Quantitative Economist & Data Scientist

A multi-agent reinforcement learning simulation of a rental housing market that asks whether a shared algorithmic pricing service (modeled on RealPage's YieldStar) can push rents above competitive levels. Uses 2,000 heterogeneous tenants from ACS Sacramento microdata and a shared-parameter PPO pricing policy. A full-factorial sweep across 144 configurations finds the shared algorithm consistently learns supra-competitive prices, with a doubling of competing landlords cutting PPO prices by roughly 55%.

A live cross-platform arbitrage bot trading weather contracts between Polymarket and Kalshi. XGBoost ensemble forecaster trained on millions of observations with daily automated fine-tuning, wrapped in a fully automated lifecycle — polling, parsing, inference, evaluation, retraining, and SQLite storage — deployed on a remote Linux server with a live monitoring dashboard. Achieved ~$3,000/day in paper-trading profit.

A Benefit-Cost Analysis of Chico's proposed Mid-Term High-Frequency Service Project for the B-Line transit system, introducing 15-minute weekday frequencies and Transit Signal Priority on Routes 3 and 14. Following 2025 USDOT guidance over a 20-year horizon with a hybrid Prophet ridership forecast, the analysis monetizes travel time savings, external highway cost reductions, and net emissions to characterize NPV and BCR under multiple discount-rate and ridership scenarios.

Replication and extension of a study on gasoline price elasticity of transit demand, using a monthly panel for 10 major U.S. urban areas (2005–2019). Baseline elasticity estimates are highly sensitive to serial correlation — Arellano cluster-robust standard errors inflate by 94–138%. A nonlinear threshold specification finds significant transit substitution (elasticity 0.115) only when fuel prices exceed $3.00/gallon. Multi-source data pipeline integrating NTD, EIA, FRED, and ACS in R.

Compares simple, expectation-augmented, and lagged formulations of the Phillips Curve. Basic models explain little of the variance in inflation, while adding expectations and lags meaningfully captures the dynamic relationship between unemployment and inflation — with implications for policy and forecasting.

A review of the history and limits of economic forecasting, from the Phillips Curve to the Lucas Critique and modern DSGE and machine learning models. Argues that while forecasting guides policy, the complexity of economies and unexpected shocks make long-horizon accuracy inherently uncertain.