Research
My PhD students and I have used machine learning in many areas of finance: algorithmic trading, directional changes, volatility forecasting, future cash flow growth, sentiment analysis for the stock market, real estate investment trusts, and temperature and rainfall weather derivatives. The work falls into four themes.
Evolutionary algorithms and genetic programming for trading
Genetic programming (GP) remains at the core of my work. Much of it uses directional changes, an event-based way of sampling prices that captures turning points that fixed time intervals miss. Recent work combines technical, fundamental and sentiment signals in strongly-typed GP, and optimises trading strategies against several objectives at once.
- Evolving trading strategies using directional changes
- A novel strongly-typed Genetic Programming algorithm for combining sentiment and technical analysis for algorithmic trading
- Multi-Objective Genetic Programming-based Algorithmic Trading, using Directional Changes and a Modified Sharpe Ratio Score
- A genetic algorithm for the optimization of multi-threshold trading strategies in the directional changes paradigm
Deep reinforcement learning for high-frequency trading
We train deep reinforcement learning agents to trade high-frequency currency data sampled with directional changes. A recurring focus is making agents aware of their own position and of trading costs, which is where many published strategies fail in practice.
Large language models and NLP for finance
We use language models both as signals and as tools. Current projects integrate FinBERT sentiment into strongly-typed GP, recover the logical structure of financial filings with large language models, and use LLMs to generate trading strategies.
Forecasting for weather derivatives, real estate and company fundamentals
Weather derivatives have no accepted pricing framework, so we build ML and GP models that forecast the underlying temperature and rainfall. The same forecasting toolkit extends to real estate investment trusts, mixed-asset portfolios and fundamental valuation measures such as free cash flow growth.
- Stochastic model genetic programming: Deriving pricing equations for rainfall weather derivatives
- Machine Learning for Temperature Forecasting in Weather Derivatives
- Improving REITs time-series prediction accuracy using Machine Learning and Technical Analysis indicators
- An in-depth investigation of five machine learning algorithms for optimizing mixed-asset portfolios including REITs
Industry and consultancy
My projects with industry and the public sector are on the Industry & consultancy page.
Joining the group
If you would like to do a PhD with me in one of these areas, or a similar area of finance, please email me a short CV and a one-page outline of the problem you would like to work on.