Time SeriesQMLERobust Inference
ARMA/QMLE Under Weak Dependence
Studies inference for ARMA models when innovations may be weakly dependent rather than IID, with emphasis on robust covariance estimation and finite-sample inferential performance.
Research
Research across inference, prediction, and trustworthy systems.
My work spans statistical inference under dependence, predictive modeling and uncertainty, financial time series, and the evaluation of computational systems under realistic operating constraints.
Statistical Inference & Time Series
Inference and forecasting under realistic dependence.
ForecastingFinancial Time SeriesMachine Learning
Financial Forecasting Across Price and Return Tasks
Compares statistical and machine-learning methods using expanding-window, one-step-ahead forecasts, separating price prediction from the more difficult problem of forecasting noisy returns.
Financial ModelingStationaritySimulation
Gaussian Financial Modeling with Bond Factors
Examines the long-run behavior and uncertainty of a Gaussian financial model with bond factors, with current work focused on stationary distributions of key log-scale state variables.
Prediction, Machine Learning & Uncertainty
Reliability across forecast horizons.
Computational Social SciencePredictive ReliabilityUncertainty
Forecasting Relationship Dissolution Across Three- and Five-Year Horizons
Uses longitudinal relationship data to examine how predictive performance, explanation stability, and uncertainty behave across three- and five-year forecast horizons. The emphasis is predictive rather than causal: whether conclusions remain reliable as the forecasting horizon changes.
Trustworthy Digital Systems
Verification under uncertainty and adversarial pressure.
Protocol EvaluationGeofencing
Session-Bound Attendance Protocols
Develops and evaluates a server-authoritative attendance protocol using rotating session-bound QR credentials and uncertainty-aware geofencing, examining credential misuse and false-acceptance/false-rejection trade-offs.
Multimodal FusionFacial VerificationGeolocation
Multimodal In-Person Verification
Investigates how geolocation evidence and facial verification can be combined for in-person verification, comparing fusion strategies through false-acceptance and false-rejection behavior.
Earlier Research
Applied quantitative work.
Public HealthSimulationApplied Modeling
Malaria Modeling & Public Health Analytics
Applied modeling work using intervention analysis, district-level data, and simulation to study malaria-control questions and support quantitative public-health decision making.