An Augmented Financial Intelligence Multi-Factor Model
46 Pages Posted: 31 Jul 2024
Date Written: July 01, 2024
Abstract
Augmented Financial Intelligence combines human expert knowledge with data-driven investment processes, particularly focusing on financial market applications. It deals with the curation of human expert datasets through the Superforecasting lense, and its application within artificial intelligence models. Superforecasters are individuals with superior predictive accuracy compared to the broader public. In addition, it is essential to consider financial market specific characteristics in the modelling process to build an augmented financial intelligence system. The following analysis provides a proof-of-concept model and gives a practical introduction to this field for financial market forecasting. At first, human expert data is extracted from analyst recommendations. Afterwards two new Superforecasting factors are created. They capture the recommendations of analysts that are identified as having superior forecasting abilities. For comparison reasons, consensus and random estimate factors are added. These are then used in conjunction with traditional investment factors to create a dynamic multi-factor model using an ensemble model. The analysis shows that aligning the model calibration with financial market specific characteristics increases the accuracy of the machine learning model. Furthermore, it is possible to extract a human expert dataset from analyst recommendation data that improves upon a traditional multi-factor model. Here the right choice of augmented financial intelligence design framework is important.
Keywords: Augmented Financial Intelligence, Human Expert Data, Superforecasting, Analyst Recommendations, Multi-Factor Investing
Suggested Citation: Suggested Citation