Big Data and Machine Learning in Investing: Current Misconceptions and the Path Ahead

Rajesh T. Krishnamachari, Head of Data Science - Equities, Bank of America Merrill Lynch | October 09, 2018

View an extract of this session held at the Generation AI: The New Data-Driven Investor event in September 2018.You can also access the full video.

Rajesh clarifies the difference in the role played by new data vis-a-vis the role played by new analysis techniques. He also classifies data science techniques based on their academic provenance. He argues for certain areas/regimes possessing the most and the least potential for application of big data analysis techniques.

What is Covered in this Presentation

  • Data vs. Techniques
  • What can Machine Learning do for a Quant?
  • FOMO driving $$$ towards Big Data
  • Big Data ≠ Silver Bullet for Macro
  • "Data Science" as Art more than Science

Takeaways

  • Quant more open than ever: still risk-premia and search for orthogonality beyond EM, intraday and derivatives
  • Machine Learning defined as the art of locating non-linear risk-premia dynamically
  • Low SNR and smaller sample size can push to threshold between absurdly simple and naively simplistic models
  • Locate inherently non-linear tasks: sector rotation, portoflio construction, short vol timing
  • No neural networks yet at daily or weekly frequency
  • To come: Tactical , CNN for technical patterns and Strategic deep reinforcement learning




By submitting this form, you agree to RavenPack's terms of service and privacy policy.

Data Insights

Read More