Precious Metal Futures Arbitrage Research Based On Dynamic VECM-GARCH Model

With the development of computer and big data technology,investors gradually turn their attention from investing based on fundamentals to quantitative investment,and quantitative trading has become the development trend of future trading.As a kind of quantitative investment technology,statistical arbitrage has a long history and has been widely used by institutional investors.Statistical arbitrage strategies are based on price volatility happens in the capital market.Investors expect to study from historical data to catch the occurrence of price volatility to achieve profits by "buy the weak and sell the strong".Compared with traditional trading strategies,statistical arbitrage weakens the analysis of fundamentals,instead,focuses on the characteristics of data itself,and emphasizes the concept of "statistics".It is always combined with the idea of mathematical modeling,and is an important part of quantitative investment.Different from the strict definition of risk-free arbitrage,statistical arbitrage has certain trading risks,but it also trades for more arbitrage opportunities.The implementation of statistical arbitrage depends on the shorting mechanism of the market.In the long run,the shorting mechanism of the domestic stock market is not yet complete,which brings many obstacles to the implementation of statistical arbitrage.However,with the increasingly mature margin financing and short selling,statistical arbitrage is expected to usher in a new round of recovery.In contrast,the futures market has a natural short mechanism,which is more suitable for the use of arbitrage strategies.The main contract trading heat high trading volume,it will be more reasonable as an investment target.There are many types of futures market contracts.This paper focuses on the precious metal futures market based on the following considerations: In today’s world,trade protectionism prevails and trade wars occur frequently.The traditional developed countries set off a wave of anti-globalization,which has formed more restraints and barriers to trade circulation.Combined with the turmoil in the Middle East,the warning of all kinds of danger signals will lead to a surge in global economic uncertainty,which will lead to a rise in risk aversion among investors.In this context,the precious metal market homeopathic investors become the safe haven,gold and silver become the choice of many investors.A number of scholars have studied the prices of gold and silver futures contracts,and found that there is a strong correlation between them,and there is also a significant cointegration relationship.On the premise of stable price difference,we can make use of its mean value recovery property for arbitrage operation.However,through the analysis of price data in the past two years,this paper found that the cointegration relationship between gold and silver contracts became less and less significant.Especially since the end of 2017,the spread between gold and silver contracts showed an obvious trend of rising,showing a non-stable state,so it was difficult to make use of its mean recovery feature for arbitrage.Under this realistic background,this paper tries to change the thinking,taking the price of a single futures contract as the research object,looking for arbitrage opportunities.In this paper,it is found that many scholars build vector error correction model(VECM)based on the cointegration relationship between high and low prices of the same trade,and carry out arbitrage trade based on this design signal.Through to the number of contracts to establish a regression model of residual error diagnosis,the results show that the prediction model based on single VECM model building exist obvious heteroscedasticity characteristic,therefore this paper consider combine the VECM model with the generalized autoregressive conditional heteroscedasticity(GARCH)model to improve the forecasting model.By comparing the prediction results,we prove that prediction accuracy has been significantly increased,fully affirmed the necessity of considering heteroscedastic property exists in the original sequence.Furthermore,uses the fixed window wide and rolling prediction to update then model’s parameters to monitoring to the high and low prices timely.Through this way,the prediction accuracy has been enhanced largely.The results show that for high sequence,the advantage is more obvious.Then this paper use the point prediction and time-varying variance to construct the track trading strategy,which was applied to the precious metal market of Shanghai futures exchange.The empirical results show that the VECM model combined with the GARCH model can better fit the price trend of gold and silver futures,and significantly improve the prediction accuracy compared with the single VECM model.In addition,compared with the trading bands generated by the traditional technical indicator MA(5)and the random walk model(RW),the VECMGARCH model can better fit the trend of out-of-sample prices and capture more arbitrage opportunities.From the back test results of the strategy,the average trading odds of gold and silver futures reached 90% and 88% respectively,indicating that the strategy has a strong rationality for the signals sent by price changes.Through the horizontal comparison,in terms of the annualized return,the strategy on the gold futures contract performance better than the silver futures.To be specific,for gold futures contract,the average annual return is about 240%,for silver futures contracts,the average annual return of 120%,the former is about twice that of the latter.Combined with maximum retracement perspective,when the maximum retracement ratio is equal,the yield of gold futures contract is significantly higher than the yield of silver futures.In all,the strategy has certain stability,and performs better in the gold futures contract.The innovation of this paper are as follows: firstly,break through the limitations of the traditional matching trading,instead,focus on a single futures contracts as the research object,establish a VECM model based on the cointegration relationship between the high and low price,then considering the heteroscedasticity exists in the raw sequence,combines VECM model and GARCH model to constract a more accurate prediction model.Secondly,combining with the thought of confidence interval to formulate the corresponding strategy.In the aspect of the signal’s design,the paper is no longer use a fixed standard deviation,instead,use the time varying variance obtained from the GARCH model,which effectively enhances the ability of strategy to capture the price changes.The back test results show that the strategy applied in gold and silver futures contracts can produce a stable return on investment,which confirmed the effectiveness of the proposed strategy.