The 2024 HMDA data was just released by the CFPB March 31, 2025. See the top lenders, their national market share, and whether their production was up or down year over year.
Our (@PolygonRE) charts keep digging deep into HMDA, and therefore deep into the mortgage industry; this time on reverse mortgages (think: your equity decreases as the loan ages). https://t.co/2VAfPzBYPF
We've added to the website a free PDF of the Bayesian Workflow book (for non-commercial purposes, like research and exorcism). We put so much work into this book, hope you all find value in it. https://t.co/f472XjM6cf
Yesterday's solar eclipse didn't just block the sun — it put a dent in Internet traffic. In Iceland, Spain, and Portugal, HTTP requests dropped up to 30% right at peak totality. Not a network issue. Just millions of people stepping outside to look up. https://t.co/cAtMEYevEC
@bcherny Usage Credits seem like a bad deal if I can't dedicate them to Fable: I thought my Opus usage would continue to fill up the first (Max) bar, and only Fable would spill over. But my Opus-heavy workload (ironically, to conserve Fable) burned up that $100 in a couple hours. Am I missing anything? I am trying to get away from bouncing among CC, Codex, and Cursor - but this isn't the way.
Investment-property share of Non-QM purchase lending, 50 largest metros, Range: 4% to 54%. Conventional, 1–4 unit, first lien, closed-end, Non-QM, purchase. 2025 HMDA. HMDAVision.
@mattturck@andrewdfeldman@cerebras This was super informative wrt where Cerebras fits in the AI landscape, their back story, and technical content. Great work.
Caleb Stephen Oliven: love, peace, and joy this birthday and every day - happy birthday, son! Here's a birthday podcast for something different - I think you'll enjoy it: https://t.co/VjcRJwwO45
@eric_seufert This is a tour de force - thank you Eric for writing/posting it. I have been steeped in Galbraith's worldview, no doubt reinforced by the lyrics of (I Can't Get No) Satisfaction along the way, without realizing how things have changed. Working this into my product thesis.
Daniel Joseph Oliven: happy birthday son! I chose this year's birthday song for its effortless cool, but also for its cautionary tale. Water? All my love. https://t.co/WJvKrTLWfy
@PolygonRE has long helped lenders build capacity through mortgage market intelligence: understanding where opportunity exists, which products matter, how competitors are winning, and how to give MLOs a stronger platform for success. We have now added county recorder data, giving us a clearer view of the purchase relationships and production that individual MLOs bring to the table.
https://t.co/Au2mEnyGPJ
Mortgage lenders often frame their MLO strategy as a build vs. buy choice:
Build producers from within, or recruit proven producers from somewhere else.
There is a third way: do both, with better data.
Threading the needle in this post of anthropic has done some bad things for AI governance & the discourse but the actions of this administration are way worse so we need to get a handle on it before stronger models, open or closed, come along soon.
https://t.co/PFu1F0sbmS
This is an incredible writeup - thank-you to the https://t.co/C2Doy079YY engineering team:
mondayDB 3 – Solving HTAP for a Trillion-Table System https://t.co/GS7lyam2V3
@motherduck In DuckLake, with respect to keeping storage and compute close to each other, what are the implications of bring-your-own-bucket on R2 vs. Motherduck-managed storage? And what is the cost difference on this choice?