investment strategies of institutional investors, improves the service efficiency of funds.
First, analysis of the Chinese stock index futures market based on MF-DFA. Based
on the MF-DFA, this paper empirically studies the multifractal properties of the Chinese
stock index futures market. The author finds that the Chinese CSI300 returns exhibit
long-range correlations and multifractality by using a total of 2942 ten-minute closing
prices, making the single-scale index insufficient to describe the futures price
fluctuations. Further, the author shows the existence of two different sources of the
multifractality for the Chinese stock index futures market by comparing the original
time series with the transformed time series through the procedure of shuffling and
phase randomization. The results suggest that the multifractality is mainly due to long-
range correlations, although the fat-tailed probability distributions also contribute to
such multifractal behaviour.
Second, prediction of the CSI300 based on new EMD-RBF model. Only in the past
four years did China set up the stock index futures market displaying the non-stable and
non-linear signal features. The traditional estimation methods cannot make accurate
estimation of long-relevant sequence. Combining EMD with RBF, the author has
created a new method of estimation to predict the daily settlement price for stock index
futures. The result shows that this model has separated the original sequence with long-
relevance features into several short-relevance frequency bands, making up for the
shortage of system power information caused by the serious randomness of the original
sequence and the interruptions from nearby frequency bands. It is also compared with
other estimation models to display a relatively high degree of accuracy.
Third, test linear and nonlinear Granger causality CSI300 futures and spot markets
based on new concepts of nonlinear positive/negative spillover. Hiemstra and Jones
(1994) argued that a significant negative value of their nonlinear Granger causality test
(H-J test) means there is a confounding effect in the prediction. However, from the
theoretical analysis and Monte Carlo simulations, the author finds that H-J test is
significantly negative under the circumstance of negative volatility spillover.
Furthermore, the author puts forward the conceptions of positive/negative nonlinear
spillover, and apply H-J test to examine positive/negative nonlinear spillover effect. The
empirical study on China stock futures and spot markets shows that: 1) There is
significant positive nonlinear spillover from futures to spot market; 2) There is
significant negative nonlinear spillover from spot to futures market.
The author argues that there is “risk absorption” mechanism in information
spillover from the spot market to the futures market, which is due to the temporal
transfer of speculative trading from the analysis.
Key Word: MF – DFA, EMD, RBF neural network, H - J inspection