Machine learning in empirical asset pricing

The absorption and multiplication of uncertainty in machine‐learning‐driven finance

“…The source code, detailing the sequence of instructions a trading algorithm performs when executing orders, constitutes the “secret sauce” that firms fiercely guard and academics cannot access (MacKenzie, 2018 , p. 1637). The fact that ML is a rather broad moniker and most useful ML systems are tailored to their context of application and comprise several different elements does not make things less complicated (Weigand, 2019 , p. 85). Lowrie ( 2017 , p. 4) describes ML as a “diverse congeries of algorithmic approaches, software implementations of such approaches, and hardware configurations designed to handle such implementations” effectively capturing the intricacies of such technical assemblages.…”

Which COVID-19 information really impacts stock markets?

“…By using factor analysis, we are able to summarise the systematic influences that drive global stock markets and are able to determine the proportion of common global market movements that are attributable to the COVID-19 pandemic. Second, we contribute to the increasing application of ML methods in finance such as explaining stock price movements and variable selection (see for example Patel et al, 2015a , Chatzis et al, 2018 ), filtering information from news to evaluate its impact on stock markets (Atkins et al, 2020; Khan et al, 2020 ) and asset pricing anomalies (such as Weigand, 2019 ; Tobek & Hronec, 2020 ). We add to a growing number of studies using ML methods in various facets of COVID-19 research such as epidemiological, molecular studies and drug development, medical, socio-economic ( Lalmuanawma et al, 2020 , Peng and Nagata, 2020 ;) and financial ( Adekoya and Nti, 2020 , Baek et al, 2020 ; Costola et al, 2020b).…”

Time‐varying neural network for stock return prediction

“…By contrast, machine learning in financial markets is still in its infancy. Weigand (2019) provided a recent survey of machine learning applied to empirical finance and noted that machine learning algorithms show promise in addressing shortcomings of conventional models (e.g., the inability to model nonlinearity or handle a large number of covariates). Recent works have applied neural networks to the problem of cross‐sectional stock return prediction (e.g., Abe & Nakayama, 2018; Gu et al, 2020; Messmer, 2017).…”