Research·Human × AI Conference Submission·June 2026
Why So Serious?
Decomposing the Belief Volatility Smile in Prediction Markets
Written and submitted in four weeks for the UCLA Fink Center Human × AI Conference, this paper studies whether the volatility smile observed in prediction markets reflects genuine information or is largely a mechanical consequence of bounded prices. Using Kalshi FOMC contracts, it develops a logit-space framework to separate boundary effects from belief dispersion and examines how those effects evolve as markets approach resolution.
Developed through a human-AI research workflow spanning literature review, coding, model development, empirical testing, and drafting. The framework is now being extended to a broader universe of event markets.
Research Note·Research Note in Progress·June 2026
Structural Credit Model with Time-Varying Default Barriers
A Disclosure-Based Calibration
This note extends structural credit models by linking the shape of the default barrier to information from firms' 10-K maturity disclosures. Rather than treating the barrier as fixed or purely exogenous, the model allows it to reflect the underlying debt profile and rollover structure of the firm. The result is a more flexible framework for interpreting default risk and distance to default in a way that is closer to the firm's actual financing structure.
Currently being extended into a more detailed version and tested against CDS term-structure data.
