Thanks for being here!
This is the Alternative Data Weekly for Friday, March 22, 2024.
Announcements:
Check out the Fordham Quant Conference April 4th in NYC.
Theme that emerged in this week’s email is … AI models need data, but not just any data…”good data”.
QUOTES
“We believe that B2B and consumer enterprises will continue seeking partnerships with companies to obtain access to their data sets. Buyers and sellers should think creatively and strategically about these broader arrangements—for example, considering the value of a proprietary data-sharing deal, which would provide the buyer with exclusive access but limit the seller’s ability to further monetize, thus demanding a higher price point.” - Solomon Partners
News Articles
Podcasts
Cool Charts
Final Thoughts (Modern Data Stack)
#1 – Solomon Partners published Data Becomes the New Oil: Businesses Have a Unique Opportunity to Monetize Content and Information as Fuel for AI Models. March 2024. (h/t Dan Entrup)
My Take: The above quote was sourced from this article. AI models need data. There is a limitless supply of data, but finding “good” data that won’t leave you in legal gray area puts some limits on the amount of data readily available for training. The $60m Reddit deal got people excited, but we are just getting started. Data has value, but so will what you do with the data.
#2 – Doug Laney & Nathan Whigham published in CDO Magazine Data Is Like Charcoal — Here’s Why. March 2024.
My Take: I had no idea charcoal briquette industry was started as a result of by-products from Ford’s auto manufacturing process. The authors analogize the value seen in data that is created as a by-product of some other endeavor. Much like Ford’s associate Mr. Kingsford was able to create value out of waste by-product, there are many trying to create value from “by-product” data.
What else I am reading:
Adam Braff’s Spring Forward. March 2024.
Vin Vashishta’s Getting Buy-In For Big Requests When No One Wants To Say “Yes”. February 2024.
Ethan Mollick’s I, Cyborg: Using Co-Intelligence. March 2024. PRE-ORDER BOOK: HERE
Jon Keegan published in The Markup What Happens to Your Sensitive Data When a Data Broker Goes Bankrupt?. February 2024.
Vendia’s Navigating data monetization challenges and unlocking revenue streams. February 2024.
Deutsche Bundesbank weekly activity index that uses Fable’s data in it’s methodology. Good demo of alt data’s potential to give an accurate forecast and nowcast of the macro economic environment – in this case with regards to quarterly GDP growth.
TechCrunch published App analytics firm Sensor Tower acquires rival Data.ai and is cutting staff. March 2024.
Source: Hedgineer’s Michael Watson interviews BAM’s Carson Boneck, the Chief Data Officer of Balyasny Asset Management (BAM).
My Take: Great interview. Must listen for anyone working in the institutional investment / data space. Of most interest to me was the focus on having the conversation around “are we driving value” … be open about that question and ask it out loud. The goal for the entire firm is making money for the LPs.
Also good discussion around the importance of getting the basics right. Carson says he was lucky when he started at BAM that there wasn’t a lot of “tech debt” to unwind. The initial focus has to be on ingestion and accessing high-quality data … it starts there.
Other highlights include Carson’s thoughts on working with data, you need: Pragmatism, Humility, Grit. Better to have centralized data team supporting investment professionals across the firm. Philosophy: “If a door opens, walk through it”.
Four Phases of Growth (Minute 11):
Foundational
Operational
Differentiating
Transformational
AI. Today >50% of BAM’s employees are actively using AI. Just getting started. Deployment of AI at BAM (minute 21:00)
Started an applied AI team
PM success team
A lot of AI assistants being built
Trying to be self-service … AI helps with that
Unlocks a lot of time … let the AI do most of the time consuming admin work
Highlights (47-minute run time):
Minute 01:00 – Carson’s professional background
Minute 06:30 – lack of tech debt (open field running) as a startup inside BAM
Minute 09:00 – huge initial demand for getting data into the firm; what are next problems?
Minute 11:00 – four phases of growth
Minute 15:30 – success having (19) people more from centralized data team to investment teams
Minute 21:00 – analyst joining before PM … onboarding PM challenge / AI makes data accessible easier
Minute 27:00 – strategy around fundamental data; IR data; core blocking & tackling data; referential data
Minute 31:00 – strategy around creating a “single pane of glass” data model for the PMs; AI will help here
Minute 33:00 – “Wet CPU” replaced with “silicon CPU” (humans replaced by AI models)
Minute 37:30 – always be questioning: will this create value? Say this out loud
Minute 41:15 – ‘24 off to fast start, what does next year look like? AI guardrails & increase PM engagement
Minute 45:00 – Increase legal and compliance speed / effectiveness (how do you speed this up?)
Source: Gable.ai. The Data Contract Workflow. February 2024.
The diagram below comes from Chad and Mark’s upcoming O’Reilly book on data contracts, which highlights the data contract workflow at a high level. A few important points about this workflow:
While a contract can be introduced by a producer, we argue that it will ideally come from the consumer as data quality is about the fitness of use for a consumer.
Both a producer and consumer need to agree on a contract, and there needs to be an established owner for the contract to be accepted.
Once a contract is in place, it can be enforced even if new producers are unaware of the contract’s existence.
A huge assumption is that both the producer and consumer are within the same business, as enforcing contracts on a third party is often futile without leverage.
Thought this was funny … enjoy the weekend!