All posts by George Pruitt

George Pruitt - Author - Blogger - Programmer - Technician Bachelor of Science in Computer Science from UNC-Asheville Levo Oculos Meos In Montes

The Perfect System

WARNING CHEATING AHEAD ūüėČ There is no Holy Grail!

Having the results of a Perfect Profit System for any given market can be helpful when trying to determine the quality and viability¬† of one’s own algorithm (ala Bob PardoThe Evaluation and Optimization of Trading Strategies).¬† The level of correlation between a user defined algorithm and the PPS can indicate the quality of an algorithm – if its not been overly curve-fit.

Michael Bryant of AdapTrade.com has written an excellent article on how to use the correlation between a user defined algo and the PPS to optimize the algo to increase the correlation co-efficient between the two.  The premise being the closer you get to the PPS with your own strategy the better off you are.

Click here for Michael Bryant’s excellent article.

Bob Pardo’s perfect system simply accumulated the absolute differences between consecutive closes to create the PPS equity stream.¬† In Michael Bryant’s article he utilizes price swings based of either price points or fractions of ATR (average true range.)¬† Bryant’s version takes advantage of 20/20 hindsight and creates a more realistic trade count.¬† Bob’s version would trade every day.¬† Michael’s version trades much, much, less.

My version, with the benefit of hindsight, picks the highest close and lowest close of every month and uses the data to sellShort and buy at these levels respectively.  I wanted to actually show the trades on the charts as they maybe helpful as well.  So to do this you have to run two strategies : 1) one that picks and outputs the trades and 2) one that takes the output from one and executes the trades.

There were two ways to accomplish this : 1) calculate the monthly turning points and output the data to  a file and then insert the data as 3rd party index into another chart of the same market and program a strategy that interprets the index or 2) calculate the monthly turning points and output the information as pure EasyLanguage code snippet and paste that code into another strategy.  I chose the second method as it also helps show the reader how to do some really cool stuff.

Before I started to program this research tool I asked myself “How can you pick the monthly peaks and valleys as you run through the data?”¬† Well I knew I had to be at the end of the month and then look back enough bars to choose these points.¬† Since the number of trading days in a month are variable I knew I had to use a while loop.¬† While loops are really cool and easy to use, but you must be careful or you will land in Steve Jobs 1 infinite loop.¬† TradeStation is very patient and will let you know before crashing and burning though.¬† Here is the snippet of code that loops back in time to gather the monthly peaks and valleys:

If month(d) <> month(d[1]) then // are we in a new month?
Begin
	Value1 = 1;
	hiMonthPrice = 0;
	loMonthPrice = 999999;
	testMonth = month(date); 	
	While month(d[value1]) = month(d[value1+1]) //loop until the prior month
	Begin
 		If c[value1] > hiMonthPrice then
 		Begin
 			hiMonthPrice = c[value1]; //store the peak price
 			hiMonthBar = value1; // store the peak bar number
 		end;
 		If c[value1] < loMonthPrice then
 		Begin
 			loMonthPrice = c[value1]; //store the valley price
 			loMonthBar = value1; // store the valley bar number
 		end;
 		Value1 = value1 + 1;
	end;
//	print(d," monthHi ",hiMonthPrice," ",hiMonthBar);
//	print(d," monthLo ",loMonthPrice," ",loMonthBar);
Looping Backward - Collecting Peaks 'n Valleys

Okay we now have the prior month’s peak and valley values.¬† So we need to store this information into a list.¬† EasyLanguage allow for user defined arrays, which are really just indexable¬† (I think that is a word) lists.¬† Here is a list of arrays that we will be using and how they are declared:


 Array: equPerf[5000](0),  { equity at each bar, perfect trades }
        trIdeal[2000](0),      { equity at end of perfect trades }
        datesPerf[2000](0),   { dates of perfect trades }
        BorSPerf[2000](0),  { buys or sells of perfect trades}
        pricePerf[2000](0); { prices of perfect trades}
I over dimensioned these Arrays - 2 per month right?

To declare an array you use the Keyword array and then give the array a name and then use square brackets to tell the computer the maximum number of elements in the array then instantiate each array element to a value that is surrounded by parenthesis.  Here we assigned each value in the arrays to zero.

Now that we have our lists and we are running through the chart and stopping at the end of each month and collecting peaks and valleys how do we turn this into an EasyLanguage strategy?¬† I had to think about this which I usually don’t do I just start programming.¬† I did just start programming but I didn’t get the results I wanted so then I had to think about it.¬† I knew this strategy needed to be in the market 100% of the time so it would be a stop and reverse system.¬† The first trade would be important so I needed to know which occurred first in the month – the peak or the valley.¬† If it was a peak then the first trade would be a sell and the second would be a buy, since we are simply selling peaks and buying valleys.¬† I then extrapolated this idea out to each month.¬† If I exited a month in a long position than I would want the following month’s valley to be first.¬† In a perfect would it would be peak, valley, peak, valley and so on.¬† That doesn’t always happen.¬† Sometimes you buy a valley late in a month and the market starts off the next month in a decline.¬† So you have to wait for the monthly peak to sell short and this could lead to a sizable loss – remember we are looking for an Almost Perfect System.¬† I could have programmed it so that the¬† buying late in the month and subsequent market decline scenario didn’t occur, but it would require eliminating some trades.¬†Since there’s not that many trades to start with I decided to go a different route.¬† So what I did was program an exception when there was a flux in the peak-valley continuum.¬† If I was exiting the month in a long position and a valley preceded a peak in the following month I simply reversed my position on the first day of the month.¬† Easy fix right?¬† At the beginning of the month I wanted all my trades set up from the preceding¬† month – I did this to just make things simpler.¬† Here is the snippet of code that puts the data into the arrays we will use later to create the EasyLanguage code:

If tempBors = 1 then
	Begin
		If hiMonthBar > loMonthBar then
		Begin  // long and peak occurs first
			BorSPerf[trdCnt]  = -1;
			datesPerf[trdCnt] = d[hiMonthBar];
			pricePerf[trdCnt] = c[hiMonthBar];
			print(datesPerf[trdCnt]," Sell ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;
			BorSPerf[trdCnt]  = 1;
			datesPerf[trdCnt] = d[loMonthBar];
			pricePerf[trdCnt] = c[loMonthBar];	
			print(datesPerf[trdCnt]," Buy ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;
		end
		else
		begin // long and valley occurs first
			BorSPerf[trdCnt]  = -1;
			datesPerf[trdCnt] = d[value1];
			pricePerf[trdCnt] = c[value1];
			print(datesPerf[trdCnt]," Sell Conflict ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;
			BorSPerf[trdCnt]  = 1;
			datesPerf[trdCnt] = d[loMonthBar];
			pricePerf[trdCnt] = c[loMonthBar];
			print(datesPerf[trdCnt]," Buy ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;	
			BorSPerf[trdCnt]  = -1;
			datesPerf[trdCnt] = d[hiMonthBar];
			pricePerf[trdCnt] = c[hiMonthBar];
			print(datesPerf[trdCnt]," Sell ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;	
		end;
	end;
Decision Process When Coming Into Month Long

Once you go through the entire chart you should have all of you trade arrays set up.  So on the last bar of the chart you need to print this data out to a file.  But you want to print it out so you can simply copy and paste the output into another strategy Рin other words you want to be syntactically correct.

Use the keyword LastBarOnChart to predicate the output process.¬† You can only print out strings to a file in EasyLanguage so all numbers have to be converted to strings – use NumToStr to accomplish this feat.¬† I think I have discussed formatted output somewhere in a post, but here is a quick reminder.¬† Let’s say you want to convert 1234.5678 to the string 1234.56, you would use NumToStr(value1 , 2).¬† The second argument in the function call informs the computer to truncate the decimal to two places. If you want to concatenate multiple strings together (i.e. connect strings together) you simply use the + sign.¬† Look at this line of code:

StrOut = “DateArr[“+NumtoStr(value1,0)+”] = ” + numToStr(19000000+datesPerf[value1], 0) + “;TradeArr[“+numtoStr(value1,0)+”] = ” +numToStr(BorSPerf[value1],0) + “;PriceArr[“+NumtoStr(value1,0)+”] = ” + numToStr(pricePerf[value1],2) + “;”+ Newline;

This is what it creates:

DateArr[1] = 20001211;TradeArr[1] = -1;PriceArr[1] = 1368.50;

This is a neat line of code where I mix hard coded word strings with converted numerical values to strings to create syntactically correct EasyLanguage in one complete, albeit long, string.

