“How do you come to a rational conclusion as to what a company is worth?” A seemingly simple question with little-to-no clear answer.
For John Alberg, a background in computer science and a passion for machine learning led him to view the problem through the lens of data. “If it is true that you can use publicly available information to buy companies for less than their economic worth,” he thought, “then you should be able to see it in the data.”
And thus was born Euclidean, an investment firm that marries machine learning with a deep value mentality.
Our conversation spanned more than 2.5 hours and covered everything from the basics of machine learning, to the evolution of Euclidean’s approach over the last decade, to the implications of adversarial examples in neural networks.
This podcast, an abridged version of our conversation, picks up the thread mid-way through, where I have asked John to expand upon his experience with his startup, Employease, and how it influenced his value-based thinking at Euclidean.
I hope you enjoy.
If you are not already, subscribe to Flirting with Models here:
Show Notes
1:58 – John discusses selling his startup and the eventual formation of Euclidean.
3:09 – John postulates that if value investing is possible, machine learning should be able to find it in the data.
5:41 – The difference between classification and regression problems.
7:33 – Where do advanced statistical techniques end and machine learning begin?
10:41 – What is deep learning?
14:39 – John discusses how the process at Euclidean has changed over time as the landscape of machine learning techniques has evolved.
16:38 – How deep learning allows you to perform less factor engineering and other interesting features of deep learning.
19:47 – Corey invites some thoughts on the complexity of the problem being asked and its ultimate tractability.
21:35 – “We started to reinvestigate whether deep learning could be successfully used to forecast excess returns, and what we found was that deep neural networks were essentially no better than linear regression at this problem.”
24:56 – Why isn’t this all just an exercise in overfitting?
29:11 – Is machine learning applicable in finance at all where data tends to be non-stationary?
32:19 – John explains what ties Euclidean to the legacy of value investing.
34:10 – John discusses the results of his paper Improving Factor Based Quantitative Investing by Forecasting Company Fundamentals.
38:23 – Corey asks John to discuss the current “p-hacking” debate and whether there are techniques that traditional factor research can learn from machine learning.
42:40 – Corey sends John off on a tangent to discuss adversarial examples and their implications for model fragility.
45:06 – Corey asks John whether machine learning is just an algorithmic arms race or whether it offers a sustainable edge for investors.
47:01 – John offers some suggestions for people who are interested in learning more about machine learning.
48:14 – If you were an investment strategy, what would you be and why?