Michael Recce is Chief Data Scientist with Neuberger Berman and is leading the firm’s effort into incorporating data signals into the investment process. When working on predictive algorithms for the advertising industry, he realised that these applications would work just as well in finance. After all, if you know which products customers are interested in then you know which companies are winning, he says.
Michael Recce Podcast overview:
0:30 What is a data scientist?
2:00 Has your study of neuroscience given you new insights into how computers think?
3:00 Predicting what ads people are interested in shows you which companies are winning; that is not priced into stocks.
4:20 Data-science driven investing will be bigger than the quant revolution was
5:00 Bootstrapping problems in data science; you don’t know what you don’t know
5:30 Most of the data that people want to sell is not useful
7:30 Asking questions around time series are not the right questions
10:00 We start with a fundamental model and the fact that it is quantitative means we are using computers.
13:20 What are some of the datasets that you are most excited about today?
15:00 Signal decay, how do you deal with it?
17:00 Data-driven investing is very different from the way most people think about investing
18:30 I pay 1/20th of what a hedge fund pays for the data
20:00 Are we solving the right problems in investing?
21:00 The different time domains of data.
25:00 How to deal with bias in data.
29:00 To what degree is creativity a part of data science?
33:00 Can you build the values that an organisation might have into the algorithms?
34:00 Partnership with the UN on sustainability data
36:00 Is this just added complexity for boards?
40:00 I just finished the first course in teacher portfolio manager how to code.
45:00 My bet is on fundamental manager that learn to code.