Date of Degree
Finance and Financial Management
fundamentals, company valuation, factor model, cointegration
This dissertation examines the structural identification of pair trades based on company fundamentals, stock price paths, and company’s capacity to transform fundamentals into value. The dissertation consists of four chapters. Chapter 1 lays the foundation for the study of pair identification. It designs the pair trading procedure and defines the trading performance measure. It also reviews and compares the performance of commonly used pair identification metrics in the literature, including normalized price squared distance, return correlation, and co-integration tests. Among the three metrics, the squared price distance represents the most effective metric and generates the best pair trading performance. The chapter also uses a simulation exercise to highlight the danger of false identification for all three metrics.
Chapter 2 proposes a set of new pair identification metrics based on pairwise similarity measures on value multiples. The chapter examines how the fundamental-based distance measure can enhance the accuracy of the pair identification and the performance of the pair trading strategy, in combination of the price squared distance measure and industry classification. The analysis shows that the value multiple squared distance measure provides useful information on the pair identification, and the information content is different from that of the price squared distance measure. Combining the two can significantly enhance the pair trading performance.
Chapter 3 constructs 8 valuation factors based on 23 firm characteristic descriptors, and proposes a consolidated fundamental squared distance metric that captures the contribution of firm fundamentals to company valuation. The consolidated squared distance metric also shows significantly positive contribution to the pair identification.
Chapter 4 proposes a pair trading return factor model to examine the contributions of the long list of distance measures in a multivariate setting. The model combines the different sources of information – price path, industry classification, value multiples, and valuation factors – to generate an aggregate prediction on the pair trading returns.
Liu, Yi, "Structural Identification of Pair Trades" (2023). CUNY Academic Works.