Algorithmic Trading Commentary, Research in books

Build your Technical/Trading Library for less than $150

Leave a comment

Great books that are dirt cheap!

These books cover a wide range of technical analysis and trading topics.  I have read and/or utilized ideas from these books for developing trading ideas and in help writing my own books.  I wanted to see what I could get for less than $13 a book.  I was pleasantly surprised at what I could get at Thriftbooks.com for this price or lower.  If you go a little higher in price, you will get some really great deals too.  Yes, these books are old, and I am sure I have left out some great names, but the bang for the buck is ridiculous.

Note – I captured the Add to Cart button in the IMAGES – IT DOES NOT WORK.  I wanted to show the various editions of each book.  Go to www.thriftbooks.com and search for the title or author and you should be able to find these books just like I did.  Also check out amazon.com

The Gallacher, Taleb, McKay, Schwager, and Livermore (Lefevre) books are fun reads that give an honest view of trading in general.  Elders’ book just needs to be any trader’s library.

The Edwards and Magee, Sklarew and Darvas books are classics.

Sunny Harris’ book is great for beginners.

The Stridsman and Rotella books are concrete methods to systematic trading.

Kaufman and McCormick/Katz are great reference books.

The Conway book has great EasyLanguage examples.

You can learn a lot from Schwager’s first book.

Learn optimization from Robert Pardo and money management from Fred Gehm’s classic.

If you go up in price a little bit, you will get Ralph Vince, Larry Williams, John Carter, Rober Carver, Andreas Clenow, and Ernie Chan.  If you are just starting out, you must have a trading library, and this would be a great start for the first shelf.

 

Algorithmic Trading Commentary, Complete EasyLanguage Systems, EasyLanguage, EasyLanguage equivalent to cool code, EasyLanguage Function, EasyLanguage Indicator, EasyLanguage Snippets, EasyLanguage Tutorial, EasyLanguage Tutorial, Free EasyLanguage, MultiCharts, Must Know, PortfolioTrader, Uncategorized

Should you use a profit taking algorithm in your Trend Following system?

Leave a comment

If letting profits run is key to the success of a trend following approach, is there a way to take profit without diminishing returns?

Most trend following approaches win less than 40% of the time.   So, the big profitable trades are what saves the day for this type of trading approach.  However, it is pure pain to simply sit there and watch a large profit erode, just because the criteria to exit the trade takes many days to be met.

Three methods to take a profit on a Trend Following algorithm

  1.  Simple profit objective – take a profit at a multiple of market risk.
  2.  Trail a stop (% of ATR) after a profit level (% of ATR) is achieved.
  3. Trail a stop (Donchian Channel) after a profit level (% of ATR) is achieved.

Use an input switch to determine which exit to incorporate

Inputs: initCapital(200000),rskAmt(.02),
useMoneyManagement(False),exitLen(13),
maxTradeLoss$(2500),
// the following allows the user to pick
// which exit to use
// 1:  pure profit objective
//     exit1ProfATRMult allows use to select
//     amount of profit in terms of ATR
// 2:  trailing stop 1 - the user can choose
//     the treshhold amount in terms of ATR
//     to be reached before trailing begins
// 3:  trailing stop 2 - the user can chose
//     the threshold amount in terms of ATR
//     to be reached before tailing begins
whichExit(1),
exit1ProfATRMult(3),
exit2ThreshATRMult(2),exit2TrailATRMult(1),
exit3ThreshATRMult(2),exit3ChanDays(5);
Exit switch and the parameters needed for each switch.

The switch determines which exit to use later in the code.  Using inputs to allow the user to change via the interface also allows us to use an optimizer to search for the best combination of inputs.  I used MultiCharts Portfolio Trader to optimize across a basket of 21 diverse markets.  Here are the values I used for each exit switch.

MR = Market risk was defined as 2 X avgTrueRange(15).

  • Pure profit objective -Multiple from 2 to 10 in increments of 0.25.  Take profit at entryPrice + or – Profit Multiple X MR
  • Trailing stop using MR – Profit Thresh Multiple from 2 to 4 in increments of 0.1.  Trailing Stop Multiple from 1 to 4 in increments of 0.1.
  • Trailing stop using MR and Donchian Channel – Profit Thresh Multiple from 2 to 4 in increments of 0.1.  Donchian length from 3 to 10 days.

Complete strategy code incorporating exit switch.  This code is from Michael Covel’s 2005 Trend Following book (Covel, Michael. Trend Following: How Great Traders Make Millions in Up or Down Markets. FT Press, 2005.)  This strategy is highlighted in my latest installment in my Easing into EasyLanguage series – Trend Following edition.


vars:buyLevel(0),shortLevel(0),longExit(0),shortExit(0);

Inputs: initCapital(200000),rskAmt(.02),
useMoneyManagement(False),exitLen(13),
maxTradeLoss$(2500),whichExit(1),
exit1ProfATRMult(3),
exit2ThreshATRMult(2),exit2TrailATRMult(1),
exit3ThreshATRMult(2),exit3ChanDays(5);

Vars: marketRisk(0), workingCapital(0),
marketRisk1(0),marketRisk2(0),
numContracts1(0),numContracts2(0);
 
//Reinvest profits? - uncomment the first line and comment out the second
//workingCapital = Portfolio_Equity-Portfolio_OpenPositionProfit;
workingCapital = initCapital;


buyLevel   = highest(High,89) + minMove/priceScale;
shortLevel = lowest(Low,89)   - minMove/priceScale;
longExit   = lowest(Low,exitLen)   - minMove/priceScale;
shortExit  = highest(High,exitLen) + minMove/priceScale;

marketRisk = avgTrueRange(15)*2*BigPointValue;
marketRisk1 =(buyLevel - longExit)*BigPointValue;
marketRisk2 =(shortExit - shortLevel)*BigPointValue; 
marketRisk1 = minList(marketRisk,marketRisk1);
marketRisk2 = minList(marketRisk,marketRisk2);

numContracts1 = (workingCapital * rskAmt) /marketRisk1;
numContracts2 = (workingCapital * rskAmt) /marketRisk2;

if not(useMoneyManagement) then
begin
    numContracts1 = 1;
    numContracts2 =1;
end;

numContracts1 = maxList(numContracts1,intPortion(numContracts1));  {Round down to the nearest whole number}
numContracts2 = MaxList(numContracts2,intPortion(numContracts1));


if c < buyLevel then buy numContracts1 contracts next bar at buyLevel stop;
if c > shortLevel then Sellshort numContracts2 contracts next bar at shortLevel stop;

buytocover next bar at shortExit stop;
Sell  next bar at longExit stop;

vars: marketRiskPoints(0);
marketRiskPoints = marketRisk/bigPointValue;
if marketPosition = 1 then
begin
	if whichExit = 1 then
		sell("Lxit-1") next bar at entryPrice + exit1ProfATRMult * marketRiskPoints limit;
	if whichExit = 2 then
		if maxcontractprofit > (exit2ThreshATRMult * marketRiskPoints ) * bigPointValue then
			sell("Lxit-2") next bar at entryPrice + maxContractProfit/bigPointValue - exit2TrailATRMult*marketRiskPoints stop;
	if whichExit = 3 then
		if maxcontractprofit > (exit3ThreshATRMult * marketRiskPoints ) * bigPointValue then
			sell("Lxit-3") next bar at lowest(l,exit3ChanDays) stop;
end;

if marketPosition = -1 then
begin
	if whichExit = 1 then
		buyToCover("Sxit-1") next bar at entryPrice - exit1ProfATRMult * marketRiskPoints limit;
	if whichExit = 2 then
		if maxcontractprofit > (exit2ThreshATRMult * marketRiskPoints ) * bigPointValue then
			buyToCover("Sxit-2") next bar at entryPrice - maxContractProfit/bigPointValue + exit2TrailATRMult*marketRiskPoints stop;
	if whichExit = 3 then
		if maxcontractprofit > (exit3ThreshATRMult * marketRiskPoints ) * bigPointValue then
			buyToCover("Sxit-3") next bar at highest(h,exit3ChanDays) stop;
end;

setStopLoss(maxTradeLoss$);

Here’s the fun code from the complete listing.

