极值理论在期货风险管理中的应用

【作者】 刘正涛；

【导师】 盛朴；

【作者基本信息】 昆明理工大学 ， 企业管理， 2011， 硕士

【摘要】 本文以极值理论为基础,探讨期货的极端价格行为,并以CARR模型描述期货日内价格的趋势,且与McNeil和Frey(2000)所提出的条件极值理论比较。本文提出两阶段法结合极值理论与CARR模型,以S&P500指数期货及轻质原油期货为样本,通过数据实证指出,无论在样本内回溯测试,或是样本外预期损失率估计,都优于McNeil和Frey(2000)的条件极值理论。同时,本文采用大贩证券交易所(OSE)推出日经225股指期货与新加坡国际金融交易所(SIMEX)推出的日本的日经225指数期货为样本,来分析以极值理论为基础,结合CARR模型是否能避免期货价格涨跌幅限制的影响。实证结果发现,McNeil和Frey(2000)所采用的GARCH模型与极值理论常会估计出超过涨跌幅限制的价格变化,较为不合理。相对而言,极值理论结合CARR模型所估计的价格变化较为合理。此结果支持了极值理论结合CARR模型在期货保证金设定方面应用的优越性。最后,基于上述的结果,本文建议期货交易所在设定保证金时应该考虑期货日内变幅信息,并采用极值理论结合CARR模型以捕捉实时的价格信息,来制定动态保证金制度,以实时反应当下的风险变化。更多还原

【Abstract】 The article discusses future’s extreme price behavior and uses the range-based CARR model to estimate the intra-day’price’s heteroskedasticity. It also compared with the McNeil &Frey’s (2000) result which combines with the GARCH model and extreme theory. S&P 500Stock Index Futures traded on CME, and Sweet Crude Oil Futures traded on NYMEX are used. Both the in-sample back- testing and out-of-sample expect ed loss indicate that the CARR model and extreme value theory performs better than McNeil & Frey (2000) conditional extreme value theory.Furthermore, we use Osaka Stock Exchange (OSE) launched the Nikkei 225 stock index futures and the Singapore International Monetary Exchange (SIMEX) launched Japan’s Nikkei 225 index futures as a sample to analyze the extreme value theory, combined with CARR model is able to avoid the futures price impact of price limits. Empirical results, McNeil and Frey (2000) used GARCH model and estimate the extreme value theory often exceed the price limit of price changes, the more unreasonable. In contrast, extreme value theory combined CARR model is more reasonable estimate of price changes. The results support the CARR model with extreme value theory to set the application in the futures margin superiority.Finally, based on the above results, this article suggests the futures should be considered when setting bond futures days of amplitude information and the use of extreme value theory combined CARR model to capture real-time price information, to develop dynamic margin system, the immediate reaction in real time changes in risk.更多还原