Ryan Whitford

I research and develop systematic trading strategies focused on cross-sectional and time-series momentum within S&P 500 equities. My work involves designing and backtesting factor-based models, analyzing turnover dynamics, and evaluating performance across varying market regimes. I conduct risk analysis using volatility targeting, drawdown assessment, exposure decomposition (sector, beta, and factor loadings), and transaction cost modeling to ensure robustness under stressed and high-correlation environments.

At the Ozcan Lab, I conducted research at the intersection of optics and machine learning, focusing on Diffractive Deep Neural Networks (D2NNs). My work explored how ensemble learning techniques can improve training stability and generalization in physically realizable optical neural networks. I implemented and benchmarked multiple training strategies, including boosting, bagging, and joint training, using PyTorch and NumPy, evaluating model performance across feature-engineered variants and simulated optical propagation constraints. This involved integrating concepts from linear algebra, wave optics, and statistical learning to design controlled computational experiments. Through this research, I developed a deeper understanding of representation learning in physics-constrained systems and presented findings at the UCLA Undergraduate Research Conference, communicating complex ML concepts to both technical and non-technical audiences.