Five analysts, four cities, one standard. Every thesis you read in the app was written by a named human you can look up here. Past contributors are archived. The history of every handoff is preserved. We do not use anonymous or algorithmic theses. We will not start.
A thesis is a structured document, not a vibe. Fifteen fields are required at publication; one is conditional. Every field has a reason. Every reason is documented in our internal style guide, which we publish on the engineering blog when the next revision goes live.
Every thesis Tarn has ever published is here. Open positions. Closed positions. The ones that worked. The ones that didn't. The version of the document on the day the thesis broke, the post-mortem, the lessons we wrote down for next time. We do not quietly delete losers.
Four times a year, the desk publishes a long memo. It is the macro overlay that informs our bond sleeve, the sector-by-sector update on the curated equity universe, the ESG and macroeconomic risks we are watching, and a section we call "things we got wrong" — short notes on positions that didn't behave the way we said they would, before we knew why.
It runs 20–40 pages. It is the document we'd most want a thoughtful skeptic to read. It is freely available — you do not need a Core account to read it.
Read Q1 2026 memo (PDF) →A small universe, deeply known, is more useful than a large universe loosely covered. We cap our active equity coverage at roughly a hundred names. The bond sleeve is largely passive ETFs with a macro overlay. We do not cover crypto, options, margin, or leveraged products. We do not cover IPOs in their first six months. We do not cover anything we cannot describe in plain English.