Now all you need now is an interpreter strategy.  The output from the first pass program creates the arrays needed for the interpreter program and then outputs all of the trades as pre-filled arrays.  Here is the output program in its entirety:

 input:  FName       ("c:\PerfectSystem\output.txt");     { file name to write results to }
 
 Var:  HighBar(0), LowBar(0),ibar(0), StrOut(""), 
       BorS  (0),daysInEquity(0),equCnt(0),dataIndex(0),tempMP(0),trdCnt(0),trdNum(0),
       oteEqu(0),clsTrdEqu(0), hiMonthPrice(0),hiMonthBar(0),
       loMonthPrice(0),loMonthBar(0),testMonth(0),firstTime(true);   
 
 Array: equPerf[5000](0),trPerf[2000](0),datesPerf[2000](0),   
        BorSPerf[2000](0),pricePerf[2000](0);
 
If month(d) <> month(d[1]) then // are we in a new month
Begin
	Value1 = 1;
	hiMonthPrice = 0;
	loMonthPrice = 999999;
	testMonth = month(date); 	
	While month(d[value1]) = month(d[value1+1]) //loop until the prior month
	Begin
 		If c[value1] > hiMonthPrice then
 		Begin
 			hiMonthPrice = c[value1]; //store the peak price
 			hiMonthBar = value1; // store the peak bar number
 		end;
 		If c[value1] < loMonthPrice then
 		Begin
 			loMonthPrice = c[value1]; //store the valley price
 			loMonthBar = value1; // store the valley bar number
 		end;
 		Value1 = value1 + 1;
	end;
//	print(d," monthHi ",hiMonthPrice," ",hiMonthBar);
//	print(d," monthLo ",loMonthPrice," ",loMonthBar);
	
	vars: tempBorS(0);
	If firstTime then
	begin
	    tempBorS = -1;
		If hiMonthBar > loMonthBar then tempBorS = 1;
		trdCnt = 1;
		firstTime = false;
	end
	else
	begin
		tempBorS = BorSPerf[trdCnt-1];
	end;
	
	If tempBors = 1 then
	Begin
		If hiMonthBar > loMonthBar then
		Begin  // long and peak occurs first
			BorSPerf[trdCnt]  = -1;
			datesPerf[trdCnt] = d[hiMonthBar];
			pricePerf[trdCnt] = c[hiMonthBar];
			print(datesPerf[trdCnt]," Sell ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;
			BorSPerf[trdCnt]  = 1;
			datesPerf[trdCnt] = d[loMonthBar];
			pricePerf[trdCnt] = c[loMonthBar];	
			print(datesPerf[trdCnt]," Buy ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;
		end
		else
		begin // long and valley occurs first
			BorSPerf[trdCnt]  = -1;
			datesPerf[trdCnt] = d[value1];
			pricePerf[trdCnt] = c[value1];
			print(datesPerf[trdCnt]," Sell Conflict ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;
			BorSPerf[trdCnt]  = 1;
			datesPerf[trdCnt] = d[loMonthBar];
			pricePerf[trdCnt] = c[loMonthBar];
			print(datesPerf[trdCnt]," Buy ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;	
			BorSPerf[trdCnt]  = -1;
			datesPerf[trdCnt] = d[hiMonthBar];
			pricePerf[trdCnt] = c[hiMonthBar];
			print(datesPerf[trdCnt]," Sell ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;	
		end;
	end;
	If tempBors = -1 then
	Begin
		If loMonthBar > hiMonthBar then
		Begin
			BorSPerf[trdCnt]  = 1;
			datesPerf[trdCnt] = d[loMonthBar];
			pricePerf[trdCnt] = c[loMonthBar];
			print(datesPerf[trdCnt]," Buy ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;
			BorSPerf[trdCnt]  = -1;
			datesPerf[trdCnt] = d[hiMonthBar];
			pricePerf[trdCnt] = c[hiMonthBar];
			print(datesPerf[trdCnt]," Sell ",pricePerf[trdCnt]);
			trdCnt = trdCnt+ 1;	
		end
		else
		begin
			BorSPerf[trdCnt]  = 1;
			datesPerf[trdCnt] = d[value1];
			pricePerf[trdCnt] = c[value1];
			print(datesPerf[trdCnt]," Buy Conflict ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;
			BorSPerf[trdCnt]  = -1;
			datesPerf[trdCnt] = d[hiMonthBar];
			pricePerf[trdCnt] = c[hiMonthBar];
			print(datesPerf[trdCnt]," Sell ",pricePerf[trdCnt]);
			trdCnt = trdCnt + 1;
			BorSPerf[trdCnt]  = 1;
			datesPerf[trdCnt] = d[loMonthBar];
			pricePerf[trdCnt] = c[loMonthBar];
			print(datesPerf[trdCnt]," Buy ",pricePerf[trdCnt]);	
			trdCnt = trdCnt + 1;		
		end;
	end;
	
end;
			
tempMP = 0;
vars:cntDown(0);
If lastBarOnChart then
Begin	
	FileAppend(FName, NewLine + "arrays: DateArr[5000](0),TradeArr[5000](0),PriceArr[5000](0);" + NewLine);
	For value1 = 1 to trdCnt
	Begin
   		StrOut = "DateArr["+NumtoStr(value1,0)+"] = " + numToStr(19000000+datesPerf[value1], 0) + ";TradeArr["+numtoStr(value1,0)+"] = " +numToStr(BorSPerf[value1],0) + ";PriceArr["+NumtoStr(value1,0)+"] = " + numToStr(pricePerf[value1],2) + ";"+ Newline;
   		FileAppend(FName, StrOut);
	end;
end;
Almost Perfect System EasyLanguage Creator Code

 

Here’s the output that was created when this strategy as was added to a 19 year long ES daily bar chart:

arrays: DateArr[5000](0),TradeArr[5000](0),PriceArr[5000](0);
DateArr[1] = 20001211;TradeArr[1] = -1;PriceArr[1] = 1368.50;
DateArr[2] = 20001220;TradeArr[2] = 1;PriceArr[2] = 1245.75;
DateArr[3] = 20010102;TradeArr[3] = -1;PriceArr[3] = 1265.50;
DateArr[4] = 20010105;TradeArr[4] = 1;PriceArr[4] = 1269.50;
DateArr[5] = 20010131;TradeArr[5] = -1;PriceArr[5] = 1346.50;
DateArr[6] = 20010201;TradeArr[6] = 1;PriceArr[6] = 1348.00;
DateArr[7] = 20010205;TradeArr[7] = -1;PriceArr[7] = 1328.25;
DateArr[8] = 20010228;TradeArr[8] = 1;PriceArr[8] = 1203.75;
DateArr[9] = 20010307;TradeArr[9] = -1;PriceArr[9] = 1231.25;
DateArr[10] = 20010322;TradeArr[10] = 1;PriceArr[10] = 1070.00;
DateArr[11] = 20010402;TradeArr[11] = -1;PriceArr[11] = 1100.75;
DateArr[12] = 20010403;TradeArr[12] = 1;PriceArr[12] = 1059.25;
DateArr[13] = 20010419;TradeArr[13] = -1;PriceArr[13] = 1209.75;
DateArr[14] = 20010501;TradeArr[14] = 1;PriceArr[14] = 1223.50;
DateArr[15] = 20010521;TradeArr[15] = -1;PriceArr[15] = 1265.50;
DateArr[16] = 20010530;TradeArr[16] = 1;PriceArr[16] = 1201.00;
Sample OutPut

