Fahd Chaudhry

Designed and implemented a multi-module proprietary trading system for forecasting and trading volatility across FX, commodities, indices, and US equities.
• Built and deployed advanced statistical models including GARCH, ARIMA, HAR, HAR-J, and realized volatility frameworks to forecast market variance and regime shifts.
• Engineered machine learning pipelines (e.g. XGBoost, LSTM) for directional prediction and volatility classification, integrated with a feedback loop for model refinement.
• Developed custom volatility clustering, regime persistence scoring, and implied-realized volatility divergence detection to generate alpha under varying market conditions.
• Constructed a real-time signal scanner to monitor over 1400+ instruments on MT5, including a custom-built feedback and journaling system powered by GPT AI for setup validation.

Built predictive models for oil time spreads using co-integration, mean reversion, and volatility clustering in Python (Pandas, NumPy, statsmodels).
• Analyzed CFTC positioning data, macroeconomic releases, and inventory reports to generate directional trade signals.
• Automated curve visualization and signal backtesting tools using ICE/CME tick data, improving execution speed and research turnaround.
• Refined the desk’s edge scoring framework by integrating quantitative factors with trader conviction to assess setup quality.
• Supported real-time volatility monitoring and event risk tracking through custom dashboards for spread and flat price books.

Wells Fargo Corporate & Investment Bank – London, UK
June – September 2024
• Trading Strategy Monitoring & Signal Automation: Supported systematic trading desk by helping monitor strategy behavior during live sessions and developing automated post-trade signal diagnostics, aligned with risk and alpha expectations.
• LLM-Enhanced Risk Intelligence: Developed a prototype LLM pipeline using LLaMA 3 to filter high-volume financial news and firm announcements, improving risk flagging on the equity research desk. Achieved a 56% improvement in NLP sentiment scoring and a 17% increase in risk-alert accuracy compared to baseline methods.
• Factor Signal Research & Replication: Assisted in replicating and validating quantitative factor strategies across multiple regions using Python and internal factor libraries. Reduced research processing time by 80% via integration with an internal knowledge extraction API.
• Earnings Forecast Automation: Built a machine learning forecasting model using LSTM and macroeconomic features to predict EPS outcomes for large-cap UK and US equities. Delivered a 53% reduction in mean absolute error vs. consensus benchmarks.
• Long-Term Desk Projects: Contributed to an exploratory study on systematic volatility timing signals using historical tick data and GARCH-based variance filters, laying the foundation for a real-time volatility adjustment tool.
• Technical Stack: Python, Pandas, NumPy, SQL, Bash, joblib, Matplotlib, scikit-learn, LSTM, Git, Linux.