Font Size: a A A

Research On Futures Statistical Arbitrage Based On High-Frequency Data

Posted on:2010-06-27Degree:MasterType:Thesis
Country:ChinaCandidate:R Q KangFull Text:PDF
GTID:2189360302466510Subject:Statistics
Abstract/Summary:PDF Full Text Request
In the rapidly changing financial market, the pursuit of steady excess return is all investor's goal forever. In 2008, global financial crisis swept through every corner of global financial markets. On one hand, Lehman Brothers and other large institutional investors have gone bankrupt. On the other hand, a remarkable performance has performed by institutional investors based on arbitrage trades as represented by High Bridge Capital in context of global financial crisis. This reflects the advantages of strategy based on arbitrage trades. This kind of strategy can create market-neural portfolios and avoid systematic market risk.Statistical arbitrage trading strategy is widely used by large hedge funds and investment banks in Europe and American. But because of the lack of short mechanism in main financial markets in our country, scholars of mainland China rarely processed related research. In the context of establishing of short mechanism such as financing capital and financing securities in our country in near future, this kind of research has an important forward-looking significance. The paper used statistical arbitrage in our country's futures market and tested the feasibility of statistical arbitrage in our country's futures market. The main framework of this paper is described as follows:(1) The futures Cu in Shanghai Futures Exchange is selected as our research object which has the longest trading history and is traded actively. In addition to the maturity date, two adjacent futures are identical in other fundamental and technological sides which are suitable for statistical arbitrage trading. The paper uses Cull and Cul2 futures price series to conduct empirical analysis. In order to identify as many potential statistical arbitrage opportunities as possible, we adopt the high-frequency data constituted by five-minute closing prices of Cull and Cul2 as the study sample.(2) Use cointegration to test long term equilibrium relation of Cull and Cul2 futures, further to establish the coefficient of pairs trading. Test result shows that the long term equilibrium relation between them is existed. After suited object of pairs trading have been decided, we use the sample data to construct the optimized threshold as the trigger point for the arbitrage which can provide us maximal expected arbitrage profit. In order to control arbitrage risk, we use VaR theory to decide the up and down stop-loss boundary which can provide us a stable arbitrage strategy. After the optimized arbitrage strategy has been constructed, we us the data in sample and out of sample period to simulate transaction with the optimized arbitrage strategy we have constructed to test whether it is feasible or not.(3) In addition, the paper uses two different methods for data out of sample period. One is based on constant historical volatility; the other is based on time-variety volatility which is produced by using GARCH model. The empirical result shows that the strategy based on time-variety volatility is better than the strategy based on constant historical volatility. The statistical arbitrage strategy constructed in the paper can earn a profit which is above 1.88% per day on average whether in sample data or out of sample data is used. This conclusion indicates that statistical arbitrage strategy is feasible in our country's futures market.
Keywords/Search Tags:high-frequency data, statistical arbitrage, cointegration, futures market
PDF Full Text Request
Related items