I propose a method for estimating treatment effect heterogeneity in event studies with staggered adoption. This method serves as an alternative to two-way fixed effects regressions, which have been shown to be near-term biased in these settings. I build on insights from the machine learning (ML)...