_pnl()

Get total unrealized P&L.

• Returns: Float

import_market(polymarket_url, sandbox=True)

Import a Polymarket market for trading.

• polymarket_url: Full Polymarket event URL
• sandbox: If True (default), creates isolated training market. If False, would create shared market (not yet supported).
• Returns: Dict with market_id, question, and import details

# Import 15-min BTC market for RL training
result = client.import_market(
    "https://polymarket.com/event/btc-updown-15m-1767489300",
    sandbox=True  # default - isolated training environment
)
print(f"Imported: {result['market_id']}")

find_markets(query)

Search markets by question text.

• query: Search string
• Returns: List of matching Market objects

get_market_by_id(market_id)

Get a specific market by ID.

• market_id: Market ID
• Returns: Market object or None

Data Classes

Market

• id: Market ID
• question: Market question
• status: active or resolved
• current_probability: Current YES probability (0-1)
• import_source: Source platform (if imported)
• external_price_yes: External market price
• divergence: Simmer vs external price difference
• resolves_at: Resolution timestamp (ISO format)
• is_sdk_only: True for sandbox/training markets, False for shared markets

Position

• market_id: Market ID
• shares_yes: YES shares held
• shares_no: NO shares held
• current_value: Current position value
• pnl: Unrealized profit/loss

TradeResult

• success: Whether trade succeeded
• shares_bought: Shares acquired
• cost: Amount spent
• new_price: New market price after trade