vars: marketRiskPoints(0);
marketRiskPoints = marketRisk/bigPointValue;
if marketPosition = 1 then
begin
	if whichExit = 1 then
		sell("Lxit-1") next bar at entryPrice + exit1ProfATRMult * marketRiskPoints limit;
	if whichExit = 2 then
		if maxContractProfit > (exit2ThreshATRMult * marketRiskPoints ) * bigPointValue then
			sell("Lxit-2") next bar at entryPrice + maxContractProfit/bigPointValue - exit2TrailATRMult*marketRiskPoints stop;
	if whichExit = 3 then
		if maxContractProfit > (exit3ThreshATRMult * marketRiskPoints ) * bigPointValue then
			sell("Lxit-3") next bar at lowest(l,exit3ChanDays) stop;
end;

if marketPosition = -1 then
begin
	if whichExit = 1 then
		buyToCover("Sxit-1") next bar at entryPrice - exit1ProfATRMult * marketRiskPoints limit;
	if whichExit = 2 then
		if maxContractProfit > (exit2ThreshATRMult * marketRiskPoints ) * bigPointValue then
			buyToCover("Sxit-2") next bar at entryPrice - maxContractProfit/bigPointValue + exit2TrailATRMult*marketRiskPoints stop;
	if whichExit = 3 then
		if maxContractProfit > (exit3ThreshATRMult * marketRiskPoints ) * bigPointValue then
			buyToCover("Sxit-3") next bar at highest(h,exit3ChanDays) stop;
end;

The first exit is rather simple – just get out on a limit order at a nice profit level.  The second and third exit mechanisms are a little more complicated.  The key variable in the code is the maxContractProfit keyword.  This value stores the highest level, from a long side perspective, reached during the life of the trade.  If max profit exceeds the exit2ThreshATRMult, then trail the apex by exit2TrailATRMult.  Let’s take a look at the math from a long side trade.

if maxContractProfit > (exit2ThreshATRMult * marketRiskPoints ) * bigPointValue

Since maxContractProfit is in dollar you must convert the exit2ThreshATRMult X marketRiskPoints into dollars as well.  If you review the full code listing you will see that I convert the dollar value, marketRisk, into points and store the value in marketRiskPoints.  The conversion to dollars is accomplished by multiplying the product by bigPointValue.

sell("Lxit-2") next bar at
entryPrice + maxContractProfit / bigPointValue - exit2TrailATRMult * marketRiskPoints stop;

I know this looks complicated, so let’s break it down.  Once I exceed a certain profit level, I calculate a trailing stop at the entryPrice plus the apex in price during the trade (maxContractProfit / bigPointValue) minus the exit2TrailATRMult X marketRiskPoints. If the price of the market keeps rising, so will the trailing stop.  That last statement is not necessarily true, since the trailing stop is based on market volatility in terms of the ATR.  If the market rises a slight amount, and the ATR increases more dramatically, then the trailing stop could actually move down.  This might be what you want.  Give the market more room in a noisier market.  What could you do to ratchet this stop?  Mind your dollars and your points in your calculations.

The third exit uses the same profit trigger, but simply installs an exit based on a shorter term Donchian channel.  This is a trailing stop too, but it utilizes a chart point to help define the exit price.

Results of the three exits

Exit 1 – Pure Profit Objective

Take a profit on a limit order once profit reaches a multiple of market risk aka 2 X ATR(15).

Pure profit object. Profit in terms of ATR or perceived market risk.

The profit objective that proved to be the best was using a multiple of 7.  A multiple of 10 basically negates the profit objective.   With this system several profit objective multiples seemed to work.

Exit – 2 – Profit Threshold and Trailing Stop in terms of ATR or market risk

Trail a stop a multiple of ATR after a multiple of ATR in profit is reached.

Trailing Stop using ATR
3-D view of parameters
3D view of parameter pairs

This strategy liked 3 multiples of ATR of profit before trailing and applying a multiple of 1.3 ATR as a stop.

Like I said in the video, watch out for 1.3 as you trailing amount multiple as it seems to be on a mountain ridge.

Exit – 3 – Profit Threshold in terms of ATR or market risk and a Donchain Channel trailing stop

Trail a stop using a Donchian Channel after a multiple of ATR in profit is reached.  Here was a profit level is reached, incorporate a tailing stop at the lowest low or the highest high of N days back.

Using Donchian Channel as trailing stop.
3-D view of parameters
3D view of parameters for Exit 3.

Conclusion

The core strategy is just an 89-day Donchian Channel for entry and a 13-Day Donchian Channel for exit.  The existing exit is a trailing exit and after I wrote this lengthy post, I started to think that a different strategy might be more appropriate.  However, as you can see from the contour charts, using a trailing stop that is closer than a 13-day Donchian might be more productive.   From this analysis you would be led to believe the ATR based profit and exit triggers (Exit #2) is superior.  But this may not be the case for all strategies.  I will leave this up to you to decide.  Here is the benchmark performance analysis with just the core logic.

Core logic results.

If you like this type of explanation and code, make sure you check at my latest book at amazon.com.  Easing into EasyLanguage – Trend Following Edition.

Complete EasyLanguage Systems, EasyLanguage, EasyLanguage Snippets, EasyLanguage Tutorial, EasyLanguage Tutorial, Free EasyLanguage, MultiCharts, Must Know

Buy in November, Sell in May Strategy Framework

2 Comments

Thanks to Jeff Swanson for the basis of this post

I like to post something educational at least once a month.  Sometimes, it’s difficult to come up with stuff to write about.  Jeff really got me thinking with his Buy November… post.  Check out his post “Riding the Market Waves:  How to Surf Seasonal Trends to Trading Success.”  Hopefully you have read his post and now have returned.  As you know, the gist of his post was to buy in November and sell in May.  Jeff was gracious enough to provide analysis, source and suggestions for improvement for this base strategy.

Why Change Jeff’s Code to a Framework?

I found Jeff’s post most intriguing, so the first think I start thinking about is how could I optimize the buy and sell months, a max loss, the three entry filters that he provided and in addition add a sell short option.  If you have read my books, you know I like to develop frameworks for further research when I program an algorithm or strategy.  Here is how I developed the framework:

  1. Optimize the entry month from January to December or 1 to 12.
  2. Optimize the exit month from January to December or 1 to 12.
  3. Optimize to go long or go short or 1 to 2 (to go short any number other than 1 really).
input: startMonth(11),endMonth(5),
       longOrShort(1),


currentMonth = Month(Date of tomorrow);
If currentMonth = startMonth and mp = 0 and entriesThisMonth = 0 Then
begin
// a trade can only occur if canBuy is True - start month is active as
// long as the filtering allows it.  Until the filter is in alignment
// keep looking for a trading during the ENTIRE startMonth
    if longOrShort = 1 and canBuy then
    	entriesThisMonth = 1; 
	if longOrShort = 1 and canBuy then
 		buy("Buy Month") iShares contracts next bar at market;
 	if longOrShort <> 1 and canShort then
 		sellShort("Short Month") iShares contracts next bar at market;
 	if longOrShort = -1 and canShort then
 		entriesThisMonth = 1;
end;
 
if CurrentMonth = endMonth Then
begin
	if longOrShort = 1 then
 		sell("L-xit Month") currentShares contracts next bar at market
 	else
 		buyToCover("S-xit Month") currentShares contracts next bar at market;
 end;
 
 if mp = 1 then 
 	sell("l-xitMM") next bar at entryPrice - maxTradeRisk/bigPointValue stop;
 if mp =-1 then
 	buyToCover("s-xitMM") next bar at entryPrice + maxTradeRisk/bigPointValue stop;
Snippet of the bones with extra flavor to enter and exit on certain months

You can see that I have provided the three inputs:

  1. startMonth
  2. endMonth
  3. longOrShort

I get the currentMonth by peeking at the date of tomorrow and passing this date to the month function.  If tomorrow is the first day of the month that I want to enter a long or short and the current market position (mp), and entriesThisMonth = 0, then a long or short position will be initiated.  If the filters I describe a little later allow it, I know that I will be executing a trade tomorrow, and I can go ahead assign a 1 to entries this month.  Why do I do this?  Just wait and you will see.   Long entries depend on the variable longOrShort being equal to 1 and the toggle canBuy set to True.  What is canBuy.  Just wait and you will see.  The sell short is similar, but conversely longOrShort needs to not equal 1.  In addition, canShort needs to be true too.

If the currentMonth = endMonth, then based on the market position a sell or a buy to cover will be executed.

