Research On Co-integration Arbitrage Strategy Based On Two Kinds Of Nonlinear Transformations And Its Application

The statistical arbitrage strategy based on cointegration analysis method has become a common and important method in modern econometrics,which is widely used in many economic and financial fields,among which linear cointegration has developed very maturely,there are a lot of research results,and there are many application areas in practical research.In contrast,non-linearity is more common in financial time series,and the research based on nonlinear cointegration and its statistical arbitrage strategy needs to be further strengthened.This paper intends to study in this aspect.This paper mainly studies the cointegration arbitrage strategy based on two kinds of non-linear transformation and its application in practice.Firstly,two kinds of nonlinear transformations,namely Logistic transformation and threshold transformation,are studied.Especially,when studying the threshold model transformation,the parameter estimation and model checking of the two-mechanism high-order model and multi-mechanism case of SETAR are elaborated in detail.Then,based on the above research results,linearize it by corresponding non-linear transformation,and then analyze it by co-integration,and establish corresponding strategies.In this process,in order to verify the feasibility of the method,this paper carries out corresponding stochastic simulation.By generating nonlinear time series,transforming them into linear series,co-integration analysis and formulating statistical arbitrage strategy to simulate transactions,it achieves considerable profits from the trading results.Finally,this method is applied to the five-minute stock price data of Bank of China and Agricultural Bank of China,and the five-minute stock price data of Sinopec and China Life.The experimental results show that the strategy achieves considerable profits.From this point of view,this method is effective and feasible.On the other hand,this paper also optimizes the threshold selection of paired trading strategy based on co-integration.When constructing the statistical arbitrage strategy based on co-integration pairing,most traders use subjective experience to artificially set the ope-ning and closing threshold of the trading strategy,and usually choose the standard deviation of several times the price difference as the threshold of their trading strategy.However,this method can not guarantee maximum profit.If the threshold is set improperly,it may lead to serious losses.To solve this problem,this paper uses intelligent optimization algorithms---particle swarm optimization,simulated annealing algorithm and genetic algorithm to optimize,search for the optimal threshold to maximize the profit of the trading strategy,and make an empirical study on it.