Multi-Strategy
Trading Bot

Overview

This quantitative project implements a multi-strategy trading framework designed to deploy equity and options-based signals across multiple horizons. The system coordinates market data ingestion, dynamic signal generation, and stringent portfolio governance.

Key Features & Upgrades

• Heston Stochastic Volatility Engine: Migrated from standard Black-Scholes to a Heston model backed by a Longstaff-Schwartz Monte Carlo solver to capture American-style execution and volatility smiles.
• Diverse Strategy Portfolio: Features cointegrated pairs mean reversion, Donchian channel breakouts, macro lead-lag arbitrage, and options volatility arbitrage (gamma scalping).
• Vectorised Batch Processing: Highly optimised data alignment and path bundle generation for the LSM solver keeps spread bootstrapping execution under a few hundred milliseconds.

System Architecture

The bot relies on a strict separation of concerns. The Data Layer interacts directly with the Alpaca API for deterministic sampling. The Execution Layer handles routing, whilst Portfolio Governance enforces per-strategy capital allocations and limits margin exposure in real-time.

Backtesting & Validation

To aggressively mitigate curve-fitting, the framework incorporates advanced robustness tests rather than relying solely on historical equity curves. This includes permutation testing (separating overnight gaps from intraday moves), synthetic data stress testing (injecting Brownian motion jumps), and Monte Carlo equity cones to calculate precise risk-of-ruin probabilities.