In　consideration　of　the　deficiency　of　existing　research,　Mean-DCCA　and　Mean-MF-DCCA　models　are　proposed　by　combining　fractal　research　method　with　traditional　portfolio　theory　in　order　to　meet　the　actual　demand　of　investors　in　different　transaction　cycles.　The　factors　of　time　scales　and　different　fluctuation　ranges　are　taken　into　account　in　these　models.　The　models　are　used　to　conduct　portfolio　strategies　in　Shanghai　&　Hong　Kong　stock　markets　by　means　of　Shanghai-Hong　Kong　stock　connect　program,　after　which　the　effects　of　out-of-sample　are　tested　and　analyzed.　The　empirical　results　show　that,　first,　the　market　structure　of　Shanghai-Hong　Kong　presents　scales　effect,　long　memory　and　multifractal　characteristics;　second,　compared　with　the　traditional　strategies,　the　Mean-DCCA　portfolio　strategies　are　proved　to　achieve　better　effects;　finally,　the　Mean-MF-DCCA　portfolio　strategies,　by　choosing　an　appropriate　multifractal　q-order,　will　significantly　improve　the　single　fractal　portfolio　strategies,　enhance　investment　project　profitability　and　Sharp　Ratios,　and　create　additional　utility　for　investors　with　different　risk　preferences.　This　study　is　of　great　practical　significance　to　the　optimal　allocation　of　assets,　risk　measurement　and　management,　as　well　as　the　delineation　of　dependence　structure　of　Shanghai　&　Hong　Kong　stock　markets.　©　2018,　Editorial　Board　of　Journal　of　Systems　Engineering　Society　of　China.　All　right　reserved.