Interactions and Interconnectedness Shape Financial Market Research

Abstract

In this article, the authors investigate two fields that might be relevant for financial data sciences. The first issue covers the entire production chain “from orders to prices” by realistically modeling stock exchanges’ microstructure (e.g., NASDAQ and Xetra). Specifically, the authors show how data-driven research can model decisions to place orders and to generate prices by matching orders accordingly. The other issue is price interconnectedness at markets by networks. The authors show that interactions shape a market’s performance. Emergence comprises the interactions at markets; as such, the collective may not be equal to the sum of individual activities. As a consequence, the assumption that markets are in equilibrium and that arbitrage opportunities do not exist can be replaced by more realistic working hypotheses. The authors show with the two examples that market participants interact, learn, and trade. These individual interactions can be described as organized complexity. Whereas calculus may not support explicit modeling of interactions, the age of big data permits their modeling and application of innovative concepts, such as network solutions for asset allocation, which can be modeled using machine learning. This article illustrates that assertion with concrete examples.

TOPICS: Equity portfolio management, statistical methods, simulations, big data/machine learning

Key Findings

• • Market performance at a real micro level becomes a key issue for modern asset management as order book analysis becomes vital, because transactions mean interactions. Big data, computer power, and machine learning allows analysis of this organized complexity.

• • There are two approaches to modelling market performance: “from orders to prices” by doubly stochastic Markov processes and by measuring the interconnectedness of prices and markets via networks; these approaches can be complementarily used in modern research.

• • Network analysis for factors and assets and centrality-based solutions in portfolio management are advantageous for capturing nonlinearity in data. Asset allocation decisions and risk management are examples to consider in machine learning.