Here is the interpreter strategy code.  I simply copied and pasted the output from the file as the header to this strategy.

arrays: DateArr[5000](0),TradeArr[5000](0),PriceArr[5000](0);//
DateArr[1] = 20001211;TradeArr[1] = -1;PriceArr[1] = 1368.50;// All these lines
DateArr[2] = 20001220;TradeArr[2] = 1;PriceArr[2] = 1245.75;//  created by Perfect
DateArr[3] = 20010102;TradeArr[3] = -1;PriceArr[3] = 1265.50;// System Out Put
DateArr[4] = 20010105;TradeArr[4] = 1;PriceArr[4] = 1269.50;//
DateArr[5] = 20010131;TradeArr[5] = -1;PriceArr[5] = 1346.50;//
DateArr[6] = 20010201;TradeArr[6] = 1;PriceArr[6] = 1348.00;//
DateArr[7] = 20010205;TradeArr[7] = -1;PriceArr[7] = 1328.25;//
DateArr[8] = 20010228;TradeArr[8] = 1;PriceArr[8] = 1203.75;//
DateArr[9] = 20010307;TradeArr[9] = -1;PriceArr[9] = 1231.25;//
DateArr[10] = 20010322;TradeArr[10] = 1;PriceArr[10] = 1070.00;//
...
...
...
...
DateArr[539] = 20180920;TradeArr[539] = -1;PriceArr[539] = 2939.50;//


vars: cnt(1);

If date +19000000 = DateArr[cnt]  then
Begin 
	print("Found Date : ",date);
	If tradeArr[cnt] = 1 then
	begin
		buy this bar on close;
	end;
	If tradeArr[cnt] = 2 then
	begin
		sell this bar on close;
	end;
	If tradeArr[cnt] = -1 then
	begin
		sellShort this bar on close;
	end;
	If tradeArr[cnt] = -2 then
	begin
		buyToCover this bar on close;
	end;
	cnt = cnt + 1;
	If DateArr[cnt] = DateArr[cnt-1] then
	Begin
		print("two trades same day ",d," ",dateArr[cnt]);
		If tradeArr[cnt] = 1 then
		begin
			buy this bar on close ;
		end;
		If tradeArr[cnt] = 2 then
		begin
			sell this bar on close;
		end;
		If tradeArr[cnt] = -1 then
		begin
	    	print("looking to go short at ",PriceArr[cnt]);
			sellShort this bar on close;
		end;
		If tradeArr[cnt] = -2 then
		begin
			buyToCover this bar on close;
		end;
		cnt = cnt + 1;
	end;	
end;
Almost Perfect System Interpreter Code

I think I discussed how this code works in a prior post.¬† If I go back and see that I didn’t I will edit and update this post.¬† So how did it do?

Here is a zip file that contains both ELDs for the strategies : PERFECTSYSTEM

EasyLanguage Code for Optimal F (Multi-Charts and VBA too!)

Optimal F in EasyLanguage for TradeStation and MultiCharts

Here is the code for the Optimal F calculation.¬† For a really good explanation of Optimal F I refer you to Ralph Vince’s Book Portfolio Management FORMULAS.¬† We had programmed this years ago for our Excalibur software and I was surprised the EasyLanguage code was really all that accessible on the internet.¬† Finding the optimal f is found through an iterative process or in programmers terms a loop.¬† The code is really quite simple and I put it into a Function.¬† I decided to create this function because I wanted to demonstrate the ideas from my last post on how a function can store variable and array data.¬† Plus this code should be readily available somewhere out there.

//OptimalFGeo by George Pruitt
//My interpretation Sept. 2018
//www.georgepruitt.com
//georgeppruitt@gmail.com

input: minNumTrades(numericSimple);
vars: totalTradesCount(0),tradeCnt(0);
array: tradesArray[500](0);

vars: iCnt(00),jCnt(00),grandTot(0),highI(0);
vars: optF(0.0),gMean(0.0),fVal(0.0),HPR(0.0),TWR(0.0),hiTWR(0.0);
vars: biggestLoser(0.0),gat(0.0);

totalTradesCount = totalTrades;
If totalTradesCount > totalTradesCount[1] then
begin
	tradeCnt = tradeCnt + 1; 
	tradesArray[tradeCnt] = positionProfit(1);
end;

// Taken from my Fortran library - GPP and Vince Book PMF

optF = 0.0;
gMean = 1.00;
gat   = 0.00;
//Only calculate if new trade
IF(tradeCnt>minNumTrades and totalTradesCount > totalTradesCount[1]) then 
Begin
	biggestLoser = 0;
	grandTot = 0;
	For iCnt = 1 to tradeCnt //get the biggest loser
	begin
   		grandTot = grandTot + tradesArray[iCnt];
   		IF(tradesArray[iCnt]<biggestLoser) then biggestLoser = tradesArray[iCnt];
	end;
//	print(grandTot," ",biggestLoser);
	IF({grandTot > 0 and} biggestLoser <0) then 
	begin
//		print("Inside TWR Calculations");
		highI = 0;
		hiTWR = 0.0;
		for iCnt = 1 to 100
		begin
			fVal = .01 * iCnt;
			TWR = 1.0;
			for jCnt = 1 to tradeCnt // calculate the Terminal Wealth Relative
			begin
    			HPR = 1. + (fVal * (-1*tradesArray[jCnt]) / biggestLoser);
    			TWR = TWR * HPR;
 //   			print(fVal," ",iCnt," " ,jCnt," Trades ",tradesArray[jCnt]," HPR ",HPR:6:4," TWR : ",TWR:6:4," hiTWR",hiTWR:6:4," bl ",biggestLoser);
			end;
//			print(iCnt," ",TWR," ",hiTWR);
			IF(TWR>hiTWR) THEN
			begin
    			hiTWR = TWR;
    			optF = fVal;    	// assign optF to fVal in case its the correct one		
			end
			else
    			break;                     //highest f found - stop looping
		end;		
		If (TWR <= hiTWR or optF >= 0.999999) then
		begin
			TWR  = hiTWR;
			OptimalFGeo = optF;  //assign optF to the name of the function
		end;	
		gmean = power(TWR,(1.0 / tradeCnt));
		
		if(optF<>0) then GAT   = (gMean - 1.0) * (biggestLoser / -(optF));		
		print(d," gmean ",gmean:6:4," ",GAT:6:4);  // I calculate the GMEAN and GeoAvgTrade
	end;
end;
Optimal F Calculation by Ralph Vince code by George Pruitt

VBA version of Optimal F

For those of you who have a list of trades and want to see how this works in Excel here is the VBA code:

Sub OptimalF()

    Dim tradesArray(1000) As Double
    i = 0
    biggestLoser = 0#
    Do While (Cells(3 + i, 1) <> "")
        tradesArray(i) = Cells(3 + i, 1)
        If tradesArray(i) < bigLoser Then biggestLoser = tradesArray(i)
        i = i + 1
    Loop
    tradeCnt = i - 1
    highI = 0
    hiTWR = 0#
    rc = 3
    For fVal = 0.01 To 1 Step 0.01
        TWR = 1#
        For jCnt = 0 To tradeCnt
            HPR = 1# + (fVal * (-1 * tradesArray(jCnt)) / biggestLoser)
            TWR = TWR * HPR
            Cells(rc, 5) = jCnt
            Cells(rc, 6) = tradesArray(jCnt)
            Cells(rc, 7) = HPR
            Cells(rc, 8) = TWR
            rc = rc + 1
        Next jCnt
        Cells(rc, 9) = fVal
        Cells(rc, 10) = TWR
        rc = rc + 1

        If (TWR > hiTWR) Then
            hiTWR = TWR
            optF = fVal
        Else
            Exit For
        End If

    Next fVal
    If (TWR <= hiTWR Or optF >= 0.999999) Then
        TWR = hiTWR
        OptimalFGeo = optF
    End If
    Cells(rc, 8) = "Opt f"
    Cells(rc, 9) = optF
    rc = rc + 1
    gMean = TWR ^ (1# / (tradeCnt + 1))
    If (optF <> 0) Then GAT = (gMean - 1#) * (biggestLoser / -(optF))
    Cells(rc, 8) = "Geo Mean"
    Cells(rc, 9) = gMean
    rc = rc + 1
    Cells(rc, 8) = "Geo Avg Trade"
    Cells(rc, 9) = GAT

End Sub
VBA code for Optimal F

I will attach the eld and .xlsm file a little later.

