DualMomentumV1.0.c

// Dual Momentum V1.0 trading, implementation by Hredot ///////////////////
// Disclaimer: Use at your own risk!                    ///////////////////
// You alone are responsible for your gains or losses.  ///////////////////

// Let's have space for up to 100 assets, just in case
#define NUMBEROFASSETS  100

var assetWeight[NUMBEROFASSETS];// Will hold the index and performance stat of up to 1000 assets
int assetSorted[NUMBEROFASSETS];// Will hold the index of assetWeight sorted in decreasing order of performance
var myAssetType[NUMBEROFASSETS];// Stores information whether asset in assetWeight is stock =1 or bond =0
vars Price[NUMBEROFASSETS];     // Price history for each asset
int assetNum;             // Will contain the actual number of assets
int assetPlusNum;         // Will contain the actual number of assets with positive return

function bubbleSort(){    // sorts assetSorted to contain assetWeight indices in decreasing weight order
	int i,s,tmp;
	s=assetNum-1;
	while(s>0){
		for(i=0;i<s;i++){
			if(assetWeight[assetSorted[i]]*pow(100,myAssetType[assetSorted[i]]) <
			   assetWeight[assetSorted[i+1]]*pow(100,myAssetType[assetSorted[i+1]])){ 
				// stocks weigh "100 times" more while sorting, since they grow better than bonds
				tmp=assetSorted[i];
				assetSorted[i]=assetSorted[i+1];
				assetSorted[i+1]=tmp;
			}
		}
		s--;
	}
}

function run()
{
	set(LOGFILE+PRELOAD);
	BarPeriod = 1440;	// 1 day bars
	LookBack = 24*20;	// two years lookback period, just in case
	StartDate = 2010;
	Capital = slider(1, 5000, 0, 7000, "Capital", "How much money you start trading with.");
	int myLookBack = 12;   // 12 month lookback
	int myDaysUpdate = 15; // after how many days to update trades

	if(is(INITRUN)) {
		assetList("History\\AssetsZ9.csv"); // load asset list
	}

   // Implement absolute momentum:
   // portfolio loop to determine performance of each asset
	assetNum=0;assetPlusNum=0;
	while(loop(Assets))
	{
		asset(Loop1);
		assetHistory(Loop1, FROM_STOOQ);
		Price[assetNum]=series(price());
		myAssetType[assetNum] = 1-AssetVar[0];
		//Baltas and Kosowski (2012): linear regression is the best momentum indicator
		assetWeight[assetNum]=LinearRegSlope(Price[assetNum],myLookBack*20)/Moment(Price[assetNum],myLookBack*20,1);
		if(assetWeight[assetNum]>0){ assetPlusNum++; } // count up a wanted asset
		assetSorted[assetNum]=assetNum;
		assetNum++;
	}
	
   // Implement relative momentum:
   // sort AssetSorted to contain assetWeight indices in decreasing weight order	
	bubbleSort();
	
	static int dayspast=0;dayspast++;    // increment days count until trade adjustment
	int i,enterNum=min((assetNum)/2,assetPlusNum); //Participate in at most 1/2 of winning assets
	if(dayspast>myDaysUpdate){	          // update trades only once a time period myDaysUpdate
		for(i=enterNum;i<assetNum;i++){   // exit unwanted assets
			asset(Assets[assetSorted[i]]);
			exitLong();
		}
		var totalSlope=0;
		for(i=0;i<enterNum;i++){  // determine the overall weight of wanted assets
			totalSlope=totalSlope+assetWeight[assetSorted[i]];
		}
		for(i=0;i<enterNum;i++){  // enter wanted assets, each at the appropriate weight
			asset(Assets[assetSorted[i]]);
			Stop=3*ATR(24*20);
// Note: The following is optimized for 2x leveraging!
			Lots=(Capital+WinTotal-LossTotal)*Leverage/(4*enterNum*price()*LotAmount);
// drawdowns with this strategy are miniscule, therefore we reinvest all
// use Capital*sqrt(1 + (WinTotal-LossTotal)/Capital) instead of (Capital+WinTotal-LossTotal) if you feel unsafe
			if(NumOpenLong < 3 && (Equity - MarginVal)/(Capital+WinTotal-LossTotal) >0.33 ) enterLong();
		}
		dayspast=0; // reset the day counter after the trades
	}asset(Assets[0]);

   //Plot results:	
	PlotWidth = 600;
	PlotHeight1 = 300;	PlotHeight2 = 300;
	plot("$ Balance",Balance,NEW,BLUE);
	plot("Equity",Equity,0,GREEN);
	plot("BalanceInTrades",MarginVal,0,RED);

}