The Empirical Analysis for the Spread of Soya Oil and Soybean Meal Based on Wavelet Neural Network
International Journal of Economics and Finance
For the sake of a better cross-commodity arbitrage in the futures market, WNN (wavelet neural network) is adopted to analyze the previous spread and predict the future in this paper. Firstly, the correlation coefficient of previous prices between the two goods is calculated in order to examine whether there is arbitrage opportunity. Considered that the spread could be affected by many nonlinearity factors and BP neural network has slow convergence rat, BP neural network is combined with wavelet
... mbined with wavelet analysis which has excellent partial analysis ability.In this way, the prediction model about soya oil and soybean meal spreads is built based on WNN Compared the result calculated through that method with only BP neural network's: WNN is superior to neural network in predicting rapid fluctuation and secular trend. 86 Firstly, by analyzing the correlation between soya oil and soybean meal, arbitrage opportunity can be figured out. According to opening prices, high prices, low prices, closing prices, volumes and open interests,WNN and BP neural network are employed to predict the spread in the long term and short term. Then we compared the difference in precision between the two different methods. The result shows that WNN is superior to neural network in the process of long-term and short-term predictions. Although WNN is superior to neural network, it is necessary to improve accuracy of long-term prediction. because the number of factors that affects spread is more than that of individual commodity, and the spread is predicted merely according on opening price, high price low price and so on that, the information is not taken full useof.For sake of accurate prediction, more factors shoud be taken into consideration, such as import and export of related goods and interest rate. In additon, the prediction model should be connected with reality, so that the accuracy of it would be higher.