How to add filters to Determine canBuy and canShort

inputs: 
useMACDFilter(1), MACDFast(9), MACDSlow(26), MACDAvgLen(9), MACDLevel(0),
useMAFilter(0), MALength(30),
useRSIFilter(0), RSILength(14), RSILevel(50)

RSIVal = rsi(close,RSILength);
MAVal = xAverage(close,MALength);
MACDVal = macd(close,MACDFast,MACDSlow);
MACDAvg = xAverage(MACDVal,MACDAvgLen);

if useMACDFilter = 1 then
begin
	canBuy = MACDVal > MACDLevel;
	canShort = MACDVal < MACDLevel;
end;

if useMAFilter = 1 then
begin
	canBuy = close > MAVal and canBuy;
	canShort = close < MAVal and canShort;
end;

if useRSIFilter = 1 then
begin
	canBuy = RSIVal > RSILevel and canBuy;
	canShort = RSIVal < RSILevel and canShort;
end;
Calculate Filter Components and then test them

You cannot optimize a True to False toggle, but you can optimize 0 for off and 1 for on.  Here the useFilterName inputs are initially set to 0 or off.  Each filter indicator has respective inputs so that the filters can be calculated with the user’s input.  If the filters are equal to one, then a test to turn canBuy and canShort to on or off is laid out in the code.  Each test depends on either the state of price compared to the indicator value, or the indicator’s relationship to a user defined level or value.

Will this code test all the combination of the filters?

Yes!  F1 is Filter 1 and F2 is Filter 2 and F3 is Filter 3.  By optimizing each filter from 0 to 1, you will span this search space.

  • F1 = On; F2 = Off; F3 = Off
  • F1 = On; F2 = On; F3 = Off
  • F1 = On; F2 = On; F3 = On
  • F1 = Off; F2 = On; F3 = Off
  • F1 = Off; F2 = On; F3 = On
  • F1 = Off; F2 = Off: F3 = On
  • F1 = On; F2 = Off; F3 = On

You will notice I initially set canBuy and canShort to True and then turn them off if an offending filter occurs.  Notice how I AND the results for Filter 2 and Filter 3 with canBuy or canShort.  Doing this allows me to cascade the filter combinations.  I do want to test when all filters are in alignment.  In other words, they must all be True to initiate a position.

Should the Filters be Active During the Entire Entry Month?

What if the first day of the month arrives and you can’t initiate a trade due to a conflict of one of the filters.  Should we allow a trade later in the entry month if the filters align properly?  If we are testing 25 years of history and allow for entry later on in the month, we could definitely generate as close to 25 trades as possible.   This line of code keeps the potential of a trade open for the entire month.


//  only set entriesThisMonth to true
// when all the stars align - might enter a long
// trade on the last day of the month

if longOrShort = 1 and canBuy then
    	entriesThisMonth = 1;
Keep the entire start month active

Some Tricky Code

I wanted to allow a money management exit on a contract basis.  I had to devise some code that would not allow me to reenter the startMonth if I got stopped out prematurely in the startMonth (the same month as entry.)

if entriesThisMonth = 1 and monthOfTomorrow <> startMonth then
	entriesThisMonth = 0;
This code resets entriesThisMonth

If a position is initiated, I know entriesThisMonth will be set to one.  If I enter into another month that is not the startMonth then entriesThisMonth is set to 0.  This prevents reentry in case we get stopped out in the same month we initially enter a position.  In other words, entriesThisMonth stays one until a new month is observed.  And we can’t enter when entriesThisMonth is equal to one.

Full Code

input: startMonth(11),endMonth(5),
       longOrShort(1),
       useMACDFilter(1),MACDFast(9),MACDSlow(26),MACDAvgLen(9),MACDLevel(0),
       useMAFilter(0),MALength(30),
       useRSIFilter(0),RSILength(14),RSILevel(50),
       startAccountSize(100000),
       marketRiskLen(30),
       riskPerTrade(5000),
       maxTradeRisk(5000);

vars: currentMonth(0),mp(0),iShares(0),entriesThisMonth(0),monthOfTomorrow(0),
      RSIVal(0),MAVal(0),MACDVal(0),MACDAvg(0),canBuy(True),canShort(True);

mp = marketPosition;
iShares = riskPerTrade/bigPointValue/avgTrueRange(marketRiskLen);

RSIVal = rsi(close,RSILength);
MAVal = xAverage(close,MALength);
MACDVal = macd(close,MACDFast,MACDSlow);
MACDAvg = xAverage(MACDVal,MACDAvgLen);

canBuy = True;
canShort = True;

mp = marketPosition;

monthOfTomorrow = month(date of tomorrow);

if entriesThisMonth = 1 and monthOfTomorrow <> startMonth then
	entriesThisMonth = 0;
	
if useMACDFilter = 1 then
begin
	canBuy = MACDVal > MACDLevel;
	canShort = MACDVal < MACDLevel;
end;

if useMAFilter = 1 then
begin
	canBuy = close > MAVal and canBuy;
	canShort = close < MAVal and canShort;
end;

if useRSIFilter = 1 then
begin
	canBuy = RSIVal > RSILevel and canBuy;
	canShort = RSIVal < RSILevel and canShort;
end;

currentMonth = Month(Date of tomorrow);
//print(d," ",currentMonth," ",startMonth," ",entriesThisMonth);
If currentMonth = startMonth and mp = 0 and entriesThisMonth = 0 Then
begin
//	print(d," ",currentMonth," ",canBuy);
    if longOrShort = 1 and canBuy then
    	entriesThisMonth = 1;
	if longOrShort = 1 and canBuy then
 		buy("Buy Month") iShares contracts next bar at market;
 	if longOrShort <> 1 and canShort then
 		sellShort("Short Month") iShares contracts next bar at market;
 	if longOrShort = -1 and canShort then
 		entriesThisMonth = 1;
end;
 
if CurrentMonth = endMonth Then
begin
	if longOrShort = 1 then
 		sell("L-xit Month") currentShares contracts next bar at market
 	else
 		buyToCover("S-xit Month") currentShares contracts next bar at market;
 end;
 
 if mp = 1 then 
 	sell("l-xitMM") next bar at entryPrice - maxTradeRisk/bigPointValue stop;
 if mp =-1 then
 	buyToCover("s-xitMM") next bar at entryPrice + maxTradeRisk/bigPointValue stop;
Complete Code Framework

Here is the best equity curve I uncovered when I optimized the startMonth from 1 to 12 and the endMonth from 1 to 12 and the maxTradeRisk per contract and the three entry filters.  Entering in November when the moving average filter aligns and exiting on the first day of August and risking $5,500 per contract produced this equity curve.

Enter November get out the beginning of August

The test returned what would basically be similar to a buy and hold scenario; the difference being you only hold the trade between seven and eight months of the year and risk only $5,500 per contract.  If you get stopped out, you wait until November to get back in – whenever the moving average filter allows.  Net profit to draw down ratio is north of 4.0.

Last Comment

If I optimize from 1 to 12 for the start month and 1 to 12 for end month, will this not cause an error?  What if the two values equal?  I mean I can’t enter and exit in the same month – a one-day trade?  You could make the code smarter, but it doesn’t matter.  As a user you will know better than to use the same number and the optimizer will test the combination with the same number, but the results will fall off the table.   In this case, error trapping doesn’t prevent a necessarily unwanted or dangerous scenario.

Algorithmic Trading Commentary, Complete EasyLanguage Systems, EasyLanguage, EasyLanguage equivalent to cool code, EasyLanguage Function, EasyLanguage Snippets, EasyLanguage Tutorial, EasyLanguage Tutorial, Free EasyLanguage, MultiCharts, Must Know, Must Know EasyLanguage, PortfolioTrader, Trend Following

Prune Your Trend Following Algorithm

Leave a comment

Multiple trading decisions based on “logic” may not add to the bottom line

In this post, I will present a trend following system that uses four exit techniques.  These techniques are based on experience and also logic.  The problem with using multiple exit techniques is that it is difficult to see the synergy that is generated from all the moving parts.  Pruning your algorithm may help cut down on invisible redundancy and opportunities to over curve fit.  The trading strategy I will be presenting will use a very popular entry technique overlaid with trade risk compression.

Entry logic

Long:

Criteria #1:  Penetration of the closing price above an 85 day (closing prices) and 1.5X standard deviation-based Bollinger Band.

Criteria #2:  The mid-band or moving average must be increasing for the past three consecutive days.

Criteria #3: The trade risk (1.5X standard deviation) must be less than 3 X average true range for the past twenty days and also must be less than $4,500.

Risk is initially defined by the standard deviation of the market but is then compared to $4,500. If the trade risk is less than $4,500, then a trade is entered. I am allowing the market movement to define risk, but I am putting a ceiling on it if necessary.

