Alternative Data Weekly #270

Theme: Unwrapping the Boring Stuff That Actually Matters

Dec 26, 2025

I’ll be attending BattleFin Miami - hope to see you there!

Special thanks to our sponsor Maiden Century.

Unwrap a point-in-time score for every ticker that distils the data into a clear long, hold, or short view with measurable conviction - say hello@maidencentury.com.

QUOTES

“To realize AI’s full potential, a strong data foundation isn’t optional—it’s mission critical. If your C-suite still considers data engineering as a support role, you’re already five years behind—and probably training your future competitors.” Chris Child, Vice President of Product, Data Engineering, Snowflake

News
Pods
Charts
Final Thoughts (Merry Christmas)

#1 – The Terminalist published Asymetry is All You Need. December 2025.

My Take: The Terminalist publishes must-read content for anyone in the financial data world. There are quite a few interesting takeaways from this most recent essay.
One idea that resonated with me is that markets are fundamentally driven by information asymmetry. This is the gap between what different participants know & when they know it. The Terminalist walks through some market history and market structures, moving from telegraphs to real-time terminals to searchable research platforms. The advantage has always been in the ability to compress latency and get actionable insight ahead of the competition. This search for an information advantage is always going to new heights, with massive rewards for those who unlock the advantage.

#2 – Shaili Guru published Data Fundamentals for AI Product Managers: What You Need to Know Before Your First Project. December 2025.

My Take: I loved this article. The author keeps it simple while addressing the issues. Getting things organized is the hard part, it is also the important part. The sexy stuff comes later and will fail without a good foundation (see below themes for 2026).

#3 – Adrian Krebs published Alternative Data and AI Trends in 2026. December 2025.

My Take: The author shares six themes for 2026. The two the resonate with me are “Foundations Matter”, and “Centralizing Data”. Both have to do with improving the quality of the AI output. Making data more accessible to both humans and agents (they need different things). There will be renewed focus on getting one’s house data in order before handing over more responsibility to AI systems. I the meantime human “augmentation” will continue to be a big deal.

BONUS: Joseph Miller published AI Engineers are Making a Key Mistake with Context Engineering: They Need a World Model. December 2025. “AI agents need that level of accountability. They shouldn’t demand trust — they should earn it.”

What else I am reading:

Abigail Stewart published Stop Building Faster Horses: What AI Should Really Change in Market Research. December 2025.

Lewis Baker published The Real and the Hard Problem of Data Modelling. December 2025.

Didier Lopes published The bitter lesson of context metadata. December 2025.

Nick Craig & Nakul Kapoor published Data standardization is the ‘trust accelerator’ for broader AI adoption. December 2025.

Alex Pentland and Alexander Lipton published Transformative AI in Financial Systems. December 2025.

Matt Harney published SaaSletter - Best Software + AI Content Of 2025. December 2025.

Daniel Beach published Revisiting Data Quality. November 2025.

Ian from ScrapeOps published Scraping Shock: Why Web Data Is Getting Too Expensive to Scrape. October 2025.

Source: The World’s Largest Open Source Entity Graph | Jose Plehn, AI By the Bay 2025. December 2025.

The AI Alliance and OpenData.org are launching the world’s largest open source entity graph, spanning over 300 million companies, 500 million places, and 1 billion people, including relationships among them.

This data can be used for entity resolution and identity verification in data and AI systems. Jose Plehn is the Founder and CEO of BrightQuery (“BQ: The Factual AI Company”) and oversees BQ’s open data initiatives (opendata.org).

BQ’s mission is to make the world more factual, one dataset at a time.

Jose also serves on the Board of Directors of the AI Alliance (aialliance.org), along with IBM and Meta.

My Take: I really like what Jose is catalyzing with the OpenData.org initiative.
People have the right to be identified correctly by all these systems (AI and otherwise) that are being built. This includes the right be not be found by these systems. This will be the single version of truth and provenance, a key to avoiding AI slop (word of the year!).

“Trying to make the world more factual one dataset at a time.”

Currently identified:
- 300M companies globally
- 500M places of business (addresses)
- 1.25B humans associated with a company
But this data doesn’t just need to be public, data needs to be linked (in a graph). This is the big step in making data AI ready (minute 24:00 - entity resolution from AI prompt to resolution).
This comes from both a legal view (gov’t filings) vs business view (public web). Also a give-to-get process (25:50) … if you add to the data, you’ll get premium data back.
Five pillars of the economy are the building blocks for the graph (legal view vs business view):
- 1. people
- 2. organizations
- 3. legal entities
- 4. locations
- 5. addresses (all addresses are geocoded; lat-long)