 

 

 

Function Variable Data Survives Between Calls

Function Variable Data Survives from One Call to the Next – A Pretty Nifty Tool in EasyLanguage!

Creating a function that can store data and then have that data survive on successive function calls without having to pass information back and forth is really a cool and powerful tool in EasyLanguage.  In most programming languages, the variables defined in a function are local to that particular bit of code and once program execution exits the function, then the data is destroyed.  There are two exceptions (in other languages) that come to mind Рif the variable is passed back and forth via their addresses, then the data can be maintained or if the variable is global in scope to the function and the calling program.  EasyLanguage prevents you from having to do this and this can definitely save on headaches.  I wrote a function that defines an array that will hold a list of trades.  Once the number of trades reaches a certain level, I then calculate a moving average of the last 10 trades.  The average is then passed back to the calling strategy.  Here is the simple code to the function.

 

{Function Name:   StoreTradesFunc by George Pruitt}
{Function to Calculate the average trade for past N trades.
 ----------------------------------------------------------
 Function remembers the current trade count in tradeCnt.
 It also remembers the values in the array tradesArray.
 It does this between function calls. 
 Values - simple and array - undoubtedly are global to the function}
 
input: avgTradeCalcLen(numericSimple);
vars: totalTradesCount(0),tradeCnt(0);
array: tradesArray[500](0);

totalTradesCount = totalTrades;
If totalTradesCount > totalTradesCount[1] then
begin
	tradeCnt = tradeCnt + 1;
	tradesArray[tradeCnt] = positionProfit(1);
//	print("Storing data ",tradesArray[tradeCnt]," ",tradeCnt);
end;

If totalTrades > avgTradeCalcLen then
begin
	Value2 = 0;
	For value1 = totalTrades downTo totalTrades - avgTradeCalcLen
	begin
		Value2 = value2 + tradesArray[value1];
	end;
	print("Sum of last 10 Trades: ",value2);
	StoreTradesFunc = value2/avgTradeCalcLen;
end;
Store A List of Trades in a Function

I call this function on every bar (daily would be best but you could do it on intra-day basis) and it polls the function/keyword totalTrades¬†to see if a new trade has occurred.¬† If one has, then the variable tradeCnt is incremented and the trade result is inserted into the tradesArray array by using the tradeCnt as the array index.¬† When you come back into the function from the next bar of data tradeCnt and tradesArray are still there for you and most importantly still intact.¬† In other words there values are held static until you change them and remembered.¬†¬†This really comes in handy when you want to store all the trades in an array and then do some type of calculation on the trades and then have that information readily available for use in the strategy.¬† My example just provides the average trade for the past ten trades.¬† But you could do some really exotic things like Optimal F.¬† The thing to remember is once you define a variable or an array in a function and start dumping stuff in them, the stuff will be remembered throughout the life of the simulation.¬† The function data and variables are encapsulated to just the function scope – meaning I can’t access tradesArray outside of the function.¬† One last note – notice how I was able to determine a new trade had occurred.¬† I assigned the result of totalTrades to my own variable totalTradesCount and then compared the value to the prior bar’s value.¬† If the values were different than I knew a new trade had just completed.

Python Script To Import List of Trades into TradeStation’s EasyLanguage – Sort of

Converting A List of Trades, Dates and Prices Into EasyLanguage Arrays:

As the old saying goes “a picture is worth a thousand words!”¬† Have you ever been given a list of trades like this:

Sell Short,20010622,1178.50 
Buy to Cover,20010626,1159.75 
Sell Short,20010801,1150.00 
Buy to Cover,20010807,1139.75 
Sell Short,20010814,1129.00 
Buy to Cover,20010816,1117.25 
Sell Short,20011001,976.75 
Buy to Cover,20011004,1016.75 
Sell Short,20011107,1053.00 
Buy to Cover,20011123,1069.50 
Sell Short,20011219,1076.25 
Buy to Cover,20020102,1075.00 
Sell Short,20020129,1067.25 
Buy to Cover,20020131,1046.75 
Sell Short,20020131,1046.75 
Buy to Cover,20020205,1026.75 
Sell Short,20020520,1033.25 
Buy to Cover,20020522,1011.50 
Sell Short,20020731,832.00 
Buy to Cover,20020805,792.50 
Sell Short,20020812,834.00 
Buy to Cover,20020814,811.75 
Sell Short,20020911,838.50 
Buy to Cover,20020913,816.75 
List of Trades : Order, Date, Price

But really wanted to see this:

I have created a small Python script that will take a list of trades like those listed in table above and create the following EasyLanguage:

arrays: DateArr[500](0),TradeArr[500](""),PriceArr[500](0);
DateArr[0]=1010622;TradeArr[0]="SS";PriceArr[0]=1178.5;
DateArr[1]=1010626;TradeArr[1]="SX";PriceArr[1]=1159.75;
DateArr[2]=1010801;TradeArr[2]="SS";PriceArr[2]=1150.0;
DateArr[3]=1010807;TradeArr[3]="SX";PriceArr[3]=1139.75;
DateArr[4]=1010814;TradeArr[4]="SS";PriceArr[4]=1129.0;
DateArr[5]=1010816;TradeArr[5]="SX";PriceArr[5]=1117.25;
DateArr[6]=1011001;TradeArr[6]="SS";PriceArr[6]=976.75;
DateArr[7]=1011004;TradeArr[7]="SX";PriceArr[7]=1016.75;
DateArr[8]=1011107;TradeArr[8]="SS";PriceArr[8]=1053.0;
DateArr[9]=1011123;TradeArr[9]="SX";PriceArr[9]=1069.5;
DateArr[10]=1011219;TradeArr[10]="SS";PriceArr[10]=1076.25;
DateArr[11]=1020102;TradeArr[11]="SX";PriceArr[11]=1075.0;
DateArr[12]=1020129;TradeArr[12]="SS";PriceArr[12]=1067.25;
DateArr[13]=1020131;TradeArr[13]="SX";PriceArr[13]=1046.75;
DateArr[14]=1020131;TradeArr[14]="SS";PriceArr[14]=1046.75;
DateArr[15]=1020205;TradeArr[15]="SX";PriceArr[15]=1026.75;
DateArr[16]=1020520;TradeArr[16]="SS";PriceArr[16]=1033.25;
DateArr[17]=1020522;TradeArr[17]="SX";PriceArr[17]=1011.5;
Converting list of trades to EasyLanguage

This just creates the arrays that you can use to graph the trades on a chart.¬† If you are using exact prices you got to make sure your data aligns with the prices in the list of trades.¬† If you are only entering on the open or the close of the bar then the price array isn’t necessary.

The following Python script will also be helpful if you want to learn how to open a file in csv format, read it into lists, convert it and then save the output to a file.

