Examines U.S. households’ interest in AI-based financial advice using nationally representative 2024 NFCS data. Finds that digital readiness, risk preferences, and behavioral frictions are key drivers of AI advice adoption.

Develops a PIDE framework for option pricing with stochastic volatility and jumps, applied to S&P 500 index options across maturities. Finds that stochastic volatility drives most pricing gains, while jumps provide modest improvements mainly for short-maturity, deep OTM options.

Examines what drives reinforcement-learning asset allocation performance across agent design, investment objectives, and rebalancing frequency. Finds that objective choice dominates performance, while frequent rebalancing helps preserve the value of forward-looking signals.

Examines risk-neutral asymmetries in short-rate lattices using option-implied probabilities across maturities and market conditions. Finds that imposing symmetric probabilities leads to bond mispricing, biased return forecasts, and less effective hedging, while option-implied probabilities improve valuation and risk management.

Develops a tilted AFNS model that incorporates option-implied asymmetries into yield-curve dynamics. Finds that these asymmetries improve yield-curve forecasts and capture economically meaningful variation in term premiums.

Demonstrates how FinBERT transforms financial news into actionable sentiment signals for portfolio and risk analysis. Provides a practical workflow for linking headline sentiment to intraday stock-price movements.

Develops a heavy-tail-aware measure of financial uncertainty using a double-subordinated Normal Inverse Gaussian model of S&P 500 returns. Introduces risk–reward measures that improve the identification of uncertainty shocks in financial markets.

Examines Pakistan-exposed ETFs from 2016–2024, assessing their risks, performance, and diversification benefits for international investors. Compares historical and dynamic portfolio optimization to evaluate how adaptive strategies perform in an emerging-market setting.

Develops a framework for distributional Granger causality that captures predictive dependence through conditional means, volatility, tails, asymmetry, and other distributional channels. Introduces an adaptive sequential testing procedure with finite-sample error control and oracle-equivalent asymptotic power, providing a complete and computationally efficient approach to testing distributional predictive dependence.

Examines how platform-specific social media use shapes speculative retail trading using nationally representative 2024 U.S. household data. Finds that community-driven platforms such as Reddit, Discord, and Telegram are associated with greater meme-stock participation, while broadcast-oriented platforms and financial influencers show little predictive power.

We examine whether common AI-generated investment advice contains information about future stock returns or largely reflects signals already embedded in prices. Using empirical evidence from financial markets, it studies how the predictive content of AI advice varies with market conditions and information redundancy, offering insights into the role of AI in price discovery and market efficiency.

This paper examines how survivorship bias distorts the performance of portfolios formed from stocks that remain successful over time, separating the effects of changing portfolio composition from the truncation of the return distribution. The central empirical result is that conditioning on surviving winners can materially alter measured portfolio performance, showing why historical analyses based only on firms that remain successful can overstate the investment characteristics of the broader market.

The paper develops an endogenous portfolio-selection framework that closes the loop between distributional learning and portfolio risk, and provides an empirical decomposition that identifies how much of the resulting change in the efficient frontier comes from the learned distribution itself versus the dependence structure used to construct it.

I study how information choice and misspecified learning interact to shape prices and beliefs in dynamic asset markets. The paper shows how limited information-processing capacity can sustain distinct equilibria and persistent valuation differences.

Studies how the equity pricing kernel changes with investment horizon using S&P 500 options and realized returns. Finds that kernel nonmonotonicity declines sharply from one month to one year, challenging standard preference and stochastic-volatility models and motivating horizon-dependent models of risk and learning.

This paper develops a recursive portfolio-learning framework in which the riskless rate is endogenously determined by the minimum-risk portfolio and the investor’s evolving estimate of return risk. The resulting Adaptive Minimum-Risk Rate (AMRR) provides a portfolio-specific benchmark that converges to a fixed point and links portfolio choice, risk estimation, and equilibrium pricing.

Examines what can be learned from jointly observing point and density forecasts, distinguishing beliefs, preferences, and forecast reporting behavior within a rank-dependent utility framework. Shows that joint forecasts provide additional identification, while some Survey of Professional Forecasters responses are inconsistent with standard preference and admissibility restrictions.

Develops a discrete-time portfolio framework with latent policy regimes in the market price of risk and long-memory FIGARCH volatility, extending dynamic programming to non-Markov volatility states. Applied to Russell 3000 portfolios during U.S. ESG policy transitions, the model finds regime-dependent risk pricing and a sign reversal in the winners–losers spread, implying state-contingent optimal allocation.

Examines how professional forecasters allocate probability across future outcomes using density forecasts from the Survey of Professional Forecasters. Finds systematic tail-risk compression for GDP growth and core inflation, while beliefs for other variables remain close to benchmark predictive distributions.

This paper examines how prediction-market prices respond to public information, distinguishing belief updating from changes in market participation. Using every trade across 240 NFL prediction markets, the study shows that participation and trader composition play an important role in shaping price dynamics.

Develops an intraday framework for forecasting S&P 500 flash crashes using one-minute returns and a double-subordinated Lévy process with stochastic trading intensity and heavy-tailed risk. Shows that model-based tail-risk forecasts, combined with structural-break, downside-risk, and distance-based indicators, can provide economically meaningful warning signals up to 30 minutes before crash onset.

Examines the impact of the 2010 Live Nation–Ticketmaster merger on concert ticket prices using multiple empirical approaches. Finds substantial and persistent price increases, highlighting the need for stronger merger scrutiny and systematic post-merger evaluation.

Develops a portfolio optimization framework that extracts implied probability weighting functions from optimal portfolios under Gaussian and NIG return distributions. Shows that fat-tailed returns and changing risk-free rates amplify belief distortions, with implications for portfolio risk management under extreme market conditions.