Font Size: a A A

State Transition-based Time-varying Combination Of The Stock Index Futures Volatility Prediction

Posted on:2012-09-27Degree:MasterType:Thesis
Country:ChinaCandidate:H L WangFull Text:PDF
GTID:2189330335951756Subject:Quantitative Economics
Abstract/Summary:PDF Full Text Request
The official Shanghai and Shenzhen 300 stock index futures listed on thedevelopment of China's capital market is an important milestone. Its launch furtherimproves China's capital market structure, and provides a comprehensive hedging tool,but also greatly increased the liquidity of capital markets and effectiveness. However,because the Shenzhen 300 stock index futures trading on margin, daily is at city andforcibly flat (minus) warehouse system, the futures price fluctuations can causeinvestors doubling of profits or losses, and intense and frequent amplified thefluctuation will also affect the healthy development of the national economy, and eventrigger a recession. So the stock index futures returns to the characteristics andfluctuation characterized by fluctuating behavior, simulation and forecast is particularlyimportant.At present, the stock market fluctuation for characterization and describe the singlemodel mainly has three kinds, namely exponential smoothing class model. GARCHclass model and SV class model. However, each of a single wave model to describe thestock market fluctuation has certain limitation. For example, exponential smoothingclass model hypothesizes yields fluctuations for constant fluctuations, according to theperiodicity use yield sequence of historical volatility for future point the unbiasedestimation return volatility, however, the reality of the efficiency for the fluctuationsequence, when earnings sequence of time-varying fluctuations in each pointdistribution is not the same, namely the distribution of standard deviation changes withtime. GARCH model conditional variances of the residual defined as the square of theconditional variances and antecedent of conditional variances deterministic function,with the past observations estimates, although solved directly related to thetime-varying fluctuate, but when yields sequence, abnormal observation will makeconditions variance estimation sudden changes. SV class model conditional variance isno longer a deterministic function, but add random item to reflect the Influence ofrandom factors on volatility. Apply these three models in to volatility, simulation and forecast to describe the process and no kind of the prediction effect is absolutelysuperior to other models.Due to the modeling mechanism, assumptions and the sources of information isdifferent, any single wave model can only contain and reflect the local information,adopts wave combined forecasting could more reasonable description and depict thecharacteristics of fluctuation. In fluctuation forecast fields, usually adopts combinationforecast model prediction of fluctuation, this model through the mix multiple singlemodel, regroup single model contains information to enhance the forecast accuracy. Atpresent, according to the combination of single forecasting model in a different way,combination forecast can be divided into variable weight combination forecast andconstant weight combination forecast. Single predictive results, with economic change,it is always a good show "ring" sex, and variable weight combination forecast modelcan be on different economic period, according to the single model goodness-of-fitchanges, realize the weight coefficient change single model, compared with constantweight combined forecasting model is more scientific . However, the existingtime-varying combination forecast model in determining weights not single modelabout the conditions satisfying practical is proposed in this paper, based on state transferof time-varying combination forecast method, which is applied to the Shenzhen 300stock index futures fluctuations forecastThis paper firstly stock index futures and fluctuations in the basic concept and thespecific characteristics of futures fluctuations comprehensive review of a single wave,and the advantages and disadvantages of prediction model comparison analysis;Secondly, in domestic and abroad the latest stock index futures fluctuation combinationforecast method carries on the summary and the evaluation of the foundation, on thetraditional combination forecast model and time-varying combination forecast modeladvantages and disadvantages of comparison and analysis, found time-varyingcombination forecast model can accurately describe more characteristics of fluctuation.Finally, the paper proposes a method based on state transfer of time-varyingcombination forecast method, and the model is applied to Shenzhen 300 of stock indexfutures in forecast, obtained better fitting and prediction effect.
Keywords/Search Tags:Markov-Regime-Switching, Stock index futures, forecast combining
PDF Full Text Request
Related items