Stock Index Futures Arbitrage Strategy Based On SVR With Mix Frequency Data

Stock index futures are not only the important object for Unilateral speculation,but also the significant tool for hedging and arbitrage on financial secondary market.Stock index futures play an essential part in price discovery and smoothing trend to China stock market.However,in the past recent few years(especially in 2015),China stock market experienced a violent fluctuation which not only exists in stocks price,but also in the basis spread between stocks and index futures.According to the arbitrage theory,the basis of spots and futures should be in stable range due to the high relativity and spread convergence.The fierce vibration of basis brings the arbitrage investors both risk and opportunity.An accurate prediction of the basis is an important guarantee to the success of arbitrage strategy.In the past decades,big-data technology,statistical learning and deep learning skills have improved a lot,and more and more predicting models from machine learning are applied in quantitative investment.Support vector regression is an excellent forecasting method due to its structural risk minimization and kernel-trick.This paper collects mix frequency price information as input features,and uses SVR to predict both the basis and the fluctuation range.Adding the range to the basis we get an interval,which known as the no arbitrage interval.We can form a trade strategy based on this no arbitrage interval.This paper tests the strategy in the empirical part.