Welcome!

I lead Finance Data Science & Strategy at Airbnb. Most of my work is on Bayesian methods for compositional and hierarchical time series, plus Bayesian decision theory. Lately I've been writing more about forecast governance, the practical question of how teams use, communicate, and act on the predictions they generate. Ph.D. in Statistics from UCLA.

✉ Email 📚 Scholar 💻 GitHub 🔗 LinkedIn

News+

Jun 2026Directional-Shift Dirichlet ARMA Models for Compositional Time Series with Structural Break Intervention is forthcoming at the International Journal of Forecasting.

Jun 2026New arXiv preprint: When Should Forecasting Models Be Re-Specified? A Cost-Sensitive Trigger for Adaptive Model-Form Updating (link).

Apr 2026New article in Foresight: Retraining as Approximate Bayesian Inference (link).

Mar 2026B-DARCH paper published in the International Journal of Forecasting (article, preprint).

Research+

Most of my work falls into four areas: compositional and hierarchical forecasting, delay-aware evaluation, platform accommodation markets, and Bayesian decision theory.

Compositional and Hierarchical Forecasting

Directional-Shift Dirichlet ARMA Models for Compositional Time Series with Structural Break Intervention

Forthcoming, International Journal of Forecasting.

arXiv Code

A Bayesian Dirichlet Auto-Regressive Moving Average Model for Forecasting Lead Times

With Kai Brusch & Robert Weiss. International Journal of Forecasting, 2024.

Article arXiv Code

A Bayesian Dirichlet Auto-Regressive Conditional Heteroskedasticity Model for Forecasting Currency Shares

With Robert Weiss. International Journal of Forecasting, 2026.

Article arXiv Code

Centered MA Dirichlet ARMA for Financial Compositions

Revise and resubmit, Journal of Forecasting.

arXiv Code

Sensitivity Analysis of Priors in the Bayesian Dirichlet Auto-Regressive Moving Average Model

With Liz Medina & Robert Weiss. MDPI: Forecasting, 2025.

Article arXiv Code

Forecasting the U.S. Renewable-Energy Mix with a Bayesian Dirichlet ARMA Model

With Thomas Maierhofer. Renewable Energy Forecasting: Innovations and Breakthroughs, 2025.

Article arXiv Code

Delay-Aware Evaluation

Distributional Fitting and Tail Analysis of Lead-Time Compositions: Nights vs. Revenue on Airbnb

With Jess Needleman & Liz Medina. Under review, Annals of Tourism Research: Empirical Insights.

arXiv

Two-Part Forecasting for Time-Shifted Metrics

With Erica Savage & Kai Brusch. Foresight: The International Journal of Applied Forecasting, 2025.

Article arXiv Code

Impact by Design: Translating "Lead Times in Flux" into an R Handbook

Monitors full lead-time distributions and turns them into a pickup-risk bound. Note, 2025.

arXiv Code

Platform Accommodation Markets

Coupled Supply and Demand Forecasting in Platform Accommodation Markets

Revise and resubmit, Tourism Management.

arXiv

Forecasting the Evolving Composition of Inbound Tourism Demand: A Bayesian Compositional Time Series Approach Using Platform Booking Data

Working paper.

arXiv

Lead Times in Flux: Analyzing Airbnb Booking Dynamics During Global Upheavals (2018–2022)

With Erica Savage & Peter Coles. Annals of Tourism Research: Empirical Insights, 2025.

Article

Slomads Rising: Structural Shifts in U.S. Airbnb Stay Lengths During and After the Pandemic (2019–2024)

With Erica Savage. MDPI: Tourism & Hospitality, 2025.

Article

Bayesian Decision Theory

Cost-Sensitive Retraining via Posterior Learning Debt

Working paper.

arXiv

Re-Specification as Approximate Bayesian Inference

Under review, Foresight: The International Journal of Applied Forecasting.

When Should Forecasting Models Be Re-Specified? A Cost-Sensitive Trigger for Adaptive Model-Form Updating

Technical note, 2026.

arXiv

Retraining as Approximate Bayesian Inference

Foresight: The International Journal of Applied Forecasting, Q2 2026.

Article

Writing+

Less formal writing, mostly on the Airbnb Engineering Blog, about how forecasting actually gets used in practice.

When History Fails You, Borrow from Geography

Airbnb Engineering Blog, 2026.

Read

What COVID Did to Our Forecasting Models — and What We Built to Handle the Next Shock

Airbnb Engineering Blog, 2026.

Read

Talks & Media+

Jun–Jul 2026A Bayesian Dirichlet Framework for Compositional Time Series Forecasting. 46th International Symposium on Forecasting, Montreal, Canada.

Jun 2026When Forecasting Isn't About Accuracy. Guest appearance, IIF Forecasting Impact podcast (listen).

Apr 2026Forecasting in the Wild. Practitioner talk, North Carolina State University, Raleigh, NC.