Evaluating the Performance of Machine Learning Algorithms in Financial Market Forecasting: A Comprehensive Survey

Abstract

With increasing competition and pace in the financial markets, robust forecasting methods are becoming more and more valuable to investors. While machine learning algorithms offer a proven way of modeling non-linearities in time series, their advantages against common stochastic models in the domain of financial market prediction are largely based on limited empirical results. The same holds true for determining advantages of certain machine learning architectures against others. This study surveys more than 150 related articles on applying machine learning to financial market forecasting. Based on a comprehensive literature review, we build a table across seven main parameters describing the experiments conducted in these studies. Through listing and classifying different algorithms, we also introduce a simple, standardized syntax for textually representing machine learning algorithms. Based on performance metrics gathered from papers included in the survey, we further conduct rank analyses to assess the comparative performance of different algorithm classes. Our analysis shows that machine learning algorithms tend to outperform most traditional stochastic methods in financial market forecasting. We further find evidence that, on average, recurrent neural networks outperform feed forward neural networks as well as support vector machines which implies the existence of exploitable temporal dependencies in financial time series across multiple asset classes and geographies.

The quest for business value drivers: applying machine learning to performance management

“…Our first research question stemmed from the observation that the identification and measurement of business value drivers is characterized by high levels of complexity, of which methodological and analytical ones constitute the core of PMS development (Okwir et al 2018, Santos, Kelly 2010, Belton and Howick 2002. In this respect, one of the main issues of conventional approaches is that the crucial drivers of performance are not always identified (Kaplan and Norton 1996) and their impact on each other and the overall performance is often overlooked (CIMA 2014, Franco-Santos, Bourne andLucianetti 2012) and not adequately measured (Silvi et al 2015, López-Ospina et al 2017.…”

“…According to Okwir et al (2018) systematic literature review, issues related to the identification of business value drivers and their measurement can be linked to two forms of complexity: methodological and analytical. Methodological complexity is associated with the varied, often conflicting and dynamic nature of organisational objectives (Sundin, Grandlund, and Brown 2010), which challenges the managerial ability to identify the business value drivers (Benson-Rea, Brodie andSima 2013, Santos, Belton andHowick 2002). Aside of bounded rationality arguments (Cyert and March 1963), their choice and understanding would reflect managers' mental models (Vandenbosch and Higgins 1996) and their own views and interpretation of organisational goals and what determines business success (Hall 2011, Ferreira andOtley 2009).…”

“…More specifically, methodological complexity refers to problems associated with the identification of business value drivers. These arise from the multiple internal and external demands organisations typically face and which make their performance the result of a number of different factors (Benson-Rea, Brodie andSima 2013, Santos, Belton andHowick 2002). On the other hand, analytical complexity concerns the specific selection of the related measures and the understanding of the interdependences between different variables and aspects of the business.…”

“…For instance, ML algorithms outperform conventional statistical approaches when there are a multitude of variables, their relationships are not linear (Granger and Terasvirta 1993), the value of the variables continuously evolves and data availability is extensive (Sokolova and Lapalme 2009). A key strength is their ability to learn from past data and adapt, which allow them to make predictions in conditions that mimic real-world, changeable and complex combinations of factors (Ryll andSeydens 2019, Syam andSharma 2018). As explained above, the identification of value drivers falls in this class of issues, thus leading to hypothesize a potential role of ML in supporting the identification of the value drivers.…”

Cryptocurrency market trend and direction prediction using Machine Learning: A Comprehensive Survey

“…Therefore, a researcher must embed the ML prediction model in a real-world trading strategy and then evaluate its performance utilizing standard trading-related metrics (e.g., ROI, RSI, and Sharpe ratio). For a relative perspective on performance, the proposed ML models should be benchmarked against the previously presented ideal classifier [89]. Movements in the crypto-market are interlinked with the news on the cryptocurrency [90].…”