Short:

Criteria #1:  Penetration of the closing price below an 85 day (closing prices) and 1.5X standard deviation-based Bollinger Band.

Criteria #2:  The mid-band or moving average must be decreasing for the past three consecutive days.

Criteria #3:  Same as criteria #3 on the long side

Exit Logic

Exit #1:  Like any Bollinger Band strategy, the mid band or moving average is the initial exit point.  This exit must be included in this particular strategy, because it allows exits at profitable levels and works synergistically with the entry technique.

Exit #2:  Fixed $ stop loss ($3,000)

Exit #3:  The mid-band must be decreasing for three consecutive days and today’s close must be below the entry price.

Exit #4:  Todays true range must be greater than 3X average true range for the past twenty days, and today’s close is below yesterday’s, and yesterday’s close must be below the prior days.

Here is the logic of exits #2 through exit #4.  With longer term trend following system, risk can increase quickly during a trade and capping the maximum loss to $3,000 can help in extreme situations.  If the mid-band starts to move down for three consecutive days and the trade is underwater, then the trade probably should be aborted.  If you have a very wide bar and the market has closed twice against the trade, there is a good chance the trade should be aborted.

Short exits use the same logic but in reverse.  The close must close below the midband, or a $3,000 maximum loss, or three bars where each moving average is greater than the one before, or a wide bar and two consecutive up closes.

Here is the logic in PowerLanguage/EasyLanguage that includes the which exit seletor.

[LegacyColorValue = true]; 
Inputs:	maxEntryRisk$(4500),maxNATRLossMult(3),maxTradeLoss$(3000),
		indicLen(85),numStdDevs(1.5),highVolMult(3),whichExit(7); 
		
Vars: 	upperBand(0), lowerBand(0),slopeUp(False),slopeDn(False), 
		largeAtr(0),sma(0),
		initialRisk(0),tradeRisk(0),
		longLoss(0),shortLoss(0),permString("");

upperBand = bollingerBand(close,indicLen,numStdDevs); 
lowerBand = bollingerBand(close,indicLen,-numStdDevs);
largeATR = highVolMult*(AvgTrueRange(20));

sma = average(close,indicLen);

slopeUp = sma>sma[1] and sma[1]>sma[2] and sma[2]>sma[3];
slopeDn = sma<sma[1] and sma[1]<sma[2] and sma[2]<sma[3];

initialRisk = AvgTrueRange(20);
largeATR = highVolMult * initialRisk; 
tradeRisk = (upperBand - sma);
// 3 objects in our permutations 
// exit 1, exit 2, exit 3
// perm #     exit #
//      1          1
//      2        1,2
//      3        1,3
//      4          2
//      5        2,3
//      6          3
//      7      1,2,3

if whichExit = 1 then permString = "1";
if whichExit = 2 then permString = "1,2";
if whichExit = 3 then permString = "1,3";
if whichExit = 4 then permString = "2";
if whichExit = 5 then permString = "2,3";
if whichExit = 6 then permString = "3";
if whichExit = 7 then permString = "1,2,3";



{Long Entry:}
If (MarketPosition = 0) and
	Close crosses above upperBand and slopeUp and 
	(tradeRisk < initialRisk*maxNATRLossMult and tradeRisk<maxEntryRisk$/bigPointValue) then 
	begin
		Buy ("LE") Next Bar  at Market;
	End;


{Short Entry:}

If (MarketPosition = 0) and slopeDn and 
	Close crosses below lowerBand and
	(tradeRisk < initialRisk*maxNATRLossMult and tradeRisk<maxEntryRisk$/bigPointValue) then 
	begin
		Sell Short ("SE") Next Bar  at Market; 
	End;


{Long Exits:}

if marketPosition = 1 Then
Begin
	longLoss = initialRisk * maxNATRLossMult;
	longLoss = minList(longLoss,maxTradeLoss$/bigPointValue);
	
	If Close < sma then
			Sell ("LX Stop") Next Bar  at Market;;
	
	if inStr(permString,"1") > 0 then
		sell("LX MaxL") next bar at entryPrice - longLoss stop;

	if inStr(permString,"2") > 0 then
		If sma < sma[1] and sma[1] < sma[2] and sma[2] < sma[3] and close < entryPrice then
			Sell ("LX MA") Next Bar  at Market;
	if inStr(permString,"3") > 0 then
		If TrueRange > largeATR and close < close[1] and close[1] < close[2] then
			Sell ("LX ATR") Next Bar  at Market;
end;

{Short Exit:}

If (MarketPosition = -1) Then
Begin

	shortLoss = initialRisk * maxNATRLossMult;
	shortLoss = minList(shortLoss,maxTradeLoss$/bigPointValue);
	if Close > sma then
		Buy to Cover ("SX Stop") Next Bar  at Market;

	if inStr(permString,"1") > 0 then
		buyToCover("SX MaxL") next bar at entryPrice + shortLoss stop;

	if inStr(permString,"2") > 0 then
		If sma > sma[1] and sma[1] > sma[2] and sma[2] > sma[3] and  close > entryPrice then
			Buy to Cover ("SX MA") Next Bar  at Market;
	if inStr(permString,"3") > 0 then
		If TrueRange > largeAtr and close > close[1] and close[1] > close[2] then
			Buy to Cover ("SX ATR") Next Bar  at Market;
end;
Trend following with exit selector

Please note that I modified the code from my original by forcing the close to cross above or below the Bollinger Bands.  There is a slight chance that one of the exits could get you out of a trade outside of the bands, and this could potentially cause and automatic re-entry in the same direction at the same price.  Forcing a crossing, makes sure the market is currently within the bands’ boundaries.

This code has an input that will allow the user to select which combination of exits to use.

Since we have three exits, and we want to evaluate all the combinations of each exit separately, taken two of the exits and finally all the exits, we will need to rely on a combinatorial table.    In long form, here are the combinations:

3 objects in our combinations of exit 1, exit 2, exit 3

  • one  – 1
  • two  – 1,2
  • three  –  1,3
  • four –  2
  • five  – 2,3
  • six –  3
  • seven  –  1,2,3

There are a total of seven different combinations. Given the small set, we can effectively hard-code this using string manipulation to create a combinatorial table. For larger sets, you may find my post on the Pattern Smasher beneficial. A robust programming language like Easy/PowerLanguage offers extensive libraries for string manipulation. The inStr string function, for instance, identifies the starting position of a substring within a larger string. When keyed to the whichExit input, I can dynamically recreate the various combinations using string values.

  1. if whichExit = 1 then permString = “1”
  2. if whichExit = 2 then permString= “1,2”
  3. if whichExit = 3 then permString = “1,2,3”
  4.  etc…

As I optimize from one to seven, permString will dynamically change its value, representing different rows in the table. For my exit logic, I simply check if the enumerated string value corresponding to each exit is present within the string.

	if inStr(permString,"1") > 0 then
		sell("LX MaxL") next bar at entryPrice - longLoss stop;
	if inStr(permString,"2") > 0 then
		If sma < sma[1] and sma[1] < sma[2] and sma[2] < sma[3] and close < entryPrice then
			Sell ("LX MA") Next Bar  at Market;
	if inStr(permString,"3") > 0 then
		If TrueRange > largeATR and close < close[1] and close[1] < close[2] then
			Sell ("LX ATR") Next Bar  at Market;
Using inStr to see if the current whichExit input applies

When permString = “1,2,3” then all exits are used.  If permString = “1,2”, then only the first two exits are utilized.  Now all we need to do is optimize whichExit from 1 to 7.  Let’s see what happens:

Combination of all three exits

The best combination of exits was “3”.  Remember 3 is the permString  that = “1,3” – this combination includes the money management loss exit, and the wide bar against position exit.  It only slightly improved overall profitability instead of using all the exits – combo #7.  In reality, just using the max loss stop wouldn’t be a bad way to go either.  Occam uses his razor to shave away unnecessary complexities again!

If you like this code, you should check out the Summer Special at my digital store. I showcase over ten more trend-following algorithms with different entry and exit logic constructs.  These other algorithms are derived from the best Trend Following “Masters” of the twentieth century.  IMHO!

Here is a video you can watch that goes over the core of this trading strategy.

 

Algorithmic Trading Commentary, Python, TradingSimula18, VBA for Excel

Is your Big Point Value right?

Leave a comment

Thinking of using futures data from a vendor such as CSI, Pinnacle, Norgate…?