#-------------------------------------------------------------------------------
# Name:        Read csv file via askOpen and save txt file via askSave
# Purpose:     Read the trade metrics from a TradeStation csv format
#              and build arrays from the information to display on charts in
#              TradeStation
# Author:      georg
#
# Created:     29/08/2018
# Copyright:   (c) georg 2018
#-------------------------------------------------------------------------------
import csv
import tkinter as tk
import os.path
from tkinter.filedialog import askopenfilenames
from tkinter.filedialog import asksaveasfilename

tradeType = list()
tradeDate = list()
tradePrice = list()

def main():
    root = tk.Tk()
    root.withdraw()
    files = askopenfilenames(filetypes=(('CSV files', '*.csv'),
                                       ('TXT files', '*.txt')),
                                       title='Select CSV format only!')
    fileList = root.tk.splitlist(files)
    fileListLen = len(fileList)


# make sure you know the format ahead of time
# I know "Buy",20180828,2745.75
#
    cnt = 0
    for files in range(0,fileListLen):
        head,tail = os.path.split(fileList[files])
        with open(fileList[files]) as f:
            f_csv = csv.reader(f)
            for row in f_csv:
                numCols = len(row)
                tradeType.append(row[0])
                tradeDate.append(int(row[1]))
                tradePrice.append(float(row[2]))
                cnt += 1
        f.close


    filename = asksaveasfilename(title="Will Save File with '.txt'",defaultextension=".txt")
#    filename = filename + '.txt'
    target1 = open(filename,'w')
    outString = 'arrays: DateArr[500](0),TradeArr[500](0),PriceArr[500](0);\n'
    target1.write(outString)
    for x in range(0,cnt):
        if tradeType[x] == "Sell Short": tradeType[x] = "SS"
        if tradeType[x] == "Buy": tradeType[x] = "B"
        if tradeType[x] == "Buy to Cover": tradeType[x] = "SX"
        if tradeType[x] == "Sell": tradeType[x] = "LX"
        outString = 'DateArr['+str(x)+']='+str(tradeDate[x]-19000000)+';TradeArr['+str(x)+']="'+tradeType[x]+'";PriceArr['+str(x)+']='+str(tradePrice[x])+';\n'
        target1.write(outString)
    target1.close


if __name__ == '__main__':
    main()
Python Script Open, Read, Convert and Write A File Using TK Dialogs

And here is the EasyLanguage code that will step through the arrays and place the trades accordingly.  I noticed that sometimes two trades could occur on the same bar, but only two and you will notice in the code where I programmed this occurrence.

vars: cnt(0);

If date of tomorrow = DateArr[cnt] then
Begin
	print("Inside: ",d," ",dateArr[cnt]);
	If tradeArr[cnt] = "B" then
	begin
		buy next bar at PriceArr[cnt] stop;
	end;
	If tradeArr[cnt] = "LX" then
	begin
		sell next bar at PriceArr[cnt] stop;
	end;
		If tradeArr[cnt] = "SS" then
	begin
		sellShort next bar at PriceArr[cnt] stop;
	end;
	If tradeArr[cnt] = "SX" then
	begin
		buyToCover next bar at PriceArr[cnt] stop;
	end;
	cnt = cnt + 1;
	If DateArr[cnt] = DateArr[cnt-1] then
	Begin
		print("two trades same day ",d," ",dateArr[cnt]);
		If tradeArr[cnt] = "B" then
		begin
			buy next bar at PriceArr[cnt] stop;
		end;
		If tradeArr[cnt] = "LX" then
		begin
			sell next bar at PriceArr[cnt] stop;
		end;
		If tradeArr[cnt] = "SS" then
		begin
	    	print("looking to go short at ",PriceArr[cnt]);
			sellShort next bar at PriceArr[cnt] stop;
		end;
		If tradeArr[cnt] = "SX" then
		begin
			buyToCover next bar at PriceArr[cnt] stop;
		end;
		cnt = cnt + 1;
	end;	
end;
EasyLanguage Snippet To Execute Trades Stored in Arrays

 

 

Using TradeStation’s COT Indicator to Develop a Trading System

TradeStation’s COT (Commitment of Traders) Indicator:

TradeStation COT Indicator

TradeStation now includes the historic COT (Commitment of Traders) report in the form of an indicator.

If you can plot it then you can use it in a Strategy.  The following code listing takes the Indicator code and with very few modifications turns it into a trading system.

{
Net positions of various groups of traders from the CFTC's weekly Commitments of
Traders report.  "Net" positions are calculated by taking the number of contracts
that a group of traders is long and subtracting the number of contracts that that
group of traders is short.

The user input "FuturesOnly_Or_FuturesAndOptions_1_or_2" determines whether the
CFTC's "Futures Only" report is used, or the "Futures and Options" report is
used to determine the positions of the various groups of traders.  By default, the
"Futures Only" report is used.

Plot1:  Commercial traders' net position
Plot2:  Non-commercial traders' net position
Plot3:  Speculators' net positions, for speculators not of reportable size
Plot4:  Zero line

If an error occurs retrieving one of the values used by this study, or if the value
is not applicable or non-meaningful, a blank cell will be displayed in RadarScreen or
in the OptionStation assets pane.  In a chart, no value will be plotted until a value
is obtained without generating an error when retrieved.
}

input:  FuturesOnly_Or_FuturesAndOptions_1_or_2( 1 ) ; { set to 1 to use the CFTC's
 "Futures Only" report, set to 2 (or to any value other than 1) to use the "Futures
 and Options" report }

variables:
	Initialized( false ),
	FieldNamePrefix( "" ),
	CommLongFieldNme( "" ),
	CommShortFieldNme( "" ),
	NonCommLongFieldNme( "" ),
	NonCommShortFieldNme( "" ),
	SpecLongFieldNme( "" ),
 	SpecShortFieldNme( "" ),
    CommLong( 0 ),
	oCommLongErr( 0 ),
	CommShort( 0 ),
	oCommShortErr( 0 ),
	NonCommLong( 0 ),
	oNonCommLongErr( 0 ),
	NonCommShort( 0 ),
	oNonCommShortErr( 0 ),
	SpecLong( 0 ),
	oSpecLongErr( 0 ),
	SpecShort( 0 ),
	oSpecShortErr( 0 ),
	CommNet( 0 ),
	NonCommNet( 0 ),
	SpecNet( 0 ) ;

if Initialized = false then
	begin
	if Category > 0 then
		RaiseRuntimeError( "Commitments of Traders studies can be applied only to" +
		 " futures symbols." ) ;
	Initialized = true ;
	FieldNamePrefix = IffString( FuturesOnly_Or_FuturesAndOptions_1_or_2 = 1,
	 "COTF-", "COTC-" ) ;
	CommLongFieldNme = FieldNamePrefix + "12" ;
	CommShortFieldNme = FieldNamePrefix + "13" ;
	NonCommLongFieldNme = FieldNamePrefix + "9" ;
	NonCommShortFieldNme = FieldNamePrefix + "10" ;
	SpecLongFieldNme = FieldNamePrefix + "16" ;
 	SpecShortFieldNme = FieldNamePrefix + "17" ;	
	end ;

CommLong = FundValue( CommLongFieldNme, 0, oCommLongErr ) ;
CommShort = FundValue( CommShortFieldNme, 0, oCommShortErr) ;
NonCommLong = FundValue( NonCommLongFieldNme, 0, oNonCommLongErr ) ;
NonCommShort = FundValue( NonCommShortFieldNme, 0, oNonCommShortErr );
SpecLong = FundValue( SpecLongFieldNme, 0, oSpecLongErr ) ; 
SpecShort = FundValue( SpecShortFieldNme, 0, oSpecShortErr ) ;

if oCommLongErr = fdrOk and oCommShortErr = fdrOk then
	begin
	CommNet = CommLong - CommShort ;
	Print ("CommNet ",commNet);
	end ;

if oNonCommLongErr = fdrOk and oNonCommShortErr = fdrOk then
	begin
	NonCommNet = NonCommLong - NonCommShort ;
	end ;

if oSpecLongErr = fdrOk and oSpecShortErr = fdrOk then
	begin
	SpecNet = SpecLong - SpecShort ;
	end ;
If CommNet < 0  then sellShort tomorrow at open;
If CommNet > 0 then buy tomorrow at open;


{ ** Copyright (c) 2001 - 2010 TradeStation Technologies, Inc. All rights reserved. ** 
  ** TradeStation reserves the right to modify or overwrite this analysis technique 
     with each release. ** }
COT Indicator Converted To Strategy

Line numbers 90 and 91 informs TS to take a long position if the Net Commercial Interests are positive and a short position if the Commercials are negative.  I kept the original comments in place in  case you wanted to see how the indicator and its associated function calls work.  The linchpin of this code lies in the function call FundValue.  This function call pulls fundamental data from the data servers and provides it in an easy to use format.  Once you have the data you can play all sorts of games with it.  This is just a simple system to see if the commercial traders really do know which direction the market is heading.

if you test this strategy on the ES you will notice a downward sloping 45 degree equity curve.  This leads me to believe the commercials are trying their best to  use the ES futures to hedge other market positions.  If you go with the non Commercials you will see  a totally different picture.  To do this just substitute the following two lines:

If CommNet < 0 then sellShort tomorrow at open;
If CommNet > 0 then buy tomorrow at open;

With:

If NonCommNet < 0 then sellShort tomorrow at open;
If NonCommNet > 0 then buy tomorrow at open;

I said a totally different picture not a great one.  Check out if the speculators know better.

