Study On Stock Index Futures Margin Levels Setting Based On Extreme Value Theory

Margin level setting is the most important factor in the stock index futures market system development. The article does empirical research on the margin level setting of Shanghai and Shenzhen 300 index futures. We choose 1279 full samples'daily returns from 5 January 2005 to 15 April 2010. We obtain the conclusion that index data can instead of index futures. Then, 1279 full samples'daily returns are chosen to describe statistical analysis and test of normality, we obtain that it doesn't obey normal distribution; furthermore, it has a "peak and fat-tail". Secondly, we use Hill estimation method and VaR-x estimation method which both base on extreme value theory to solve the margin level, and then we solve margin level's coverage under different default rates to compare with two methods. The conclusion is as follows:The margin level can both cover the price fluctuations by using Hill and VaR-x estimation methods when the default rate is 1%. On one hand, Hill estimation method's applicable default rate area is [0.01,0.022], the corresponding margin level is 5.835%, it has a highly cover rate and good stability. However, it needs large samples to support it. It is suit for a lower default rate investment environment. On the other hand, VaR-x estimation method's applicable default rate area is [0.01,0.03], the corresponding margin level is 5.334%. It has good calculate general, but its stability is not so good. The article combines the advantages of the two extreme value methods, and gives suggestions in setting margin level. Furthermore, we do the empirical analysis among EWMA; GARCH and G-H methods which are not belong to extreme value methods, and find that extreme value theory method is more stability and effective. It could provide useful guidance to the domestic stock index futures market which is not mature yet.