You would be surprised how many quants still use Excel or they own homegrown back-testing software.  Determining trade execution is simple.  On a stop order, all you need to do is see if today’s bar is higher than the prior bars calculated trade signal – right.  And if it is then a buy stop order is executed and the calculated signal price.  Store the entry price and when the exit occurs do the math to calculate the profit or loss.  Before you can do this, you must know two very important contract specs on the futures data you are testing:

  1. big point value – I will explain how to calculate this a little later
  2. minimum tick move

That’s it.  You don’t need to know anything beyond this to properly calculate trade signals and trade P and L.  As long as you store this information for each futures/commodity market you are testing, then you will be good to go.  Many testing platforms store more information such as contract size, trading hours, expiration date and several other contract specs.  You don’t need to know this information if you all you are interested is proper signal placement and accurate trade accounting.  Once you set the big point value and minimum tick move you are good to go, right?  Yes, if you stick with the same data vendor.  I kid you not.  Data vendors often quote the same exact market with different decimal places.  If you look up a sugar quote at the CME website, you will see it quoted like 0.1908.  If you look it up on TradeStation it is quoted as 19.08.  It’s the same size contract 112,000 pounds, but if you use the big point value at the CME, it will not produce accurate calculations on the TradeStation quote.  You can’t use the contract size of 112,000 pounds to help with the math either.  You have to delve into how each data vendor quotes their data and this will impact the big point move and minimum tick move.  Most vendors are gracious enough to provide a futures dictionary with these specs.  But it is wise to know how to do this by hand as well.

If you had to take a test, could you calculate the profit or loss from a trade in soybeans, in sugar or in euro currency?

Are you kidding me this is super simple.  Just program the strategy in TradeStation or Amibroker or Multicharts or TradingSimula-18.  I took the NFA Series 3 test more than two decades ago, and this type of question was on the test.  The following information was given:

  • contract size – 5000 bushels
  • minimum tick or move – 1/4th of a cent or penny
  • long entry price 8.32’2 and sell exit price 8.48’6 (per bushel)

The entry and exit price may look a little funny.  The entry price is eight dollars and 32 1/4 cents.   Many still quote soybeans in 1/8ths.  So, 2 X 1/8th = 1/4th.  For fun let’s calculate the notional value of the entry and exit prices.  The notional value of a soybean contract at the entry price is $8.3225 * 5000 = $41,612.50.  As you know futures are highly leveraged and you can control this quantity of soybeans for a small percentage of the notional value (margin.)  Now the exit price is $8.4875 (6*1/8th = 3/4th of a cent) and the notional value of the contract at exit is $8.4875 * 5000 = $42,437.5.  The result of the trade was $42,437.50 – $41,612.50 = $825.  Or you could simply multiple the difference between entry and exit by 5000.  This would $0.165 * 5000.  This makes perfect sense, but as a broker it was difficult to keep the contract size of a market and its minimum tick move in your memory.  Well, if you did it long enough it wasn’t that difficult.  You can reduce the math down to one easy to remember value and quickly do the math in your head.  The concept of the big point move allows for this.  If you download your data from Pinnacle or CSI the price of soybeans is usually quoted like this:  848.75.  This would be eight hundred and 48.75 cents.  The big point move is the amount of dollars required to lift (or drop) the first digit to the left of the decimal by one.  The first digit to the left of the decimal in beans is a penny or one cent.  A one cent move in beans is 5000 * $0.01 or $50.  Going back to our trade example 848.75 – 832.25 = 16.5 – multiply this by $50 you get $825.  You can also derive the minimum move dollar value too.  If the minimum move is 1/4th of a cent, then you can multiply 5000 * $.0025 and this yields $12.5.

Why is this important to know?

You should know this if you are trading the market, period.  Also, if you are a quant and are using some back testing software that requires you to set up the database outside the purview of the back-tester, then these values must be known, and must also be accurate.  Since TradeStation integrates its own data, you don’t have to worry about this.  But if you want to use a database from Pinnacle, BarChart, CSI or Norgate, then you have to take the time to set this up, right off the bat.  I feel like the leading data vendors, provide accurate data.  But there are differences from one vendor to another.  Not all data vendors use the same number of decimal places in their market quotes, so you must be able to determine the big point value and minimum move from the data.   You can do all the math to determine profit and loss from the big point value.  Here is a snapshot of a few markets in the TS-18 dataMasterPinnacle and the TS-18 dataMasterCSI text files. Like I said, all vendors will provide this information for you, so you can set your database properly.  Some back-testing platforms require more contract specs, but TS-18 just needs this information to carry out accurate calculations.  Here TS-18 wants to know the symbol, big point value, minimum tick move, and market name.

Pinnacle Data
AN,1000,0.01,Aus.Dol
ZL,600,0.01,BeanOil
BN,625,0.01,B.Pound
ZU,1000,0.01,Crude
ZC,50,0.25,Corn
CC,10,1,Cocoa
CL,1000,0.01,Crude
CN,1000,0.01,Can.Dol
CT,500,0.01,Cotton
FN,1250,0.005,EuroCur


CSI Data
AD,100000,0.0001,Aus.Dol
BO,600,0.01,BeanOil
BP,62500,0.0001,B.Pound
CL,1000,0.01,Crude
C2,50,0.25,Corn
C_,50,0.25,Corn
CC,10,1,Cocoa
CD,100000,0.0001,Can.Dol
CL,1000,0.01,Crude
CT,500,0.01,Cotton
CU,125000,0.00005,EuroCur
EC,125000,0.00005,EuroCur

Other than the symbol names the values for each symbol are very similar.  Most data vendors, including CSI quote the Euro currency like this:

1.08735 and a move to 1.08740 = 0.00005 *$125,000 = $6.25

But Pinnacle data quotes it like this:

108.735 and a move to 108.740 = 0.005 * $1,250 = $6.25

The size of the contract isn’t different here.  It is the quote that is different.  The big point value is tied to the contract size and the format of the quote.  This is why knowing the big point value is so important.  If you don’t set up the contract specifications correctly, then you will receive inaccurate results.

One more test in sugar – from the CME website

  • contract size – 112,000 pounds
  • minimum tick or move – $0.0001
  • long entry price $0.1910 and sell exit price $0.1980 (per pound)
  • big point value = $112,000
  • $0.0070 * 112,000 = $784

From TradeStation

  • contract size – 112,000 pounds
  • minimum tick or move – $0.01
  • long entry price $19.10 and sell exit price $19.80 (per pound??)
  • big point value = $1,120
  • $0.70 * 1,120 = $784

Most fractional quotes are delivered in decimal format and the also very important minimum tick move.

The interest rate futures have minimum tick moves ranging from 1/256ths to 1/32nds.  Most vendors will give you a quote like 120.03125 for the bonds.  If you test in TradersStudio or Excel or any other platform where the data is not integrated and are using CSI (or any other vendor), then you must force your calculated prices to fall on an even tick.  Assume you do a calculation to buy a bond future at 120.022365895.  If you don’t take the minimum tick move into considerations, you might be filled at this theoretical price.  In reality you should be filled on an even tick at 120.03125.  This is worse but realistic fill. You could create what you think is a really awesome strategy where you are shaving off the price on every trade – getting a better fill on every trade.

Thinking of purchasing and using back-testing software or Excel and data from a data vendor?

Become your own quant and do this.  You will learn a lot about algorithm testing and development.  But first things first.  Get your database set up according to your data vendor.  Once this chore is complete, testing becomes smooth sailing.  I provide the databases for Pinnacle and CSI with TS-18. Other software provides these as well, and a way to create your own databases.

Algorithmic Trading Commentary

Check out my interview in Traders’ Magazine [Germany]

Leave a comment

INTERVIEW WITH THE GERMANY BASED TRADERS´ Magazine 05.2024 (May 2024 Issue)

traders-mag.com

Note from George – Original text was written in German.  The text below is translated without modification.

George Pruitt was Director of Research at Futures Truth for 30 years.

During this time he checked the performance of more than 1000 trading strategies that were commercially available. He had contact with many prominent people in the algorithmic trading industry. He has also advised domestic and foreign customers and published a total of eight books. We talk to George Pruitt about his experiences during more than three decades of trading system development.

Traders’ media Gmbh – Eisenbahnstraße 53
97084 Würzbug

info@traders-mag.com

TRADERS´: YOU BEGAN WORKING FOR FUTURES TRUTH IN 1989, WHILE YOU WERE STUDENT. HOW DID THAT HAPPEN?

Pruitt: I was in my final year of college and looking for an internship. I saw an ad for Futures Truth in the newspaper. They were looking for a part-time programmer. After the interview with founder John Hill and programmer John Fisher, I got the job.

