NLP, Machine Learning, and Quant – Top Recent Papers
Nowadays, many discussions of quantitative methods revolve around factor investing, but quant can also take on a variety of other forms. Machine learning (ML) techniques are increasingly being used to enhance financial models, finding new statistically significant variables that would likely go undetected by the human eye.
State Street looks at how we can validate the output from ML-enhanced models as both trustworthy and applicable. McKinsey & Company explains the basic premises of quantum computing and the beneficial characteristics of qubits for investment managers of the future. Meanwhile, investment managers of the present who are at all curious about the formulas that constitute index mathematics should turn their gaze to the second paper on the below list from S&P Dow Jones Indices.
RECENT QUANTITATIVE PAPERS
Despite the recent underperformance of quantitative strategies, David Blitz of Robeco believes that the quant winter could soon be coming to an end.
Ever wondered about the mathematics that underlies equity index calculations? In this 71-page publication, S&P Dow Jones Indices explains the methodology behind every single one of their indices.
This paper uses a type of real asset sensitivity analysis to demonstrate how chief investment officers can build custom commodity benchmarks.
For compliance reasons, this paper is NOT accessible in the United States
Amundi seeks to propose a framework for liquidity stress testing. This article is the first of a three part series on liquidity risk in the asset management industry. It contains an in-depth analysis of redemption shocks.
FTSE Russell examines the behaviour of four factors (Size, Value, Momentum, and Quality) across the past 60 years of economic cycles.
The author uses machine learning to build an index of financial conditions, composed of variables that are able to predict future unemployment data.
FactSet uses machine learning to enhance a model that assesses the potential for outperformance within the China A-share market. Five lessons are then provided as a guide for machine learning implementation.
Qubits could hold the key to an immense amount of processing capability. This has the potential to benefit complex models within the investment industry and many others that rely on large data sets.
Are machines reliable stock pickers? State Street looks at the drivers behind the predictions of machine learning models in order to reveal output that is interpretable, investable, and interesting.
Beginning with the work of J. Piotroski on the value factor, the author uncovers several statistically significant variables to build upon earlier work and formulate a new and improved model.
This two-part series discusses the definition of alternative data, the forms that it can take on, and many of its applications for investment professionals.
Private equity firms have the ability to manipulate the very distributions and capital calls that impact their internal rates of return, and many choose to do so. How can sophisticated investors better account for this phenomenon?
A factor-based model from Axioma is used to examine the importance of fixed income sector allocations in credit portfolios, with surprising results.
For compliance reasons, this paper is only accessible in certain geographies
Invesco’s Q4 edition of their Risk & Reward magazine focuses upon the acceleration of innovation, via new inventions and digitalisation. The second article discusses the application of Natural Language Processing techniques to uncover investment trends.
SAVVY INVESTOR AWARDS
The winner of the 2020 Savvy Investor Awards "Best Quant" paper was CFA Institute Research Foundation. For more information about their paper, as well as the other Highly Commended papers recognised in the Awards, click on the link above.