Writing

The covenant extractor: how it reads a credit agreement, how it's scored against labels I haven't reviewed yet, what 184 documents cost to run, and where it's still wrong.

Building a complete financial dataset out of primary sources: point-in-time fundamentals and full statements for every US filer, Treasury and Fed credit history, and a corpus of credit agreements underneath structured covenant terms that cite the exact line they came from.

A portfolio system I built for the UCLA Anderson Student Investment Fund: accounting off one transaction ledger, per-name research and notes, attribution and risk, and the weekly board report.

Two pitches on the same credit four weeks apart, the 2028s at +394 and then the 2029s at +520. Both were through target by Aug-2026, and from a common December start the 2029s returned 390bp more.

The 2031 first-lien term loan at par against the same issuer's unsecured notes at 104–105, where the call schedules capped total return.

Written the week DLO fell from $12.60 to $8.87 on a short report and a soft print. The take rate call was right and compression ran further than I expected. The conclusion I drew from it, that growth would stall without new merchants, was wrong.