| Black series futures are commodity futures based on steel industry chain products.Due to the fact that most of the steel industry chain products are black and related,they are called "black futures".Since the black series futures such as coke,coking coal and deformed steel bars were listed in Dalian Commodity Exchange,China’s futures market has formed a complete steel industry chain,which provides an effective means for China’s steel upstream and downstream enterprises to conduct price discovery and hedging in the market,and also provides more investment options for investors.Through the study of black commodity and black commodity futures,this paper finds that the upstream and downstream of the black commodity industry chain are closely connected,and black futures have large trading volume and good liquidity in the futures market.Iron ore,coke and rebar futures are selected as cross-variety arbitrage objects.The daily line data of main futures contracts from December 2019 to March 2022 are selected,and the co-integration test of the three futures shows that there is a long-term equilibrium relationship between iron ore,coke and steel rescrew futures,and the arbitrage ratio among the three futures is determined.The five-minute closing price data of futures from August 6,2020 to December 1,2022 were divided into training sets and test sets for neural network prediction and strategy demonstration.BP neural network,RNN neural network and LSTM neural network were built to predict the price difference of the black futures industry chain successively,and the predicted results were compared.Finally,according to the mean recovery characteristics of the price difference of the black chain,arbitrage strategies under various parameters were constructed for experiments,and the empirical results were compared.The results show that LSTM is superior to BP model and RNN model in terms of error size and fitting accuracy in predicting the spread trend.Among the futures crossvariety trading strategies,LSTM neural network model is superior to BP model and RNN model in terms of yield,trading win rate and control risk.The black series futures arbitrage strategy constructed based on LSTM neural network can achieve positive returns with less risks.Finally,this paper looks forward to the future research on strategy design and futures varieties,expands the research ideas of futures cross-variety arbitrage,and provides suggestions for investors aiming at different risk points to help investors better use the cross-variety arbitrage method,which makes this paper have certain theoretical and practical significance. |
