Pairing transaction co-integration relationship acceleration verification method based on E-G two-step method

Abstract

The invention discloses a pairing transaction co-integration relationship accelerated verification method based on an E-G two-step method. The method comprises the steps of obtaining two time sequences of a to-be-verified co-integration relationship; carrying out augmented Dickey-Fuller test on a residual error and a difference of the residual error, wherein a regression analysis result corresponding to the maximum lag order is solved by utilizing an LDLT decomposition method; obtaining regression analysis results corresponding to all lag orders according to the regression analysis result corresponding to the maximum lag order, and obtaining a corresponding error quadratic sum; utilizing an optimized akaike information criterion formula to calculate akaike information criterion function values corresponding to all the lag orders; selecting the lag order corresponding to the minimum akaike information criterion function value as the optimal lag order; and obtaining a regression coefficient corresponding to the optimal lag order. According to the invention, algorithm intensity reduction and approximate calculation are performed by applying a common least square method algorithm for many times in a traditional algorithm, and a verification speed of the co-integration relationship of the two time sequences is increased.

Description

technical field [0001] This application relates to the field of verification of cointegration relationship of paired transactions, and in particular to an accelerated verification method of cointegration relationship of paired transactions based on the E-G two-step method. Background technique [0002] Pair trading is a statistical arbitrage strategy widely used in financial markets. Based on two highly correlated stocks or other securities, their trends remain close for a long time. Once there is a divergence, it can be considered that the divergence is short-lived and can be Predict that their trends will return in the future, so as to make profits by judging their future trends. [0003] An important link in pair trading is how to find such two stocks with long-term correlation. The more classic and commonly used method is the E-G two-step method proposed by Engle and Granger. The cointegration relationship verification is used to determine the two stocks with long-term c...

AI Technical Summary

Problems solved by technology

[0004] However, due to the large amount of stock market data and the frequent changes in stock prices, and the E-G two-step method has the characteristics of high computational complexity and loops, when the length of the price sequence is very long or the number of stocks to be selected is large, the CPU Or running directly on the GPU will cause a very large system delay; and its parallel calculation will consume relatively large resources and generate large power consumption. At the same time, due to the matrix inversion and logarithmic operations in the algorithm, the hardware is Very unfriendly operation exists, it is very difficult to deploy it on FPGA or ASIC platform

Too long calculation time will reduce the pairing efficiency, so that profit opportunities are missed or even losses are caused by the end of the pairing relationship

Images