TRADERS´: FUTURES TRUTH HAS BEEN AN INSTITUTION IN THE INDUSTRY FOR A LONG TIME. WHAT DID THE PROVIDERS OF TRADING STRATEGIES THINK ABOUT THE PERFORMANCE OF THEIR APPROACHES BEING SUBJECT TO EXTERNAL VALIDATION AND THEREFORE MADE TRANSPARENT?

Pruitt: Many system providers used the results published in the magazine as a springboard to become leaders in the heyday of the trading systems industry.

Other providers were not so happy about their results being made available to the public. Particularly in the early days of system trading, only a few traders had the computing power, data or software to check the claims about the strategies made in the providers’ advertisements. Sometimes the results shown were not accurate at all. Futures Truth used a rigorous backtesting methodology, using tick data when necessary to create authentic track records. These sometimes contradicted the advertising material of the providers, who were of course not that enthusiastic about them.

TRADERS: WHAT SURPRISED YOU THE MOST WHILE TESTING ALL THESE STRATEGIES OVER THE YEARS?

Pruitt: I was surprised that some very expensive trading systems were not worth the price the customer paid for them. In fact, price had very little to do with success. One reason for this was that many strategy providers were not traders in the true sense, but rather engineers or scientists.

TRADERS’: IN THE 1990’S YOU TESTIFIED TO YOUR FEDERAL GOVERNMENT ON THE SUBJECT OF ALGORITHMIC TRADING. WHAT WAS IT ABOUT THEN?

Pruitt: In this particular case, I was asked to essentially answer the following question: “Can an algo that has low trading success have periods of success – and could it in the future-be more successful at all?” Of course, any trading system, good or bad, can have profit and loss margins. However, in my experience at the time, an algo with a negative expected value usually didn’t last long after a winning streak.

TRADERS: YOU SAID “ACCORDING TO MY EXPERIENCE AT THE TIME”. HAS YOUR OPINION LATER CHANGED?

Pruitt: Back then, markets were relatively inefficient. One might expect a solid backtest to be pretty indicative of future performance. In the meantime, however, the markets have evolved. Don’t get me wrong: all we can lean on is the backtest, and it will always stay that way. But I don’t think it’s as meaningful today as it was in the 1990s.

TRADERS: DO YOU REMEMBER A SUPERIOR SYSTEM THAT GOT INTO TROUBLE?

Pruitt: One example was Keith Fitschen and his “aberration” system. It was a breakout system based on Bollinger bands with a period length of 80 days and two standard deviations. The exit was at the moving average (middle band). In my opinion, it was one of the most successful trading systems of all time. From 2010 to 2020, however, the strategy went through a difficult phase. Keith believed that markets have become much more efficient. With his “Paradigm Shift” system, he basically turned the trading rules around, although not exactly. This resulted in it having a really poor performance in the backtest, but succeeding after that. But only temporarily. The success of the reverse approach did not last.

The graphic shows a comparison of Keith Fitschen’s original aberration system and his practically reversed paradigm shift strategy, each applied to crude oil. For a while, it seemed like the market had undergone a paradigm shift. But then the new strategy also disappointed.

This is what the capital curve of a typical trend-following strategy looks like across a portfolio of different futures contracts. There were big movements in 2008, 2010 and 2014. After that, it became more difficult

TRADERS: YOU ARE AN EXPERT IN THE DEVELOPMENT OF TRADING SYSTEMS. WHAT ABOUT YOUR PERSONAL TRADING?

I was two years into working at Futures Truth. It took this long to complete the CompuTrac-M conversion. Futures Truth was started by John Hill who also owned Hill Financial Group and Commodity Research. At this time the managed futures business was booming and I was asked to sit for the Series 3 exam and to help start trading Fisher and Hills algorithms. John Hill, in my opinion, was the greatest market technician of his time. He had several books on bar patterns and he exposed me to discretionary trading. I couldn’t see the same things he did on a chart. However, Fisher had canned many of John’s concepts and programmed them into Fortran. I am a programmer and a computer scientist first and foremost and could see how the ideas, expressed by code, could work and they did for many years. I am an algorithmic trader and will always be one. I never traded futures for my personal accounts due to the conflict of interests. I spend most of my time trading stocks and working on my first love – programming.

TRADERS: WHAT WAS IT LIKE BACK THEN? YOUR APPROACH TO STOCK TRADING?

Pruitt: It was the dot-com era. We simply bought everything and sold it again a little later, which had a certain dynamism. In doing so, we also saw the signs of the times and held back a lot before the end of the bubble.

Shown is the capital curve of a short-term mean reversion strategy on the Nasdaq E-mini future. It buys pullbacks and trades in the direction of the long-term moving average. However, the drawdowns can be significant. For example, there was a 30 percent setback from trade 98 to 107.

TRADERS: YOU LED THE BACK-OFFICE TEAM OF A $180 MILLION HEDGE FUND IN THE 1990S AND DEVELOPED THE CORE ALGORITHMS OF ANOTHER HEDGE FUND IN THE 2000S. HOW DID YOU MANAGE THE GREAT SUCCESS AND LATER DISILLUSIONMENT WITH TREND-FOLLOWING STRATEGIES?

Pruitt: In the early 2000s, trend-following strategies became more and more popular. Back then, the managers of funds of funds didn’t even look at you if you were trading discretionarily, with options or anything else to follow trends. I, too, had included trend following in our programs and had some success with it. But at some point the zenith was passed. After the last major movements in raw materials and the stock market in the early 2010s, many hedge funds and commodity trading advisors (CTAs) had to give up. A two-percent fixed fee and a 20 percent performance fee are great when you’re making money, but it’s very hard when you’re not. The High Watermark became an obstacle that was difficult to overcome. At the same time, small hedge funds and CTAs in particular had to spend far too much money in relative terms to comply with the now stricter regulations.

TRADERS: NEVERTHELESS, TREND FOLLOWING IS STILL ONE OF THE MOST WELL-KNOWN TRADING STRATEGIES TODAY. WHY IS THAT?

Pruitt: For one thing, timing isn’t that important. And on the other hand, you do what the commandments of trading tell us: be selective when you enter, let the profits run, give the market room to its noise, and limit the losses. Most markets lend themselves well to trend following. And many of these systems have few parameters, so that over-optimization is hardly possible. This is also important.

TRADERS: WHAT DOES YOUR OWN TRADING APPROACH LOOK LIKE TODAY?

Pruitt: I trade a small portfolio of stocks on the daily chart. In doing so, I rely on both breakouts and mean reversion. I overlay levels on a 5-minute chart using Camarilla pivot points, for example, in order to systematically find entry levels. I derive the corresponding levels from the daily chart. The decisive factors here are indications that speak for a breakout or even a countercyclical trade. In these places, the risk can also be kept low. I also rely on the pyramiding of positions. However, my portfolio only contains stocks that I know and that I would hold for the long term.

TRADERS: IT IS SAID THAT AMONG DISCRETIONARY TRADERS ARE BOTH THE BEST AND THE WORST OF ALL. HOW MANY SUCCESSFUL DISCRETIONARY TRADERS THERE ARE IN THEIR EXPERIENCE?

Pruitt: Discretionary traders make a lot of money – but also lose it. For me, that’s not the way to go. I’ve met a lot of discretionary traders. But not many have been successful in the long run. Most eventually turn to algorithms – either as complete trading systems or as tools.

TRADERS: IT IS ALSO SAID THAT DISCRETIONARY TRADERS ALSO FOLLOW A SYSTEM. THE ONLY DIFFERENCE IS THAT THEY HAVE NOT YET SUFFICIENTLY SPECIFIED THEIR (UNCONSCIOUS) RULES. WOULD YOU AGREE WITH THAT?

Pruitt: Absolutely. We have a pattern in our head that we follow. You can’t just trade based on a gut feeling and be successful with it.

TRADERS: SYSTEMATIC TRADING HAS, BUT THERE ARE ALSO DOWNSIDES. WHAT ARE THEY?

Pruitt: It’s very challenging in a number of ways. Of course, if you’re winning non-stop with a system, it’s great. But what if ten losses occur in a row? At this point, at the latest, you start to doubt. To a certain extent, you have to enter the market with a blunt instrument and hope that your diligence in system development will pay off in the end. Remember, no matter what type of trader you are, your worst drawdown is yet to come. That’s why I believe that a multi-strategy approach works best. The hope, of course, is that diversification will improve earnings. In the end we were trading five different systems on almost the same portfolio.

