convexpi-arena 0.2.0

ConvexPi Arena — discrete-time limit-order-book exchange simulator

convexpi-arena

Discrete-time limit-order-book exchange simulator for quantitative finance education and research.

pip install convexpi-arena

Part of the ConvexPi platform. See also convexpi-lab for the daily-data research harness.

Quick start

from convexpi.arena import Market, Agent, MarketState
import random

class MyAgent(Agent):
    def on_tick(self, state: MarketState):
        if state.mid and random.random() < 0.1:
            return [self.limit('buy', round(state.mid) - 5, 10)]
        return []

market = Market(n_background_agents=20)
market.run(ticks=1000, agents=[MyAgent(cash=10_000)])

Run the Arena server

convexpi-server                          # WebSocket on :8765
convexpi-server --tick-interval 0.2     # faster
convexpi-server --admin-token secret    # instructor console

Connect a remote agent:

from convexpi.arena import RemoteAgent

class MyAgent(RemoteAgent):
    def on_tick(self, state):
        if state.mid and state.position < 50:
            return [self.limit('buy', round(state.mid) - 5, 5)]
        return []

MyAgent().run('ws://localhost:8765', name='my-agent')

Features

• Price-time priority matching engine
• Background agent population: noise traders, market makers, momentum, informed trader
• Avellaneda-Stoikov optimal market making
• TWAP execution agent
• Risk engine: drawdown limits, position limits, force liquidation
• Crypto L2 replay (Binance / Coinbase candle data)
• WebSocket server with admin-triggered volatility shocks
• Rich terminal visualizer

License

MIT © Shane Conway