Skip links

how we’re different (for those in the back): part 2


What do farmland, timberland, real estate, or oil and gas assets have in common?

I’m sure you’re on the edge of your seat here.

They are all relatively illiquid, infrequently traded, and insulated from commodities speculation, such as options trading. Most transactions occur point to point between individuals and or companies, so there’s less volatility. The constraints of real assets are really around supply limitations and their relatively high transaction costs (title). Still, if we can structure enough data, there are ways to start to overcome these constraints and make transactions more systematic – automating the entire end-to-end process from underwriting -> identifying deal flow/leads/opportunities -> capital deployment -> asset management -> back to underwriting because we learn something when we spend money (if we architect on the front end correctly and instill discipline in our team).

Let’s break this down—first, an introduction.

As I write this post from a WeCrashed in Dallas, in April of 2021, Rich Gomel, most recently managing partner of WeWork’s real estate investment platform, joined as CIO of Two Sigma Real Estate to use computer-driven research and technology to spot opportunities. I found this mind-blowing because a) I’m a nerd, and b) it was the first time I read that a quant like Two Sigma was looking at real assets.

“Real estate is rich in data, but nobody uses it to figure out what to buy or where to buy it,” Tom Hill (formerly of Blackstone) also joined Two Sigma Real Estate and said in an interview. “If you actually have predictive tools where you can forecast demand and supply, it’s really powerful.” It can also help avoid having “your talented investment professionals bark up the wrong tree, go down dead-end alleys, and waste their time on non-deals,” said Mr. Hill, who built Blackstone Group’s alternative asset management unit into one of the world’s largest hedge fund investors. Sounds like someone put on their Arthur Scherbius hat to solve this jigsaw puzzle. 

In his quote, Mr. Hill is referencing underwriting. Still, I would “bet the farm” that they have automated deal origination/lead generation and have a boiler room actively working deal flow at scale. They’ve most likely turned this process into a well-oiled automated machine with technology and data at the core. 

Just curious, has anyone read a Two Sigma LP pitch deck? I would guess that the first ten pages highlight the number of PhDs they employ in computer science, the amount of servers and data they process, and their kitchen sink carries a sign with their favorite W.E. Demming quote. Has anyone read an oil & gas LP pitch deck? I would guess that the storyline goes underwriting ——> mgmt team track record —-> market thesis with one slide at best discussing data and technology, and not a single member of the mgmt team carries the title “Chief Technology Officer,” “Chief Data Officer,” “Head of Data Science,” etc. 


As I mentioned in our first blog – at Oseberg, we have a vision for a systematic process to acquire real assets. Before I show you how we envision this systematic process, I must explain our technology and how we created the data to enable you to get from A-Z. Please allow me to introduce Oseberg’s Full-Text Search, or FTS as a precursor.

Full text search

Full text search

• Search across multiple documents using "full-text" like when using a search engine like Google.
• Quickly and easily identify documents containing people, places, entities or phrases.
• Search both courthouse and OCD documents.
• Research mineral owners to see if they own other tracts of land.

Click Me

FTS is accessible through a web browser.

Document Overload: Find the Damn Needle!

Remember, I said that private individuals comprise the largest mineral ownership component. Land grants and hundreds of years of birth, the three ds, marriage, and associated family tree complexities, have led to a complex tangle of private ownership. To further complicate matters, mineral rights ownership can be (and often is) severed from surface ownership. 100s of millions of documents in courthouse records explain how these minerals have been divided and encumbered and who currently owns them. Additionally, 100s of millions of regulatory documents describe how the states regulate the oil & gas companies trying to extract and produce from these minerals.

There were too many documents to quickly find THE documents we cared about within this massive galaxy of documentation. Did you know the Solar System is about 36 billion times larger than Earth (3.6 X 1010)? It’s not possible for the human brain to even comprehend. 

So after building the tech infrastructure to aggregate 100s of millions of images from thousands of county courthouses and 10s of regulatory agencies by state, pre-process the images to enhance their image or dpi quality, and OCR the documents – we were able to develop a Google-like search interface to locate the images that we cared about once we found our oil and gas leases, mineral deeds, acreage dedication agreements, unit declarations, nonstandard proration units, unitizations, increased densities, poolings, spacings, etc., etc. – then…we needed to work to get the data trapped within those documents, out – and structured into a format that we could do something with.

Not just the high-level structured metadata that we touched on that most of the minidata incumbents in our market focus on – that’s the inexpensive and less valuable data. We needed to crack open the document with a hammer and suck everything out of the image…at scale.


Next chapter.