TRADERS: THE MARKETS ARE CONSTANTLY CHANGING. IN 2001 YOU WROTE ABOUT IT, WHICH, FROM THEIR POINT OF VIEW, ARE THE BEST SYSTEMS FOR SHORT-TERM TRADING. HOW HAS THIS DEVELOPED OVER THE LAST TWO DECADES?

Pruitt: The best short-term swing systems in 2001 were based on breaking out of the opening range. The traders made a lot of money with this very simple approach. Toby Crabel wrote one of the best books about it. With the disappearance of the pits, the opening range was diluted. That said, I believe it still works for stocks and stock indices. Together with system developer Murray Ruggiero, we developed the Dynamic Open approach. The opening of the trading day is not based on time, but on a constant increase in volume. This also worked for a while, but unfortunately not permanently. Currently, the best short-term systems are based on mean reversion approaches.

TRADERS: CAN YOU GIVE US AN EXAMPLE?

Pruitt: You can use short-term Bollinger Bands, the Relative Strength Index, or anything else that shows a move away from the average. Buying pullbacks can be difficult because you don’t know where to end up in the downcycle.

 

TRADERS: SPEAKING OF MARKETS THAT CHANGE OVER TIME, HOW LONG DO YOU THINK IS THE IDEAL PERIOD FOR BACKTESTING?

Pruitt: The markets have undergone a change in strategy. It used to be that the more data, the better. But now I think that old data is affecting the validity of the results. A test period of ten to 15 years is more than sufficient these days.

TRADERS: ONE OF THE CHALLENGES OF BACKTESTING IS TO HAVE ENOUGH HIGH-QUALITY DATA IN THE FIRST PLACE. WHAT DO YOU WORK WITH? IS IT POSSIBLE TO ACHIEVE GOOD RESULTS EVEN WITH FREE DATA?

TO ACHIEVE THE BEST POSSIBLE RESULTS?

Pruitt: You can work with free data. This means that a positive expected value can be derived from it. But I would never trade it. Yahoo Finance and Quandl offer some free data, and good daily data is also relatively cheap. I’m using Pinnacle Data for commodities. They are very reasonably priced and very good. The same goes for CSI Data. I also test with Trade Station and their data. The integration of data, testing and trading in one package like TradeStation or MultiCharts is the main reason why I wrote my book series on EasyLanguage. Despite its limitations, it is probably the best programming language for beginners.

This capital curve looks too good to be true. That’s why caution is advised when it comes to data mining. However, it can also be the birth of a trading idea – used consciously – by optimizing all possible entry points and holding periods. Every once in a while, you’ll find something that works and can withstand rigorous testing.

TRADERS ́: WHAT ABOUT THE OLD IDEA OF DEVELOPING THE WORST POSSIBLE TRADING STRATEGY AND THEN JUST TRADING THE OPPOSITE?

Pruitt: That can work. You need to keep an eye on the average trade metric. If it’s high (big avgerage loss), why not? Often, bad systems are the ones that can’t cover their trading costs, and that’s a function of trade frequency. So a bad system that trades too much will always be a bad system.

TRADERS ́: WHAT INDICATORS WERE YOUR FAVORITES OVER THE YEARS?

Pruitt: I like Bollinger Bands, Keltner Channels, Moving Averages, and the MACD. Through Murray Ruggiero, I was introduced to John Ehlers’ indicators such as the Laguerre RSI and the Dominant Cycle. RSI and stochastic are good for the entry point, but not necessarily for making the entry decision.

TRADERS: DAY TRADING STRATEGIES ARE VERY POPULAR. HOWEVER, MANY PEOPLE UNDERESTIMATE HOW DIFFICULT IT IS TO EARN MONEY IN THE LONG RUN. WHAT ARE THE BIGGEST CHALLENGES?

Pruitt: Day trading can be successful, but you can’t trade too many times a day, and you can’t trade every day. The algorithms need appropriate filter rules. However, it’s hard for many traders to skip a day if they’re day traders. You have to wait for a setup. Waiting is perhaps the hardest part.

TRADERS: DO YOU HAVE ANY TIPS FOR DAY TRADERS THAT CAN BE USED TO IMPROVE RESULTS?

Pruitt: Identify support and resistance levels and trade around those levels. Be selective about which trades you make and don’t trade more than three times a day. In addition, filters should be developed to avoid bad trades in the first place.

TRADERS: WHAT ARE THE ANOMALIES IN RETURNS? THAT STILL EXIST TODAY?

Pruitt: There are still patterns in the market. They are a reflection of human nature. And there are persistent anomalies that can be detected with data mining tools. One example is the turn-of-the-month strategy. For day traders, there are specific times on the trading day that are suitable for long or short trades.

And the aforementioned open range breakout, which I still believe in, as well as the trade in the opposite direction in case the initial setup fails.

TRADERS: NO STRATEGY ALWAYS YIELDS PROFITS. THAT’S WHY SYSTEMATIC TRADERS NEED DIFFERENT APPROACHES THAT COMPLEMENT EACH OTHER. CAN YOU SHOW US A COMBINATION TO START WITH?

Pruitt: I would combine a trend-following approach with mean reversion and day trading. The trend-following portfolio should be the largest.

Mean Reversion only works in a handful of markets, and you should test this to see which markets are in this mode. When day trading, you should stick to markets with high volume and high volatility. As a rule, these are stocks and stock indices.

TRADERS: ONE CONTROVERSIAL ELEMENT OF STRATEGY DEVELOPMENT IS OPTIMIZATION. HOW MUCH DO YOU THINK SHOULD BE OPTIMIZED?

AND WITH WHAT APPROACH?

Pruitt: At the end of the day, all system developers are optimizing. A common question is whether you should use different parameters for different markets. When I was still working with managed futures, we didn’t allow for different parameters for different markets. We programmed concepts that were functions of the market, such as the Average True Range (ATR).

However, I believe that different parameters can work as long as you don’t overdo it, for example by optimizing individual parameters over too large a space. Data mining using genetic optimization can also be useful. On the other hand, I don’t believe in walk-forward optimization. This means that you are always behind the curve.

TRADERS: ONE METHOD OF OPTIMIZATION IS TO FILTER OUT THE LESS ATTRACTIVE SIGNALS WITHOUT COMPROMISING THE GOOD ONES. IS IT POSSIBLE TO IMPLEMENT THIS CONSCIENTIOUSLY?

Pruitt: Often, losing signals go hand in hand with a certain type of market behavior. In this case, filtering is the best approach to improve.

Let’s take a day trading system that trades every day but makes a loss overall. If you add that it only trades after days with a tight trading range, the performance of the system often increases.

TRADERS: YOU HAVE EXTENSIVE PROGRAMMING SKILLS IN DIFFERENT LANGUAGES. WHICH OF THEM ARE FOR YOU TODAY

MOST USEFUL?

Pruitt: Definitely Python and EasyLanguage. The latter is more or less identical to PowerLanguage from MultiCharts. Most trading strategies can be programmed in EasyLanguage to test and generate signals. Python allows you to test ideas outside of EasyLanguage, mainly because of the data limitations and access to academic libraries.

TRADERS: TODAY, ARTIFICIAL INTELLIGENCE (AI) IS ON EVERYONE’S LIPS. YOU HAVE ALREADY DEALT WITH SIMILAR IDEAS A FEW YEARS AGO.

IN WHICH AREAS CAN AI HELP?

Pruitt: We were already working with AI in the 1990s. She can help the trader develop the code and explain the huge library of indicators. Genetic optimization is helpful to some extent. As far as machine learning goes, I haven’t had any personal experience, but I’ve seen some disappointing results.

TRADERS: WITH YOUR KNOWLEDGE OF MORE THAN 1000 DIFFERENT ALGORITHMS, YOUR PROGRAMMING SKILLS AND ALL THE DATA AT YOUR DISPOSAL, DO YOU HAVE THE HOLY GRAIL FOUND?

Pruitt: In quantum circles, I was known as a “bubble brusher.” Often, developers and traders tested with table calculations and didn’t realize that they were cheating by unconsciously using future information. It’s easy to do. For example, you make the decision to trade at the open, but you accidentally use the daily close in your calculations for the signal. Most test platforms nowadays prevent such leaks, in the case of EasyLanguage, for example, through the “next bar” paradigm. However, if you have multiple signals that can be triggered on the same day, it is difficult to determine which one actually took place from a back testing perspective. In other words, during the back test if you need to know which long position was initiated from the various potential trade signals, the next bar paradigm makes it difficult to ascertain this.

