Abstract
Pairs trading is a well-known statistical arbitrage strategy used in the financial market to detect pricing anomalies among financial assets using statistical techniques. This research aims to develop a pairs trading strategy in two steps (formation and trading) and reproduce these steps several times from 27-09-2018 to 30-12-2021.
In the first step, we gather multiple cryptocurrencies' historical minute-by-minute closed prices and discover nonlinear cointegrated pairs using the Kapetanios-Snell-Shin (KSS) test. Among them, the nominated pair is selected with the highest Kendall's Tau correlation.
Then, an appropriate bivariate copula family is specified to demonstrate the dependence structure of two assets, and the corresponding parameters are obtained by canonical maximum likelihood estimation. In the second step, the information from the dependence structure is used to determine trading signals for the crypto coins that are overvalued or undervalued compared to their pairs.
The strategy is backtested by generating trading signals and executing the simulated trades.