Field Notes, Volume Zero
This is a public data journal. The idea is simple enough: I spend time downloading, parsing, and analyzing financial datasets, and instead of letting those notebooks sit in a folder on my laptop, I put them here.
What you’ll find here
Mostly credit markets. The kind of data that shows up in analyst reports as a single chart with a footnote that says “Source: Federal Reserve.” I’m interested in where those numbers come from and what the raw data actually shows when you look at it directly.
The main datasets right now:
- ABS-EE — loan-level auto credit data from SEC EDGAR securitization filings
- SLOOS — the Federal Reserve’s survey of bank lending standards
- FRED — macroeconomic series (NFCI, credit spreads, delinquency rates)
More as the pipelines get built.
What this isn’t
Not financial advice. Not a trading signal service. Not trying to have takes. The analysis here is descriptive — here’s what the data shows, here’s the methodology, here’s where it’s incomplete. Conclusions are, per the tagline, questionable.
How the charts work
The dashboards on this site query live Parquet data from a private Cloudflare R2 bucket. The queries run in your browser via DuckDB-WASM — a WebAssembly build of DuckDB that runs client-side with no backend server. The data travels from R2 to your browser and nowhere else.
The data lake pipeline that ingests this data lives separately at bens-data-lake. This repo is just the visualization layer.
That’s it. More notes when there’s something worth writing.