Font Size: a A A

High-frequency Trading Strategies Of SPIF Based On SVM Model

Posted on:2016-05-28Degree:MasterType:Thesis
Country:ChinaCandidate:H YuFull Text:PDF
GTID:2309330452965055Subject:Statistics
Abstract/Summary:PDF Full Text Request
With the rapid development of financial market, high-frequency trading of the SharePrice Index Futures (SPIF) has become more and more attention. More and more investorsbegin to explore a new profit model in high frequency trading of the SPIF. Speculativetrading is an expression of deep cognition to the trading market, which plays an importantrole in enhancing the market liquidity. This paper mainly studies high-frequency tradingstrategies of SPIF based on the view of the speculative trading.Considering the successful probability and maneuverability of the high-frequencytrading strategies, the paper selects the main stock index futures contracts as the researchobject. In order to guarantee the high accuracy, the paper confirms that the variation trendsof SPIF price is predicted by pattern classification and recognition and it’s decided thatusing suppor vector machine (SVM) mdel, which has the outstanding performance in thearea of classification and recognition, to classify and recognize the variation trends of SPIFprice. In the paper, it first uses median price momentum changes to classify training sets,and the methods, such as owe sampling based on clustering and so on, to balance trainingsets. And then through the characteristic index selection and parameter optimization, itforecasts the variation trends of price in the main contract. At last, a single multiclassification SVM model, a model consisting of multi classification SVM and binaryclassification SVM, and a dynamic SVM model are established. Applying prediction resultsof the three models, combined with the trading strategy, back analysis tests have been gone.Through the analysis of the test results, a well performanced trading strategy is got, whichincreases the winning percentage from56%to80%. This strategy plays a role of earlywarning effectively and, to a certain extent, reduces the risk of trading slippage. The tradingstrategy has been carried on the comprehensive reading, with illustrated shortcomingsand possible risks further explained of the strategy.
Keywords/Search Tags:SVM, classification and recognition, SPIF, high-frequency trading, speculative trading, prediction
PDF Full Text Request
Related items