Thanks for being here!
Announcements:
I plan to be in NY May 16-17-18. Happy to get together if you will also be in the area.
We have a new daily data product that offers end-of-day account balances across various types of accounts (savings, checking, personal, business, etc). Anonymous small regional & community banks. No PII. Very granular.
Contact me with questions.
Theme that emerged this week is … the tools to engage with Alternative Data are improving.
News Articles
Podcasts
Cool Charts
Final Thoughts (The World is Getting Better)
#1 – Qaisar Hasan & Stan Altshuller of Maiden Century published Challenges of Implementing Alternative Data in Investment Management. April 2023.
My Take: Great title. There is a long list of challenges. Incorporating new data into the investment decision process is hard to do, but the payoff is becoming more evident. If done well, this data will become an essential part of your day-to-day decision-making process. If you do this poorly, you (the data professional) will lose the confidence of the end user, something that is tough to regain. Remember the portfolio manager is taking career risk with every buy/sell decision. Your data needs to deliver. Every time.
#2 – Matthew Bernath published Data Ecosystems: Privacy-Preserving Data Sharing for a Better Tomorrow. April 2023.
My Take: Privacy is a big deal, but regular readers know that I believe attitudes are rapidly changing about privacy, specifically with the younger generation caring a lot less about privacy of their data (that said, obviously no one wants to be robbed or doxed). The article points out that there is a TON of valuable data at big companies that can be more easily shared and used once privacy concerned are removed. All data is more easily used once PII can reliably be removed without impacting the quality of the data and insights. We are early, but better tools to ensure data privacy will open up monetization opportunities.
#3 – Maxime Beauchemin published Mastering AI-Powered Product Development. April 2023.
My Take: We are all prompt engineers. “Prompt engineering” is essentially asking good questions of your AI models. I am seeing in my daily use of ChatGPT that the better, more detailed question I ask, the better the response. Domain expertise becomes more important for data professionals as the tools to extract the answers becomes more easily accessible & available via tools like ChatGPT.
BONUS: Marina Goche published in AlphaWeek How Investors Can Use Digital Sentiment OT Achieve Above-Benchmark Returns. April 2023. “Investors today need to supplement their analysis and due diligence with alternative data sets that monitor investment signals as they emerge in digital channels.”
BONUS 2: The Bid By Blackrock published in Seeking Alpha How Geospatial Data Can Inform Investing Decisions?. April 2023. “Alternative data can be big or small, but usually it's strange, unstructured, not mapped to an investment that you're trying to analyze.”
BONUS 3: Alt Credit Models and Platforms Improve Financial Inclusion for Credit Marginalized Consumers. April 2023. “A growing movement to use alternative forms of payments data — phone and electric bills, and other utilities, rent, and personal loans — can help would-be lenders broaden the field of credit-worthy borrowers.”
What else I am reading:
Doodling Data Cards: Cocktails, Spirits, & Data
The Web Scraping Club: Web Scraping and Alternative Data for Financial Markets
IDSO TOPICS: Personally Identifiable Information (PII) – yes, this is me dorking out
#1 – Adel Nehme of DataFramed published Building Great Machine Learning Products at Opendoor. April 2023.
My Take: This is an interview with Sam Stone, Director of Product Management at OpenDoor. I was drawn to this interview by my interest in better understanding real world use cases for ML products. Two things of particular interest. First, the importance of building trust with users by showing your work. Second the benefit of cross-functional teams that have both the data skills AND the domain experience. That combination is unicorn status.
Three keys for building a great ML product:
build a product that stems from deep understanding of user behaviors
optimizing for accuracy in a way that is aligned with what user wants
build feedback loops from the beginning
Highlights (39-minute run time):
Minute 03:20 – interview starts; What makes a great ML product?
Minute 04:40 – how ML drives value at OpenDoor (pricing, recommendation, personalization)
Minute 08:30 – importance of understanding the user (customer)
Minute 12:30 – incorporating user feedback
Minute 14:30 – “model interpretability” … can the user understand what you are delivering?
Minute 20:30 – don’t assume what is interpretable by you and your team is interpretable by end-users
Minute 22:20 – discussion of metrics & evaluation (what are “NorthStar” metrics)
Minute 27:00 – common pitfalls on “NorthStar” metrics (i.e. too academic)
Minute 29:00 – what does getting user feedback look like in practice
Minute 30:30 – importance of cross-functional partnerships (“throw it over the wall” approach does not work)
Minute 32:00 – important “rituals” to build culture
BONUS: The new podcast I See Data People has been posting a number of solid interviews … short & to the point. These posted just since April 8th:
08- Isabel Yishu Yang
07- Dan Huss Episode
06- Mark Fleming-Williams Episode
05- Yedidya Gorsetman Episode
04- Jame Kardatzke Epidsoe
03- Ed Lavery Episode
02- Anju Marempudi Episode
01- Dan Entrup Episode
Source: Qaisar Hasan & Stan Altshuller of Maiden Century published Challenges of Implementing Alternative Data in Investment Management. April 2023. This was summarized above too.
Data discovery is a broken process.
More data produces better results, but this is tough to do well (MC can help!).
Source: The World is Getting Better
It is easy to get caught up in the latest crisis: Global warming; The lingering effects of Covid; The war in Ukraine; Classified documents in the wrong hands; Chaos at the border.
If you listen to the nightly news for 20 minutes you’d think we were living in a wasteland. It is enough to make you depressed.
Don't be fooled. We are living during wonderful times. There has never been a better time to be alive.
Understand that good news often happen slowly & we are impatient.
More people than ever have access to education.
Human health is better.
Child mortality has vastly improved.
More people are living in democracy.
More people have more resources (money, plentiful food, clean water, healthcare).
The world is generally at peace.
Zoom out. The picture gets better.