Analysis On Characteristics Of Corners In Commodity Futures Based On K-Means Clustering Algorithm

With the improvement of the futures regulatory system and the continuous strengthening of the exchange supervision,the "hard corner"has been withdrawn from the historical stage.However,the "soft corner",that is,some social groups control the quantity of the available spot by their advantage in the spot market.When the contract is close to delivery,the way to maintain the position is to enter the delivery month,forcing the bull to liquidate because it cannot be delivered in full.Therefore,the author selects some characteristic variables of futures contract,and through the K-Means clustering algorithm,we observe the clustering characteristics of the characteristic values in different state of the corner,and obtain the variables that have the ability to explain the action of the forced action as the target of identifying the corner and perfect the current means of futures exchange.In this paper,based on the definition of forced warehouse in the academic circle,this paper uses Pirrong(2004)to determine several futures contracts in 1989 Ferruzzi soybeans,and defines several futures contracts with high or high risk in China.There are three kinds of small probability,and then the sample data is analyzed by the three dimensions of the K-Means clustering algorithm,which are the deviation degree of the immediate contract and the next contract price,the deviation of the spot price from the spot price in the non cut spot and the proportion of the excess warehouse.The clustering results show that the three dimensions can accurately identify the depressed warehouse behavior,in which the recognition accuracy of the contract for methanol and screw steel is up to 80-90%,and the identification ability of the contract for the natural rubber contract is poor.The reason may be that the import dependence of the natural rubber is high and the domestic depressing people can't control it.The spot market makes the index based on the spot dimension invalid.On the basis of the above conclusions,this paper puts forward the policy suggestions for the prevention of storehouse behavior,including the continuous improvement of the limited warehouse system,the large household reporting system without losing liquidity,the innovation of the delivery system to reduce the cost of delivery and the increase of the setting up of the delivery bank.