Joseph Chen is a quantitative finance professional with extensive experience building trading systems, backtesting frameworks, and research infrastructure for multiple financial firms.
In this episode, Joseph joins us for a technical and practical conversation about designing institutional-grade quantitative research systems, integrating artificial intelligence into established workflows, validating financial data, and moving systematic strategies from backtesting into live trading.
Joseph explains why asking an AI coding tool to build an entire trading system from scratch can lead to missing components, inconsistent assumptions, and unreliable results. Instead, he argues that AI should operate within a modular framework containing trusted data sources, reusable libraries, clearly defined interfaces, and established backtesting procedures.
We discuss the importance of data integrity and why information from even reputable vendors must still be inspected carefully. Joseph explains how corporate-action adjustments, raw versus adjusted prices, missing observations, timestamps, exchanges, currencies, and instrument specifications can materially change the outcome of a backtest.
The conversation explores the architecture of an event-driven trading system. Joseph breaks down the roles of data loaders, bar engines, strategy modules, portfolio components, order-management systems, matching engines, reporting tools, and monitoring modules. He also explains why intermediate calculations, signals, and order events should be recorded so the entire research process can be audited and reproduced.
Joseph discusses the challenge of supporting multiple asset classes and trading frequencies within one framework. Equities, futures, foreign exchange, fixed income, and options each have different conventions, data requirements, execution rules, and risk characteristics. Similarly, infrastructure designed for daily or minute-level strategies may be unsuitable for high-frequency trading, where order-book data and millisecond-level performance become critical.
We examine the transition from backtesting to simulation and live trading. Joseph explains why the same strategy logic should remain consistent across all three environments, with only the order destination and execution mechanism changing. A realistic simulation engine should reproduce broker callbacks, partial fills, matching rules, transaction costs, and market-specific execution behavior as closely as possible.
The discussion also covers the risks of using AI to optimize strategies repeatedly on the same historical data. Faster experimentation can easily accelerate overfitting, making human supervision essential. Joseph emphasizes that researchers must evaluate whether each modification is economically reasonable instead of allowing AI to search blindly for the highest historical return.
We also explore machine learning in quantitative finance, including the trade-off between predictive power and interpretability. Joseph explains why smaller datasets may be better suited to simpler tree-based models, while deep neural networks or transformers may be appropriate only when researchers have enough data and a strategy that genuinely requires that level of complexity.
Finally, Joseph outlines how he evaluates a new systematic strategy. The process begins by confirming that the existing infrastructure can support the idea, followed by controlled backtesting, parameter-sensitivity analysis, realistic simulation, and deployment with limited capital. Exposure should increase only when live behavior remains consistent with the original research.
A detailed conversation on backtesting architecture, systematic research, AI-assisted development, data integrity, execution simulation, machine learning, strategy validation, and the infrastructure required to transform a quantitative idea into a live trading system.
*DISCLAIMER*
The information shared on this podcast is for educational and informational purposes only and reflects the personal opinions of the hosts and guests at the time of recording. Nothing in this podcast constitutes financial, investment, legal, tax, or trading advice, and nothing should be interpreted as a recommendation to buy, sell, or hold any security, cryptocurrency, derivative, or financial product.
Trading and investing involve substantial risk, including the possible loss of all or part of your capital. You are solely responsible for your own decisions, and you should consult a qualified professional before making financial decisions. By listening to this podcast, you agree that the hosts, guests, and producers are not liable for any losses or damages arising from the use of any information discussed.
