William Voisine
-
Email
wv20590@essex.ac.uk -
Location
Colchester Campus
Profile
- Derivatives and Options-Informed Portfolio Strategies
- Large Language Models (LLMs) and AI Agents in Financial Forecasting
- Event-Driven and Sentiment-Aware Market Prediction
- Machine Learning for Macro-Economic and Market Signal Integration
Biography
I am a PhD researcher in Financial Forecasting and AI-Driven Portfolio Optimization at the University of Essex. Based near Washington, DC, my work focuses on how Large Language Models (LLMs) and AI agents can synthesize information from diverse financial domainsincluding macroeconomic indicators, market sentiment, derivatives data, and geopolitical eventsto improve forecasting accuracy and decision-making in modern markets. My current research explores a suite of interconnected systems that apply LLMs and agent architectures to financial forecasting and risk management. These projects investigate how market sentiment evolves and decays over time, how multimodal data such as macroeconomic releases, options activity, and policy developments can be fused to anticipate market-moving events, and how agentic frameworks can translate those forecasts into adaptive, derivatives-informed portfolio strategies and real-time risk-monitoring systems. Before joining academia, I spent over twenty years in analytical and systems-engineering roles supporting large-scale public-sector programs. My work included building econometric forecasting models to guide resource allocation, performing statistical analyses on high-volume operational datasets, and designing financial systems to strengthen oversight and governance across acquisition portfolios. My goal as a researcher is to bridge advanced AI modeling and practical portfolio management, developing explainable, data-driven methods that connect the speed of modern machine learning with the discipline of financial reasoning.
Qualifications
-
BSc Computer Information Systems (Husson University) (2002)
-
MSc Systems Engineering (The George Washington University) (2014)
Research and professional activities
Thesis
Agent-Augmented Portfolio Selection: Large Language Models for Event Forecasting and Derivative-Informed Risk Management
This research examines how Large Language Models (LLMs) and AI agents can synthesize information from macroeconomic, sentiment, and derivatives data to improve forecasting and portfolio selection. By modeling the temporal decay of sentiment and predicting event-driven risk, the work introduces adaptive, explainable methods for managing exposure in dynamic markets. It represents a step toward intelligent, forecast-aware portfolio management that bridges machine learning and real-world finance.
Supervisor: Michael Fairbank
Research interests
AI-Driven Financial Forecasting
Developing predictive models that integrate macroeconomic indicators, political and policy events, options-market data, and alternative datasets to improve market regime detection and asset forecasting.
Derivatives and Portfolio Optimization
Exploring derivatives-informed portfolio strategies and risk models that leverage options chains, volatility structures, and sentiment dynamics to enhance allocation and hedging decisions.
Large Language Models and Agent Architectures
Fine-tuning and orchestrating LLM-based agents to interpret multi-source financial data, generate forecasts, and execute adaptive, explainable trading strategies in real time.
Sentiment and Event Dynamics
Modeling how financial news sentiment evolves and decays, and applying those temporal patterns to event-driven forecasting and dynamic portfolio selection.
Contact
Location:
Colchester Campus
Working pattern:
8:00am to 4:30pm Eastern Standard Time