Why do startups increasingly choose to trade on private secondary markets (PSMs)? I examine the trade-offs associated with this decision and the characteristics of startups trading on PSMs. Using a recent regulatory change that expanded investor access to PSMs and drawing on novel startup share price data, I show that PSM-listed startups face higher fundraising uncertainty and typically experience declining share prices once listed. My results reveal a risk-return trade-off, in which startups accept increased financing risk for higher returns when fundraising succeeds and broad stock markets are supportive, benefiting from price discovery without foregoing private status.

Presentations: FMA Annual Meeting (Doctoral Student Consortium and Special PhD Paper Presentation) (2025), UZH Rising Scholar Conference in Finance (2025), Dauphine Finance PhD Workshop (2025), SFI Research Days (2025), EPFL-UNIL PhD Workshop (2024, 2025)

I study whether predictive technologies such as machine learning (ML) methods can improve deal selection in venture capital (VC) and whether investors can systematically translate predictive advantages into economic gains. Using a large panel of U.S. financing rounds, I show that portfolios constructed using ML signals derived exclusively from past and publicly available information to select out-of-sample deals outperform most investors in the U.S. VC market. I find that these gains are larger when models are trained to predict rare outcomes, such as IPOs and acquisitions, and when applied to early-stage deals. Overall, these ML-based portfolios tend to outperform a large share of investors—often those outside the top quartile in terms of success rates. Despite this potential incremental performance (“benefit”), I find that its persistence is weak within investors: once investor and time fixed effects are accounted for, benefit is strongly mean reverting, both in existence and magnitude. Exploiting a plausibly exogenous shock to the implementation cost of these technologies, I show that investors with higher benefit ex ante tend to have it quickly eroded post-shock, suggesting that the existence of exploitable predictive gains is structural rather than behavioral.

Presentations: EPFL-UNIL PhD Workshop (2024), SFI Research Days (2024), 9th Annual Cambridge Conference on Alternative Finance (2024), World Finance Conference (2024)

Institutional Venture Capital (IVC) Firms are responsible for most of the successful outcomes in the U.S. VC market, with a distinct (separable) and incremental (positive) effect toward success.

Presentations: EPFL-UNIL PhD Workshop (2023), Junior Academics Research Seminars (JARS) (2024)