Building a Production Options Trading System

Problem Statement

Over the past five years, I developed a comprehensive options trading system that addresses the complex challenges of real-time market data collection, strategy backtesting, and automated trade execution. The system needed to handle high-frequency data ingestion from multiple providers while maintaining the reliability and precision required for financial operations.

Data Architecture and Collection Strategy

Multi-Provider Data Integration

The system implements a diversified data collection strategy using three primary providers:

Charles Schwab API: Serves as the primary data source for high-volume collection, leveraging account-based access to minimize operational costs while providing comprehensive market coverage.

ORATS: Premium data provider offering superior data quality and specialized options analytics, justified by the enhanced accuracy requirements for strategy validation.

CBOE: Secondary provider maintained for redundancy, though cost considerations limit its primary usage.

Collection Infrastructure

The system performs minute-level data collection across multiple symbols, generating substantial daily data volumes that require careful architectural consideration. This high-frequency collection enables precise strategy backtesting and real-time trading decision support.

Storage Architecture Evolution

Initial Implementation: Compressed File System

The original storage approach utilized proprietary format compressed files with gzip compression. Initially, each collection cycle generated individual files, but performance optimization led to consolidating full trading day data into single files to support backtesting requirements.

This approach optimized for backtesting performance, where algorithms need access to complete daily datasets for comprehensive strategy evaluation across all expiration cycles.

Migration to MongoDB

Analysis revealed significant performance limitations when building web interfaces requiring timestamp-specific data queries. The file-based approach required loading entire daily datasets to access specific time periods, creating unacceptable latency and memory consumption.

Migration Benefits:

• Query Performance: Dramatic improvement in timestamp-specific data retrieval
• Scalability: Better support for concurrent access patterns
• Flexibility: Enhanced ability to support diverse query patterns for different system components

The MongoDB migration remains in progress, with initial performance metrics demonstrating substantial improvements in query response times.

Backtesting Framework Architecture

Strategy Validation System

The system implements over 100 distinct backtesting strategies, each representing different trading approaches and market conditions. These strategies operate sequentially, with each backtest functioning as a comprehensive integration test for the associated trading logic.

Backtesting Methodology

Date Range Processing: Each strategy iterates through specified date ranges, evaluating potential trades on each trading day based on market conditions and strategy parameters.

Range-Based Trade Generation: The system generates high and low price ranges using proprietary algorithms, placing trades designed to profit when these ranges are breached. This approach leverages statistical analysis of market movement patterns.

Position Monitoring: Once trades are placed, the system tracks position performance using collected options data, monitoring for exit criteria including:

• Range breach conditions
• Profit target achievement
• Time-based exit rules
• Risk management triggers

Performance Evaluation

Kelly Criterion Integration: Position sizing utilizes the Kelly Criterion to optimize capital allocation based on historical win rates and average returns, ensuring mathematically sound risk management.

Success Metrics: Strategy evaluation focuses on average return per trade across all positions, providing a normalized measure of strategy effectiveness independent of capital allocation.

Production Trading Bot Architecture

Strategy Deployment Pipeline

Successful backtesting strategies are deployed to production through a code migration process where backtesting logic is adapted for real-time execution. This approach ensures that production trading logic has been thoroughly validated through historical data analysis.

Order Execution System

Limit Order Strategy: The system exclusively uses limit orders to optimize fill prices, particularly important in illiquid options markets where bid-ask spreads can significantly impact profitability.

Price Optimization: Sophisticated algorithms ensure bid-ask spread reasonableness before order placement, with multiple attempts to achieve optimal pricing within acceptable latency constraints.

Execution Latency: The system tolerates 1-2 minute execution windows to prioritize price optimization over speed, appropriate for the medium-term trading strategies employed.

Position Management

Monitoring Infrastructure: The trading bot executes every minute, continuously evaluating open positions against exit criteria including range breaches and profit targets.

Automated Exit Logic: When exit conditions are met, the system automatically executes sell orders using the same price optimization algorithms employed for entry orders.

Expiration Handling: The system manages position expiration scenarios, including worthless expiration and end-of-day liquidation requirements.

Risk Management and Operational Excellence

Data Quality Management

Real-Time Validation: The system implements real-time data quality checks, with NaN detection triggering symbol-level trading suspension to prevent erroneous trade execution.

Market Open Handling: Data quality issues are most prevalent during market open periods, with automatic recovery typically occurring as market conditions stabilize.

Early Assignment Detection: Specialized monitoring identifies early assignment risk for short options positions, enabling proactive risk management.

Operational Monitoring

Comprehensive Logging: All system events are logged for audit purposes, enabling detailed post-trade analysis and system performance evaluation.

Email Alerting: Critical system events trigger email notifications, ensuring rapid response to operational issues.

Audit Trail: Complete transaction logging supports regulatory compliance and performance analysis requirements.

Web Interface Development

Strategy Visualization Platform

Developed a React-based web application hosted on Firebase for visualizing backtesting results and common strategy performance (covered calls, cash-secured puts, wheel strategies). This platform enables rapid strategy comparison and performance analysis.

Options Data Visualization Tool

Created a Think or Swim alternative for displaying collected options data, addressing data quality issues observed in commercial platforms. The interface supports:

• Time-Series Navigation: Cycling through historical timestamps to analyze market evolution
• Trade Simulation: Placing hypothetical trades and tracking performance over time
• Data Accuracy: Displaying verified market data free from the inconsistencies found in commercial platforms

Current Limitations: The interface treats each options leg independently rather than recognizing complex multi-leg strategies, representing an area for future enhancement.

Lessons Learned and Strategic Implications

Testing Strategy Trade-offs

The system employs backtesting as the primary validation mechanism, consciously trading comprehensive unit testing for development velocity. While this approach accelerates feature delivery, it represents a calculated risk in favor of rapid iteration.

Architectural Evolution

The migration from file-based to database storage demonstrates the importance of anticipating diverse access patterns during initial architecture design. Early consideration of query requirements could have avoided the need for significant refactoring.

Data Provider Strategy

The multi-provider approach provides both cost optimization and data quality assurance, demonstrating the value of diversified infrastructure in financial applications where data accuracy directly impacts profitability.

This system represents a comprehensive approach to automated options trading, balancing the competing requirements of data accuracy, execution speed, and operational reliability in a production financial environment.