Back Adjusted Continuous Contract Data from Quandl

I finally got around to programming an easy Python down loader for CME data from Quandl.¬† I promised this in my last book, but haven’t got that many requests.¬† ¬†On the Ultimate Algorithmic Trading System…. page I have provided some data going back to early 2008 on several markers.¬† This is a work in progress.¬† I also wrote a back adjuster that rolls when the Open Interest + Volume is greater in the next contract and adjusts the data retroactively (Panama process) by the discount on the roll date.¬† As I state on the web page this data is not all that great even though it looks like it comes directly from CME.¬† ¬†I do fill in gaps and try to fix glaring errors so take it for what it is.¬† Its good data for preliminary testing, but to finalize an algorithm or to trade by I would definitely go the “paid” route.¬† Although the later data looks pretty good.

Before downloading the data I would sign up for Quandl and get an API key.  You will be amazed at the amount of data they make available Рfree and paid subscriptions.

 

 

Anatomy Of Mean Reversion in EasyLanguage

Look at this equity curve:

As long as you are in a bull market buying dips can be very consistent and profitable.¬† But you want to use some type of entry signal and trade management other than just buying a dip and selling a rally.¬† Here is the anatomy of a mean reversion trading algorithm that might introduce some code that you aren’t familiar.¬† Scroll through the code and I will¬† summarize below.

inputs: mavlen(200),rsiLen(2),rsiBuyVal(20),rsiSellVal(80),holdPeriod(5),stopLoss$(4500);
vars: iCnt(0),dontCatchFallingKnife(false),meanRevBuy(false),meanRevSell(false),consecUpClose(2),consecDnClose(2);

Condition1 = c > average(c,mavLen);

Condition2 = rsi(c,rsiLen) < rsiBuyVal;
Condition3 = rsi(c,rsiLen) > rsiSellVal;


Value1 = 0;
Value2 = 0;

For iCnt = 0 to consecUpClose - 1 
Begin
	value1 = value1 + iff(c[iCnt] > c[iCnt+1],1,0);
end;

For iCnt = 0 to consecDnClose - 1 
Begin
	Value2 = value2 + iff(c[iCnt] < c[iCnt+1],1,0);
end;

dontCatchFallingKnife = absValue(C - c[1]) < avgTrueRange(10)*2.0;

meanRevBuy = condition1 and condition2 and dontCatchFallingKnife;
meanRevSell =  not(condition1) and condition3 and dontCatchFallingKnife;

If meanRevBuy then buy this bar on close;
If marketPosition = 1 and condition1 and value1 >= consecUpClose then sell("ConsecUpCls") this bar on close;

If meanRevSell then sellShort this bar on close;
If marketPosition = -1 and not(condition1) and value2 >= consecDnClose then buyToCover this bar close;

setStopLoss(stopLoss$);


If barsSinceEntry = holdPeriod then
Begin
	if marketPosition = 1 and not(meanRevBuy) then sell this bar on close;
	if marketPosition =-1 and not(meanRevSell) then buytocover this bar on close;
end;
Mean Reversion System

I am using a very short term RSI indicator, a la Connors, to initiate long trades.¬† Basically when the 2 period RSI dips below 30 and the close is above the 200-day moving average I will buy only if I am not buying “a falling knife.”¬† In February several Mean Reversion algos kept buying as the market fell and eventually got stopped out with large losses.¬† Had they held on they probably would have been OK.¬† Here I don’t buy if the absolute price difference between today’s close and yesterday’s is greater than 2 X the ten day average true range.¬† Stay away from too much “VOL.”

Once a trade is put on I use the following logic to keep track of consecutive closing relationships:

For iCnt = 0 to consecUpClose - 1 
Begin
	value1 = value1 + iff(c[iCnt] > c[iCnt+1],1,0);
end;
Using the IFF function in EasyLanguage

Here I am using the IFF function to compare today’s close with the prior day’s.¬† iCnt is a loop counter that goes from 0 to 1. IFF checks the comparison and if it’s true it returns the first value after the comparison and if false it returns the last value.¬† Here if I have two consecutive up closes value1 accumulates to 2.¬† If I am long and I have two up closes I get out.¬† With this template you can easily change this by modifying the input:¬† consecUpClose.¬† Trade management also includes a protective stop and a time based exit.¬† If six days transpire without two up closes then the system gets out – if the market can’t muster two positive closes, then its probably not going to go anywhere.¬† The thing with mean reversion, more so with other types of systems, is the use or non use of a protective stop.¬† Wide stops are really best, because you are betting on the market to revert.¬† Look at the discrepancy of results using different stop levels on this system:

Here an $1,800 stop only cut the max draw down by $1,575.  But it came at a cost of $17K in profit.  Stops, in the case of Mean Reversion, are really used for the comfort of the trader.

This code has the major components necessary to create a complete trading system.  Play around with the code and see if you can come up with a better entry mechanism.

Updated Pattern Smasher in EasyLanguage

Update To Original Pattern Smasher

What will you learn : string manipulation, for-loops, optimization

Before proceeding I would suggest reading my original post on this subject.    If you believe the relationship of the last few bars of data can help determine future market direction, then this post will be in you wheel house.  Another added benefit is that you will also learn some cool EasyLanguage.

Original post was limited to four day patterns!

This version is limitless (well not really, but pretty close).¬† Let’s stick with the original string pattern nomenclature (+ + – – : two up closes followed by two down closes.)¬† Let’s also stick with our binary pattern representation:

Pattern # 2^3 2^2 2^1 1
3 0 0 1 1
4 0 1 0 0
5 0 1 0 1
6 0 1 1 1

Remember a 0 represents a down close and a 1 represents an up close.¬† We will deviate from the original post by doing away with the array and stick with only strings (which are really just arrays of characters.)¬† This way we won’t have to worry about array manipulation.

How to create a dynamic length string pattern

This was the difficult part of the programming.¬† I wanted to be able to optimize 3, 4 and 5 day patterns and I wanted to control this with using just inputs.¬† I discovered that pattern three is different in a three day pattern than it is in a four day pattern: in a three day pattern it is 011 or – + + and in a four day pattern it is 0011 or – – + +.¬†¬†Since I am counting 0’s as down closes, pattern #3 depends on the ultimate size of the pattern string.¬† No worries I will have eventually have another version where I utilize a different value for down closes and we can then have holes in our string patterns.¬† But I digress – so to differentiate the patterns based on the pattern length I included a maxPatternLen input.¬† So if maxPatternLen is three and we are trying to match pattern #3 then we will be looking for 011 and not 0011.¬† That was an easy fix.¬† But then I wanted to build a string pattern based on this input and the pattern number dynamically.¬† Here is some psuedo code on how I figured it out.


