There　are　many　categories　in　the　financial　market,　and　futures　market　plays　a　very　important　role.　How　to　make　the　most　profit　in　futures　market　is　a　problem　that　investors,　fund　organizations　and　private　placement　teams　have　been　studying　all　along.　This　paper　constructs　a　futures　investment　decision　model　based　on　intensive　learning　model　theory,　makes　graphical　conversion　according　to　historical　time　series　data,　defines　the　return　of　strategies　for　local　strategy　banks,　trains　a　DQN　network,　automatically　switches　strategies,　and　ensures　the　maximization　of　gains.　©　2018　IEEE.