The Research Of Discount Or Premium Of Structured Fund And Arbitrage Strategy

As a kind of leveraged funds, structured funds are not originated from China. Since the first one established in 2007, structured funds have a good development in China, no matter from the scale or improvements of products. However, the research on structured funds mostly focuses on pricing and the reason why discount or premium happens. For arbitrage strategies based on discount or premium of structured funds, the academic research is relatively less.From the second half of 2014 to 2015, special market conditions draw more attention from investors to the discount or premium of structured funds, along with the corresponding arbitrage strategies, but most of the studies stay in practical usage level. Securities companies as sellers, did some research as well, yet not deep. This paper attempts to make an in-depth research on improving the arbitrage strategy.Firstly, an introductory is carried out on product design, production raising and operation, discount or premium phenomenon, and paired conversion mechanism which is really important to the arbitrage of structured funds. We make an analysis on how the arbitrage is operated and then introduce some arbitrage strategies based on discount or premium of structured funds.Based on the result of analysis on the discount or premium arbitrage process of structured funds, we try to select the indexes which may have an impact on risk and return of the arbitrage, including the two-day underlying index return, index tracking error, the estimation error of net asset value, the tracking index volatility, turnover of structured B, bull and bear index, positions and discount or premium rate.When predicting arbitrage opportunity in real cases, we find that the linear relationship between the indexes we select and the predict outcome is not significant. So we switch to nonlinear neural network model to carry out the arbitrage predict. To verify the effectiveness of the chosen indexes, we use two future data in the calculation of the two-day underlying index return, index tracking error, the estimation error of net asset value, discount or premium rate and position, tests show good result which prove the effectiveness of the chosen indicators. In order to achieve real forecasting, we replace the future data with existing data, and find the success rate of arbitrage eventually enhanced. Confusion matrix shows the type two error reduce at a large extent, which means the overall arbitrage retracement is reduced. After optimization, the result obtained by discount and premium arbitrage strategy seems to be better, which is because in relatively short period of time, the more discount and premium arbitrage opportunities make model better trained, thus more accurate the predict outcome can be.