{Psuedo code to translate pattern number into binary number}
patternNumber = 3
maxPatternLen = 3

numBits = 0    						// stick with binary representation
testValue = 0						// temporary test value
numBits = maxPatternLen-1  			// how many bits will it take to get to the
									// center of - or numBits to represent max
									// number of patterns or 2^numBits
currentBit =numBits					// start wit current bit as total numBits

value1 = patternOptTest				// value1 represents current pattern number
testString = ""  					// build test string from ground up


for icnt = numBits downto 0			//building string from left to right
begin       						// notice keyword downto
	if power(2,currentBit) > value1 then  // must use power function in EL
	begin							// if the very far left bit value > 
		testString = testString + "-"	  // patten number then plug in a "-"
	end
	else
	begin							// else plug in a "+" and deccrement by
		testString = testString + "+"	 // that bits value - if its the 3rd bit
	value1 = value1 - power(2,currentBit)// then decrement by 8
	end;
	currentBit = currentBit - 1		// move onto the next bit to the right
end;
Pseudocode for Binary Representation of Pattern #

Now if you want to optimize then you must make sure your pattern number search space or range can be contained within maxPatternLen.¬† For example, if you want to test all the different combinations of a four day pattern, then your maxPatternLen would naturally be four and you would optimize the pattern number from 0 to 15.¬† Don’t use 1-16 as I use zero as the base.¬† A five day pattern would include the search space from 0 – 31.¬† The rest of the code was basically hacked from my original post.¬† ¬†Here is the rest of the code to do optimizations on different length pattern strings.¬† Notice how I use strings, for-loops and comparisons.

input: buyPattern("+++-"),sellPattern("---+"),patternOptimize(True),patternOptTest(7),maxPatternLen(3),patternOptBuySell(1),
	   stopLoss$(2000),profitTarg$(2000),holdDays(5);
vars: buyPatternString(""),sellPatternString(""),buyPatternMatch(""),sellPatternMatch(""),numBits(0),testValue(0),currentBit(0),
      remainder(0),value(0),icnt(0),testString(""),numCharsInBuyPattern(0),numCharsInSellPattern(0);
vars:okToBuy(false),okToSell(false);

buyPatternMatch = buyPattern;
sellPatternMatch = sellPattern;
numCharsInBuyPattern = strLen(buyPatternMatch);
numCharsInSellPattern = strLen(sellPatternMatch);

If patternOptimize then
begin
	numBits = 0;
    testValue = 0;
    value = maxPatternLen;
    numBits = maxPatternLen-1;  
    currentBit =numBits;
    remainder = patternOptTest;
    testString = "";
    for icnt = numBits downto 0
    begin       
        if power(2,currentBit) > value1 then
        begin
            testString = testString + "-";
        end
        else
        begin
            testString = testString + "+";
            remainder = remainder - power(2,currentBit);
        end;
        currentBit = currentBit - 1;
	end;
	numCharsInBuyPattern = maxPatternLen;
	numCharsInSellPattern = maxPatternLen;
	if patternOptBuySell = 1 then
	Begin
		buyPatternMatch = testString;
		sellPatternMatch = "0";
	end;
	If patternOptBuySell = 2 then
	Begin
		buyPatternMatch = "0";
		sellPatternMatch = testString;
	end;
end;
	

buyPatternString = "";
sellPatternString = "";

For icnt = numCharsInBuyPattern-1 downto 0
Begin
	If close[icnt] >= close[icnt+1] then buyPatternString = buyPatternString + "+";
	If close[icnt] < close[icnt+1] then buyPatternString = buyPatternString + "-";
end;
For icnt = numCharsInSellPattern-1 downto 0
Begin
	If close[icnt] >= close[icnt+1] then sellPatternString = sellPatternString + "+";
	If close[icnt] < close[icnt+1] then sellPatternString = sellPatternString + "-";
end;


okToBuy = false;
okToSell = false;

if buyPatternMatch <> "" then
	If buyPatternString = buyPatternMatch then okToBuy = true;
If buyPatternMatch = "" then
	okToBuy = true;
If sellPattern <> "" then
	If sellPatternString = sellPatternMatch then okToSell = true;
If sellPatternMatch = "" then
	okToSell = true;
	
If okToBuy then buy next bar at open;
If okToSell then sellshort next bar at open;

If marketPosition = 1 and barsSinceEntry > holdDays then sell next bar at open;
If marketPosition = -1 and barsSinceEntry > holdDays then buytocover next bar at open;

setStopLoss(stopLoss$);
setProfitTarget(profitTarg$);

If lastBarOnChart then print(d," ",buyPatternMatch);
Final Version of New Pattern Smasher

Also see how I incorporate a profit target and protective stop.¬† I use the built in¬†BarsSinceEntry function¬†to count the number of days I am in a trade so I can utilize a time based exit.¬† Here is an interesting equity curve I developed using a two day pattern ( – –) to go long.

Register on the website and I will email you an ELD of the improved Pattern Smasher.  Or just shoot me an email.

 

 

Making Trading Decisions on Current Month’s Profit/Loss

Keeping track of intra-month profit or loss

In real time trading I have noticed that once you reach a certain loss for the month its best, sometimes, to circle the wagons and quit trading until the beginning of the next month.¬† This concept works best for very short term or day trade algorithms, as its very easy to get started back up.¬† You can do this with Trend Following, but you must build a logical and replicable process for re-entering existing positions.¬† Let’s assume a trading algorithm whose averaging losing month is $1500 and you are currently down $2000 – what are the chances that you will revert to the mean or draw down further?¬† Probably 50/50.¬† Who knows you might turn around and actually make money by month’s end.¬† If you review a track record of a hedge fund manager, trader, or algorithm and they show a bar chart of monthly returns and there sticking out like a sore thumb is a big down bar, that kind of makes you think that could happen again.¬† If you can control the monthly downside without sacrificing the existing Profit:DrawDown ratio, then why not do it.

Sample Code To Monitor IntraMonth $P/L

if month(date) <> month(date[1]) then
Begin
	begMonthProf = netProfit; 
	print(d," ",t," ",begMonthProf);
	canTrade = true;
end;
Capture Beginning Of Month Net Profit

Here I am comparing the month of the current bar against the month of the prior bar.¬† If they are not equal, then we have a new month.¬† Store the netProfit¬†in the variable begMonthProf.¬† All you have to do is compare the current bar’s netProfit to begMonthProf and make a decision.¬† Here is some code:

Making a Trading Decision Based on Monthly $P/L

		If dayOfMonth(date) > 15 and begMonthProf - netProfit >= intraMonthMaxLoss then canTrade = false;
If Down MaxLoss for Month and Past Mid-Month - Quit Trading

If the day of the month is greater than 15 (month half over) and the difference between the current netProfit and begMonthProfit is greater than a negative intraMonthMaxLoss then quit trading for the month.  Only turn it back on the first bar of the next month.  See how this works for your algos.

How to Create a Dominant Cycle Class in Python

John Ehlers used the following EasyLanguage code to calculate the Dominant Cycle in a small sample of data.¬† If you are interested in cycles and noise reduction, definitely check out the books by John Ehlers – “Rocket Science for Traders” or “Cybernetic Analysis for Stocks and Futures.”¬† I am doing some research in this area and wanted to share how I programmed the indicator/function in Python.¬† I refer you to his books or online resources for an explanation of the code.¬† I can tell you it involves an elegantly simplified approach using the Hilbert Transform.

