科研之友
微信
新浪微博
Facebook
分享链接
摘要
For high-frequency statistical arbitrage, setting the proper trading threshold for each trading period is extremely important. We find that the optimal short-run trading threshold series in the Chinese metal futures market demonstrates chaotic characteristics. Therefore, we propose a new statistical arbitrage model in which a TGARCH model is applied to capture short-term price cointegration and asymmetric price spread standard deviations between futures contracts and a wavelet neural network is utilized to predict trading thresholds. Backtesting results demonstrate that the new model provides more stable and accurate trading thresholds and therefore generates more profits compared with the backpropagation neural network and historically optimal models. Potential topics for further study, including adjustments of the model to fit different market situations and the impact of transaction costs, are discussed.
- 出版日期2015-2-1
- 单位武汉大学; rutgers
