Statistical Arbitrage Research Based On Kalman Filter

Statistical arbitrage strategy is one of the common strategy of overseas hedge funds. For a long time, due to the short mechanism is not perfect in china, there is always a huge statistical arbitrage opportunities waiting for mining in A-shares. With the maturity of domestic securities lending and refinancing mechanisms.investors are gradually involved in this field, I believe that research in this area will lead to further development of national statistical arbitrage.Classical statisitical arbitrage is based on the use of cointegration to select stock pairs and to get the spread seris. Followed by the establishment and implementation of a seris of open/close indicator to finish statistical arbitrage.This article is based on the basic cointegration, firstly establish a statistical arbitrage model by cointegration theory, and then through a series of parameter optimization process we get the most excellent learning and longest holding period together with the open point. The parameter sensitivity test and the outside samples test proved that the parameters of our investment model should be appropriate and achieved a better return on investment.We create a model of varying parameter estimation model in the next part of the paper using dynamic Kalman filter. Dynamic cointegration between the two price time series can be portrayed using the state-space model, and the Kalman filter is an important tool to solve the state space model. After obtaining a series of initial value is estimated that we can get a series of dynamic and time-varying cointegration estimation, we use this cointegration to get a better return on investment under the above extension model.Overseas research on statistical arbitrage has covered classical physics, statistical theory. In this paper, Kalman filtering here belong to the classical theory of cybernetics, the authors hope to be able to initiate, to allow more people to join scholars and practitioners to research in this field among domestic and also be able to promote the theory and practice of statistical arbitrage make a contribution to development.