PT  - JOURNAL ARTICLE
AU  - Daniel Philps
AU  - David Tilles
AU  - Timothy Law
TI  - Practical Applications of Interpretable, Transparent, and Auditable Machine Learning: An Alternative to Factor Investing
AID  - 10.3905/pa.2022.pa492
DP  - 2022 Apr 27
TA  - Practical Applications
PG  - pa.2022.pa492
4099  - https://pm-research.com/content/early/2022/04/26/pa.2022.pa492.short
4100  - https://pm-research.com/content/early/2022/04/26/pa.2022.pa492.full
AB  - In Interpretable, Transparent, and Auditable Machine Learning: An Alternative to Factor Investing, from the Fall 2021 issue of The Journal of Financial Data Science, Daniel Philps of Rothko Investment Strategies, David Tilles of Mondrian Investment Partners, and Timothy Law of Rothko Investment Strategies make a strong case for using symbolic artificial intelligence (SAI) to drive investment strategies. Artificial intelligence can offer better returns than traditional investing approaches, but many artificial intelligence techniques are not easy to explain, and many are not transparent. While there have been efforts to make such black-box methods explainable, the results can be difficult to interpret as well as misleading, presenting a potentially serious problem for investment managers. In contrast, investment choices determined using factor-based models with linear regression tend to be interpretable, yet the performance of factor-based models has been poor, falling short of expectations. An SAI investing approach offers the potential to significantly outperform traditional factor-based investing while also providing a transparent, easily understandable explanation of the decision-making process.