From May 22-24 in New York, Data Disrupt brought together 80 speakers, approximately 30 of whom were alternative data focused. We highlight key takeaways below.
Two New Providers:
Three key themes throughout the conference:
  • Technology is no longer the main objective; data quality is.
  • Investment in alternative datasets, infrastructure, and employees is growing exponentially.
  • Data loading and integration remains a big hurdle for investment teams.
Long-onlies are no longer experimenting with alternative data, they are fully invested in it. Jon Nietzel of GSAM, Jordan Vinarub of T. Rowe, and Nelson Yu of Alliance Bernstein shared various trends on how alternative data is impacting their business.
  • More data or more technology?
    • Nelson believes it depends on each specific company’s core competency, but likely more data makes more sense. Technology will be table stakes.
    • Jon highlighted the importance of process. The ability for different groups (technologists, data scientists, and investment analysts) to communicate with each other is the most important piece to the puzzle.
    • Jordan agreed that collaboration between teams of different skill sets is the most important value-add to the whole process, but also the hardest to get right.
  • Organizational structure
    • Jon: don’t separate your data teams from the investment process. You should align your entire organization’s goals to the same objective - yield better results.
    • Nelson: investors ultimately own their decisions, they can’t hide behind the data - accountability should be shared across all participants in the process.
    • Nelson: you cannot do everything at the same time. You need to evaluate and integrate datasets over time. Scalability over time.
  • Is rise of machines helping make more money?
    • Nelson: technology and data enabled companies are a lot more scalable - the investment process itself becomes more scalable and productive.
    • Jon: time is always under pressure - systems that help accelerate decision-making have a positive impact on the investment process.
  • ROI from alternative datasets?
    • Jon: best insights come from a collection of datasets, so the challenge is attribution to specific datasets.
      • Tip: Look at situations when you or your investment team are under pressure to make a decision. What is the dataset that you rely on the most?
    • Nelson: two elements to the equation: 1) what is the insight that you want to know that would influence your decision, and 2) who gets closer to addressing that insight. Understanding the value proposition.
Vin Vacanti, CEO and Co-founder of YipitData, presented on the State of The Alternative Data Industry, highlighting:
  • Importance of understanding the core competency of a “data player”. Are they data owners, data processors, or data brokers - each has its specific challenges in working with investors.
  • Presented on the most penetrated, most accurate, and least accurate data categories according to institutional investors. (Stats)
  • Shared key takeaways from the growth of the alternative data FTE analysis.
  • Shared key takeaways from the buy-side data spend survey.
Carl Reed, Global Head of Data License Product at Bloomberg, believes that the data engineer role is likely to become more central to the investment process. Is there value in alternative data providers partnering/redistributing through large platforms like Bloomberg, FactSet, or Thomson Reuters?
  • Greg Skibiski of Thasos sees a lot of value in these partnerships and announced Thasos is partnering with Bloomberg. Details to be announced mid-June.
  • Steve Albert from 1010Data doesn’t see the value add of their distribution as a key demand for customers. Not currently solving a major pain point. Does see the risk of massively commoditizing the value of the data.
What do you wish clients were doing more often?
  • That they hired a lot more data full-time employees that could work faster with the data.
  • That they reached out a lot more when they have a thesis to see how data providers can help.
Antonio Tomarchio, Founder and CEO of Cuebiq, shared the pros and cons of geolocation data methodologies:
Methodologies Pros Cons
 Consumer Apps
  • High Quality Data
  • Not Scalable
  • PII Data Collection
  • Scalable
  • Persistent
  • Accurate
  • Low Precision
  • PII Data Collection
 Ad Calls
  • Scale
  • Easy to Obtain
  • Inaccurate Data
  • Lack of Privacy
  • Risk of Fraudulent Data
  • Accurate
  • Privacy Compliant
  • Persistent
  • High Quality Data
  • Hard to Scale

  Other Data Discussions:
Yielding actual results from data requires domain-specific knowledge beyond standard data processing capabilities. Rohit Kapoor, CEO of EXL Service, an operations management company, spoke about Making Digital Transformation Real.
  • He asserted that the actual business benefit of digital transformation is a far cry from the promise of technology and data analytics.
    • 70% of transformation programs fail. (McKinsey)
    • 48% of companies are not prepared to execute business transformation. (HBR)
    • Only 30% of data assets are used. (EXL)
    • Only 20% of companies have implemented AI into a service or process at scale. (Forbes)
    • Only 3% of companies data meets basic quality standards. (BCG)
  • The missing ingredient is what they call “digital intelligence.”
    • Combine the actual data with the capability of the people they have. Marriage between domain knowledge and data drives context.
    • EXL solves this by pairing data scientists with those who run operations, bringing them together to create contextual knowledge.

Investing in culture is essential to tapping into AI on an organizational level. Jeff McMillan, Chief Data Officer of Morgan Stanley, gave his thoughts on AI Today, Tomorrow, and 2050.

  • Today, how good is AI at different tasks?
    • To detect: machine > human
    • To process: machine > human
    • To recommend: machine = human
    • To answer: human > machine
    • To reason: human >> machine
  • Tomorrow:
    • His goal is not “artificial intelligence”, it is to build an “intelligent organization.”
    • Believes that 450 25-year old data analysts with good data are better than 25 experienced data scientists.
  • 2050:
    • Computers will augment human ability. AI is not going to replace humans any time soon, so you have to invest in culture, skill set, and resources.