Kyle Harrison
talk

Rev 2 — Data Science and the Future of Investing

January 1, 2019 View original ↗

Rev 2 — Data Science and the Future of Investing

Thomas Laffont and Alex Izydorczyk of Coatue with Point72 on building a data-science function inside an investment firm — the fullest treatment in the corpus of Data in Investing as an operating problem rather than a thesis.

Notes

  • Description
    • Thomas Laffont, Co-founder & Sr. Managing Director, Coatue Management
    • Alex Izydorczyk, Head of Data Science, Coatue Management
    • Matthew Granade, Chief Market Intelligence Officer, Point72
    • Panel Discussion - Coatue Management & Point72
    • Though active asset managers have used quantitative techniques for decades, data science has presented them with new challenges and opportunities. In this session, three leaders representing different hedge fund perspectives join a conversation to answer questions and discuss topics including:
    • How are hedge funds leveraging data science to drive innovation and new strategies?
    • How are investment strategies in public and private markets changing with the evolution of data science and modeling practices?
    • What lessons and concepts from finance firms’ quant research experiences have been valuable to incorporate into the new data science regime? Conversely, what must hedge funds “unlearn” from their past ways of operating?
    • What are some concrete tactics and techniques for integrating data science into your fund? Which of those can be applied to businesses outside of the finance sector?
    • How to integrate data scientists with traditional analysts and portfolio managers.
  • “To be good at technology investing, you have to (1) be focused on technology, and (2) be good at technology
  • “What are the key tools we’ll use in our data science efforts?”
    • Our tools guide our investment decisions dogfooding
  • The thesis has evolved from “help us track these companies we’re investing in” to a much broader data science platform
  • “When we first started investing in data science we thought this would be a “nice-to-have” bolt on to our core business. What we’ve realized is that it became much more existential for us.”
  • ==“If we don’t use data in our business we will be out of business.”== -Thomas Laffont Data in Investing
  • We want to be best-in-class with every software tool we use
  • For us to be relevant in the sphere of public, private, growth and early stage companies, etc. we have to be using data
  • On top of that you can layer in an informed view from the Chinese landscape
  • That allows us to invest in a data platform that informs every function area
  • We have a fly wheel of access to best cutting-edge companies (e.g. early stage fund), best broad data sets from our public corporate access
  • Almost everything we do is tracked by someone and they have varying rights to sell that data (e.g. app usage, credit card data, etc.)
  • “You could get to a point where you could forecast the P&L”
  • Public markets are within 2% across analyst estimates
  • On the private side, it’s much more about trends
  • “Venture capitalists [using data] is much more rare today.”
  • “It’s just a matter of time before they all come along.”
  • “If you don’t believe that data can inform what you’re doing, then you haven’t found the right data.”
  • Often working with companies that have a data set that has value and the owner of the data might not even realize is valuable
  • Nielsen, IHS, these companies have all existed for 30+ years; that’s not new. What is new is the ability to deliver that data much more readily. Coatue focuses on
  • Data sourcing is becoming a role in funds; it used to be “vendor management,” but its become much more proactive in identifying data assets that might not be monetized but could be
  • Compliance around data is critical; we have a lawyer whose only job is compliance around any data that comes into the firm to ensure the vendor has the right to resell their data (user agreements, 3rd party data, etc.)
  • Believes there will be an insider-trading case in the future that revolves around inappropriate data access / rights
  • China is further ahead in terms of scale and usage of AI (e.g. facial recognition, robotics, semis, etc.)
  • One of the most interesting trends is seeing open source software communities getting built in China
  • Data driven firm vs. model driven firm - the most advanced data science company (e.g. Netflix) believes you can only inform you, not decide
  • There are some areas that are model driven (e.g. quant fund) vs. data driven (Mosaic (Coatue) insights)
  • Excel files (what we typically think of as models) are actually data models, they’re just managed very manually
  • “How do you hire data scientists who can communicate and build tools that work for non-technical people?”
    • There is a certain data scientist that works really well; someone who can balance the needs of a public analyst, venture analyst, etc.
    • “==The future of new financial analysts; they have to be data literate and in five years they’ll have to be able to write basic code. You won’t have a choice.==” Data in Investing
  • The era of fundamental analysts calling a broker asking for a model, changing a few variables, and making an investment, those days will be over shortly. I don’t think there is any more differentiation in doing that.
  • An analyst in the future will go to a portfolio management with a data-driven insight that will inform their trades. Significantly more valuable than your excel model.
  • “We need to become HBO faster than HBO can become us.” - Reed Hastings at Netflix
  • “We need to adopt and become a leader in data science faster than data science can become us.” - Thomas Laffont
  • The leader of the firm has to believe that this is the future
  • “If you’re at a firm where your skills aren’t being leveraged or you don’t have buy in, it isn’t worth your time to stay there.”
  • “We’re at a rare moment in time with technology.”
  • The Houston Astros fired their last scout Moneyball
  • Has gotten to the point in Coatue where we don’t want to hear a pitch before we see the data view of the business
  • Data Organization
    • Data Engineering: Software and infrastructure
    • Research: Statistcs & R&D
    • Insights: Data products (people with finance + data literacy)
  • Prioritizing for pragmatism; people who aren’t overly concerned with the specifics of the model and more on what can be accomplished
  • “If you had all the credit card data in the world, which company could you model and which could you not?”
    • McDonalds is a franchise; it doesn’t map up to one piece
    • Chipotle owns all their own data
  • Hiring Talent Data Science
    • Alexander Izydorczyk doesn’t just work on their data org; they get to be involved across venture, growth, public, quant, everything
    • “We want to be the best at using data science in investing.”
  • There are a lot of data scientists who are trying to drive 0.00001% efficiencies, and many of those people would rather work with Coatue
  • Technical Specs
    • They had the benefit of building everything cloud-first which has made them much more flexible
    • Heavy users of Scala and R:
    • Scala - Apache Spark; functional language, you can teach Java developers
    • Center of everything in the compute engine is Apache Spark
    • Fix, add, extend a project is much more flexible / possible
  • Either the incumbents will adopt data science, or new entrants will adopt data science and replace the incumbents. Already seeing this in insurance.
  • Companies in Silicon Valley don’t actually realize how valuable data can be to their company. They might not know the data exists or have the ability / capital to access that data
  • Showing portfolio companies the level of data they can get is unreal
  • They’ve evaluated 1.5K data vendors
  • Everyone uses their own data; Data Marketplaces are going to become much more common
  • LinkedIn has a very compelling view of data sets and managing them, making them available, etc.