Font Size: a A A

Research Of Quantitative Strategies For Chinese Commodity Futures Market Based On A Dynamic Weighted Money Flow Model

Posted on:2019-06-11Degree:MasterType:Thesis
Country:ChinaCandidate:G H LuFull Text:PDF
GTID:2429330566993792Subject:Applied statistics
Abstract/Summary:PDF Full Text Request
This paper aims to analyze the transaction data of commodity futures and construct the money flow formula which can accurately describe the law of money flow in the futures market.The quantitative trading strategy of commodity futures based on the analysis of money flow market will be established in this paper.Firstly,due to the short mechanism of the commodity futures market,a dynamic weighted money flow model is proposed in this paper.The model proposed herein is based on the original money flow model but considers the impact of changes in both open interest and price on money flow.The proposed model aptly depicts the overall law of money flow in the Chinese commodity futures market.Then,this paper studies the impact of money flow on the future prices of commodity futures.The results regarding correlation between current money flows and future futures prices show that there are 17 futures contracts with strong negative correlations and three futures contracts with strong positive correlations between 2011 and 2013.Next,the logistic regression,Bayesian discriminant,decision tree,random forest and support vector machine models are applied to validate the forecasting ability of the proposed money flow model with respect to price fluctuations.The average prediction accuracies of the above models exceed 55%,indicating that the money flow model proposed in this paper has a strong forecasting ability.In addition,we find that among the mentioned model,the logistic regression model not only has high prediction accuracy but also can achieve stable prediction effect in multiple contracts.Finally,a binary classification logistic regression strategy based on the dynamic weighted money flow model is established for back-testing in the Auto-Trader platform and is compared to the double-moving average strategy and the buy-hold strategy.The back-testing results show that the cumulative annualized yield of the portfolio that uses the strategy proposed in this paper is 281.95%.Therefore,the proposed strategy is far superior to other strategies and exhibits better profitability.
Keywords/Search Tags:Money Flow, Futures high frequency data, Classification model, Logistic Regression Strategy Model
PDF Full Text Request
Related items