 

Inputs:	Price((H+L)/2);

Vars:	Imult(.635),
		Qmult (.338),
		InPhase(0),
		Quadrature(0),
		count(0),
		Re(0),
		Im(0),
		DeltaPhase(0),
		InstPeriod(0),
		Period(0);

If CurrentBar > 8 then begin
	Value1 = Price - Price[7];
 	Inphase = 1.25*(Value1[4]  - Imult*Value1[2]) + Imult*InPhase[3];
	 	
//    print(price," ",price[7]," ",value1," ",inPhase," ",Quadrature," ",self.im[-1]," ",self.re[-1])	
//	print(d," ",h," ",l," ",c," ",Value1[4]," ",Imult*Value1[2]," ", Imult*InPhase[3]," ",Inphase);
	Quadrature = Value1[2] - Qmult*Value1 + Qmult*Quadrature[2];
	Re = .2*(InPhase*InPhase[1] + Quadrature*Quadrature[1]) + .8*Re[1];
	Im = .2*(InPhase*Quadrature[1] - InPhase[1]*Quadrature)   + .8*Im[1];
	print(d," ",o," ",h," ",l," ",c," ",value1," ",inPhase," ",Quadrature," ",Re," ",Im);
	If Re <> 0 then DeltaPhase = ArcTangent(Im/Re);

	{Sum DeltaPhases to reach 360 degrees.  The sum is the instantaneous period.}
	InstPeriod = 0;
	Value4 = 0;
	For count = 0 to 50 begin
		Value4 = Value4 + DeltaPhase[count];
		If Value4 > 360 and InstPeriod = 0 then begin
			InstPeriod = count;
		end;
	end;

	{Resolve Instantaneous Period errors and smooth}
	If InstPeriod = 0 then InstPeriod = InstPeriod[1];
	Period = .25*InstPeriod + .75*Period[1];

	Plot1(Period, "DC");
EasyLanguage Code For Calculating Dominant Cycle

In my Python based back tester an indicator of this type is best programmed by using a class.¬† A class is really a simple construct, especially in Python, once you familiarize yourself with the syntax.¬† ¬†This indicator requires you to refer to historical values to calculate the next value in the equation:¬† Value1[4], inPhase[1], re[2], etc.,.¬† In EasyLanguage these values are readily accessible as every variable is defined as a BarArray – the complete history of a variable is accessible by using indexing.¬† In my PSB I used lists to store values for those variables most often used such as Open, High, Low, Close.¬† When you need to store the values of let’s say the last five bars its best to just create a list on the fly or build them into a class structure.¬† A Class stores data and data structures and includes the methods (functions) that the data will be pumped into.¬† The follow code describes the class in two sections:¬† 1) data declaration and instantiation and 2) the function to calculate the Dominant Cycle.¬† First off I create the variables that will hold the constant values: imult and qmult.¬† By using the word self I make these variables class members and can access them using “.” notation.¬† I will show you later what this means.¬† I also make the rest of the variables class members, but this time I make them lists and instantiate the first five values to zero.¬† I use list comprehension to create the lists and zero out the first five elements – all in one line of code.¬† This is really just a neat short cut, but can be used for much more powerful applications.¬† Once you create a dominantCycleClass object the object is constructed and all of the data is connected to this particular object.¬† You can create many dominantCycleClass objects and each one would maintain its own data.¬† Remember a class is just a template that is used to create an object.

class dominantCycleClass(object):
    def __init__(self):
        self.imult = 0.635
        self.qmult = 0.338
        self.value1 = [0 for i in range(5)]
        self.inPhase = [0 for i in range(5)]
        self.quadrature = [0 for i in range(5)]
        self.re = [0 for i in range(5)]
        self.im = [0 for i in range(5)]
        self.deltaPhase = [0 for i in range(5)]
        self.instPeriod = [0 for i in range(5)]
        self.period = [0 for i in range(5)]
Data Portion of Class

 

The second part of the class template contains the method or function for calculating the Dominant Cycle.¬† Notice how I index into the lists to extract prior values.¬† You will also see the word self. preceding the variable names used in the calculations.¬† Initially I felt like this redundancy hurt the readability of the code and in this case it might.¬† But by using self. I know I am dealing with a class member.¬† This is an example of the ” . ” notation I referred to earlier.¬† Basically this ties the variable to the class.

def calcDomCycle(self,dates,hPrices,lPrices,cPrices,curBar,offset):
        tempVal1 = (hPrices[curBar - offset] + lPrices[curBar-offset])/2
        tempVal2 = (hPrices[curBar - offset - 7] + lPrices[curBar-offset - 7])/2
        self.value1.append(tempVal1 - tempVal2)
        self.inPhase.append(1.25*(self.value1[-5] - self.imult*self.value1[-3]) + self.imult*self.inPhase[-3])        
        self.quadrature.append(self.value1[-3] - self.qmult*self.value1[-1] + self.qmult*self.quadrature[-2])
        self.re.append(.2*(self.inPhase[-1]*self.inPhase[-2]+self.quadrature[-1]*self.quadrature[-2])+ 0.8*self.re[-1])
        self.im.append(.2*(self.inPhase[-1]*self.quadrature[-2] - self.inPhase[-2]*self.quadrature[-1]) +.8*self.im[-1])
        if self.re[-1] != 0.0: self.deltaPhase.append(degrees(atan(self.im[-1]/self.re[-1])))
        if len(self.deltaPhase) > 51:
            self.instPeriod.append(0)
            value4 = 0
            for count in range(1,51):
                value4 += self.deltaPhase[-count]
                if value4 > 360 and self.instPeriod[-1] == 0:
                    self.instPeriod.append(count)
            if self.instPeriod[-1] == 0: self.instPeriod.append(self.instPeriod[-1])
            self.period.append(.25*self.instPeriod[-1]+.75*self.period[-1])
            return(self.period[-1])
Dominant Cycle Method

Okay we now have the class template to calculate the Dominant Cycle but how do we us it?

#---------------------------------------------------------------------------------
#Instantiate Indicator Classes if you need them
#---------------------------------------------------------------------------------
#    rsiStudy = rsiClass()
#    stochStudy = stochClass()
    domCycle = dominantCycleClass()
#---------------------------------------------------------------------------------
#Call the dominantCycleClass method using " . " notation.
	tempVal1 = domCycle.calcDomCycle(myDate,myHigh,myLow,myClose,i,0)
#Notice how I can access class members by using " . " notation as well!
	tempVal2 = domCycle.imult
Dominant Cycle Object Creation

Here I assign domCycle the object created by calling the dominantCycleClass constructor.¬† TempVal1 is assigned the Dominant Cycle when the function or method is called using the objects name (domCycle) and the now familiar ” . ” notation.¬† See how you can also access the imult variable using the same notation.

Here is the code in its entirety.  I put this in the indicator module of the PSB.

class dominantCycleClass(object):
    def __init__(self):
        self.imult = 0.635
        self.qmult = 0.338
        self.value1 = [0 for i in range(5)]
        self.inPhase = [0 for i in range(5)]
        self.quadrature = [0 for i in range(5)]
        self.re = [0 for i in range(5)]
        self.im = [0 for i in range(5)]
        self.deltaPhase = [0 for i in range(5)]
        self.instPeriod = [0 for i in range(5)]
        self.period = [0 for i in range(5)]

    def calcDomCycle(self,dates,hPrices,lPrices,cPrices,curBar,offset):
        tempVal1 = (hPrices[curBar - offset] + lPrices[curBar-offset])/2
        tempVal2 = (hPrices[curBar - offset - 7] + lPrices[curBar-offset - 7])/2
        self.value1.append(tempVal1 - tempVal2)
        self.inPhase.append(1.25*(self.value1[-5] - self.imult*self.value1[-3]) + self.imult*self.inPhase[-3])        
        self.quadrature.append(self.value1[-3] - self.qmult*self.value1[-1] + self.qmult*self.quadrature[-2])
        self.re.append(.2*(self.inPhase[-1]*self.inPhase[-2]+self.quadrature[-1]*self.quadrature[-2])+ 0.8*self.re[-1])
        self.im.append(.2*(self.inPhase[-1]*self.quadrature[-2] - self.inPhase[-2]*self.quadrature[-1]) +.8*self.im[-1])
        if self.re[-1] != 0.0: self.deltaPhase.append(degrees(atan(self.im[-1]/self.re[-1])))
        if len(self.deltaPhase) > 51:
            self.instPeriod.append(0)
            value4 = 0
            for count in range(1,51):
                value4 += self.deltaPhase[-count]
                if value4 > 360 and self.instPeriod[-1] == 0:
                    self.instPeriod.append(count)
            if self.instPeriod[-1] == 0: self.instPeriod.append(self.instPeriod[-1])
            self.period.append(.25*self.instPeriod[-1]+.75*self.period[-1])
            return(self.period[-1])
Dominant Cycle Class - Python