Chicago Quant Alley: Crypto Trading Simulator & Strategy Optimizer

Chicago Quant Alley is a crypto trading simulator and strategy optimizer that bridges academic concepts from stochastic finance with real-world crypto markets. The project allows you to simulate, backtest, and optimize derivatives strategies using historical data, stochastic modeling, and reinforcement-based tuning techniques.

---

Project Overview

This repository includes:

• Historical data collection from real crypto exchanges
• Strategy implementation for crypto forwards and options
• Backtesting engine with performance evaluation metrics
• Simulation framework using stochastic models
• Multi-Armed Bandit (MAB) algorithms for parameter tuning and adaptive strategy optimization

---

Weekly Progress

Foundations of Quant Trading

• Read: Quantitative Trading by Ernest P. Chan
• Key concepts covered:
  • Strategy design and execution architecture
  • Risk management and capital allocation
  • Building a research and execution pipeline

---

Data Collection

• Collected Options and Forwards data from Delta Exchange API
• Cleaned and stored data in structured formats (pandas DataFrames and .parquet)
• Built utilities for querying, transforming, and visualizing this data

---

Theoretical Foundations

• Completed NPTEL courses:
  • Probability and Statistics – IIT Kanpur
  • Introduction to Research – IIT Madras
• Reading in progress: Stochastic Finance with Python by Avishek Nag
• Key focus:
  • Stochastic differential equations (SDEs)
  • Brownian motion, geometric Brownian motion (GBM)
  • Option pricing theory

---

Simulator Development

• Implemented Python-based core simulation engine for trading
• Modeled crypto asset price paths using stochastic processes (e.g., GBM)
• Developed modular system for testing custom strategies on synthetic and real data
• Plug-and-play strategy interface created for future extension

---

Strategy Optimization with MAB

• Integrated Multi-Armed Bandit algorithms (e.g., Epsilon-Greedy, UCB, Exp3)
• Simulated different reward environments (stationary & adversarial)
• Used bandits to optimize:
  • Strike price selection
  • Entry/exit timing
  • Hedging ratios
• Added performance plots and logging to analyze convergence behavior of bandit arms
• Compared MAB-optimized vs. static strategies in terms of Sharpe ratio, drawdown, and win rate

---

Resources

• Books:
  • Quantitative Trading – Ernest P. Chan
  • Stochastic Finance with Python – Avishek Nag
• Courses:
  • NPTEL - Probability and Statistics
  • NPTEL - Introduction to Research
• Data API:
  • Delta Exchange

---

🛠 Tech Stack

• Language: Python
• Libraries:
  pandas, numpy, matplotlib, scipy, seaborn, scikit-learn, gym, optuna
• Tools:
  • Jupyter Notebooks (research + prototyping)
  • Delta Exchange API (data source)
  • Modular architecture for strategy and simulation logic

---

Final Roadmap

Task	Status
Read Quant Trading by Ernest Chan	✅ Completed
Collect and clean data from Delta Exchange	✅ Completed
Complete NPTEL courses	✅ Completed
Read Avishek Nag’s Stochastic Finance book	✅ Completed
Build simulator for crypto derivatives	✅ Completed
Implement pricing models (e.g., Black-Scholes, Bachelier)	✅ Completed
Build strategy plug-in system	✅ Completed
Add backtesting with performance metrics	✅ Completed
Integrate Multi-Armed Bandit algorithms for parameter tuning	✅ Completed
Visualize performance (win rate, Sharpe, regret curves, etc.)	✅ Completed

---

Contributing

This is a research-driven project; contributions are welcome in:

• Advanced strategy modules (e.g., volatility arbitrage, spread trading)
• Risk modeling enhancements
• Visualization and dashboarding (Plotly, Dash, Streamlit)
• Reinforcement Learning-based tuning (e.g., contextual bandits, policy gradients)

---