TRADERS’: BASED ON YOUR DECADES OF EXPERIENCE, WHAT DO YOU RECOMMEND THAT ASPIRING TRADERS SHOULD START WITH?

Pruitt: John Hill’s most famous saying was, “Trading is the hardest way to make easy money!” Those of us who were able to act in the 1980s, 1990s, and a few years after were gifted. Arbitrage opportunities abounded at the time. John often said that trading bonds in those days was as easy as shooting fish in an aquarium. Today, on the other hand, it is very difficult to trade successfully. We have better instruments, but the market is more efficient. Countless market participants, access to data and fast computers have robbed us of easy profit opportunities. I would definitely recommend using an algorithm, and not trading too frequently. Don’t worry about every little movement in the market. Sooner or later, a bigger one will inevitably come. And be prepared to change your preconceived notions. Don’t try to force anything.

Infobox: Toby Crabel

Toby Crabel is a successful tennis player, US futures trader and fund manager at the hedge fund Crabel Capital Management, which he founded. In 1990, he published the book "Day Trading With Short Term Price Patterns and Opening Range Breakout", which is now considered a collector's item. On Amazon, more than 500 euros are called up, on the US site even over 1000 US dollars.

Source: Wikipedia

Infobox: Pivot-Punk Clique

Camarilla pivot points are a modified version of the classic pivot points. Introduced in 1989 by Nick Scott, a successful bond trader, they indicate potential intraday support and resistance levels. Similar to classic pivot points, they use the previous day's high, low, and close. However, Fibonacci numbers are used in the calculation of levels.

What: https://www.babypips.com
Complete EasyLanguage Systems, EasyLanguage, EasyLanguage Function, EasyLanguage Snippets, EasyLanguage Tutorial, Free EasyLanguage, MultiCharts, Must Know, Must Know EasyLanguage

Multi-Agents and the Power of the Series Function

Leave a comment

Jeff Swanson wrote a great post on multi-agent trading a few years ago.

Jeff created a simple mean reversion system and then created two derivatives that culminated in three systems (or three agents.)  Using Murray Ruggiero’s Equity Curve Feedback, he was able to poll which system was doing the best, synthetically, and execute the strategy that showed the best performance.   If memory serves, picking the highflyer turned out to be the way to go.  Jeff had just touched the surface of Murray’s tool. but it definitely did the job.  Murray contracted me to fix some problems with the ECF tool and I did, but the tool is just way too cumbersome, resource hungry and requires a somewhat higher level of EasyLanguage knowledge to be universally applicable.  I was doing similar research in the area of polling multiple strategies and picking the best, just like Jeff did, and just executing that one system, when I thought about this post.  Traders do this all the time.  They have multiple strategies in the pipeline and monitor the performance and if one is head and shoulders better than what they are currently trading, they will switch systems.  This was one of the side benefits of the ECF tool.

What is an agent

An agent is any trading system that produces a positive expectancy.  Using multiple agents in a polling process allows a trader to go with the strategy that is currently performing the best.  This sounds reasonable, but there are pitfalls.  You could always be behind the curve – picking the best system right before it has its draw down.  Agents can be similar, or they can be totally different types of systems.  I am going to follow in Jeff’s footsteps and create three agents with the same DNA.  Here is what I call the AgentSpawner strategy.

inputs: movAvgLen(200),numDownDays(3),numDaysInTrade(2),stopLoss(5000);


value1 = countIf(c < c[1],numDownDays);
if c > average(c,movAvgLen) and value1 = numDownDays then
	buy next bar at open;
if barsSinceEntry = numDaysInTrade then
	sell next bar at open;
if marketPosition = 1 then sell next bar at entryPrice - stopLoss/bigPointValue stop;
Use this template and optimize inputs to spawn new agents

This code trades in the direction of the longer-term moving average and waits for a pull back on N consecutive down closes.  I am using the neat function countIF. This function counts the number of times the conditional test occurs in the last N bars.  If I want to know the number of times I have had a down close in the last 3 days, I can use this function like this.

Value1 = countIF(c<c[1],3);

// this is what the function doues
// 1.) todays close < yesterdays close + 1
// 2.) yesterdays close < prior days close + 2
// 3.) day before yesterdays close < prior cays close + 3

//  If value3 = 3 then I know I had three conscecutive down 
//  closes.  If value3 is less than three then I did not.

If the close is greater than the longer-term moving average and I have N consecutive down closings, then I buy the next bar at the open.  I use a wide protective stop and get out after X bars since entry.  Remember EasyLanguage does not count the day of entry in its barsSinceEntry calculation.  I am not using the built-in setStopLoss as I don’t want to get stopped out on the day of entry.  In real trading, you may want to do this, but for testing purposes my tracking algorithm was not this sophisticated.  I spawned three agents with the following properties.

	Case 1: //Agent 1
		movAvgLen = 200;
		numDownDays = 2;
		numDaysInTrade = 15;
		stopLoss = 7500;
	Case 2:  //Agent 2
		movAvgLen = 140;
		numDownDays = 3;
		numDaysInTrade = 9;
		stopLoss = 2500;
	Case 3: //Agent 3
		movAvgLen = 160;
		numDownDays = 3;
		numDaysInTrade = 15;
		stopLoss = 2500;

System Tracking Algorithm

This is why I love copy-paste programming.  This can be difficult if you don’t know your EasyLanguage or how TradeStation processes the bars of data.  Get educated by checking my books out at amazon.com – that is if you have not already.  This code is a very simplistic approach for keeping track of a system’s trades and its equity.

value4 = countIf(c<c[1],4);
value3 = countIf(c<c[1],3);
value2 = countIf(c<c[1],2);
//Agent #1 tracking algorithm
if sys1Signal<> 1 and c[1] > average(c[1],160) and value2[1] = 2 then
begin
	sys1Signal = 1;
	sys1BarCount = -1;
	sys1TradePrice = open;
	sys1LExit = open - 7500/bigPointValue;
end;
if sys1Signal = 1 then
begin
	sys1BarCount+=1;
	if low < sys1LExit and sys1BarCount > 0 then
	begin
		sys1TradePrice = sys1LExit;
		sys1Signal = 0;
	end;
	if sys1BarCount = 16 and sys1Signal = 1 then
	begin
		sys1TradePrice = open;
		sys1Signal = 0;
	end;
end;
Yes, this looks a little hairy, but it really is simple stuff

I am pretending to be TradeStation here.  First, I need to test to see if Agent#1 entered into a long position.  If the close of yesterday is greater than the moving average, inclusive of yesterdays close, and there has been two consecutive down closes, then I know a trade should have been entered on todays open.  EasyLanguage’s next bar paardigm cannot be utilized here.  Remember I am not generating signals, I am just seeing if today’s (not tomorrows or the next bars) trading action triggered a signal and if so, I need to determine the entry/exit price.  I am gathering this information so I can feed it into a series function.  If a trade is triggered, I set four variables:

  1. sys1Signal – 1 for long, -1 for short, and 0 for flat.
  2. sys1BarCount – set to a -1 because I immediately increment.
  3. sys1TradePrice – at what price did I enter or exit
  4. sys1LExit – set this to our stop loss level

If I am theoretically long, remember we are just tracking here, then I need to test, starting with the following day, if the low of the day is below our stop loss level and if it is I need to reset two variables:

  1. sys1TradePrice – where did I get out
  2. sys1Signal – set to 0 for a flat position

If not stopped out, then I start counting the number of bars sys1Signal is equal to 1.  If sys1BarCount = 16, then I get out at the open by resetting the following variables:

  1. sys1TradePrice = open
  2. sys1Signal = 0

If you look back at the properties for Agent#1 you will see I get out after 15 days, not 16.  Here is where the next bar paradigm can make it confusing.  The AgentSpawner strategy says to sell next bar at open when barsSinceEntry = 15.  The next bar after 15 is 16, so we store the open of the 16th bar as our trade price.

Now copy and paste the code into a nice editor such as NotePad++ or NotePad and replace the string sys1 with sys2.  Copy the code from NotePad++ into your EasyLanguage editor.  Now back to NotePad++ and replace sys2 with sys3.  Copy that code into the EL edition too.  Now all you need to do is change the different properties for each agent and you will have three tracking modules.

The Power of the EasyLanguage Series Function

The vanilla version of EasyLanguage has object-oriented nuances that you may not see right off the bat.  In my opinion, a series function is like a class.  Before I get started, let me explain what I mean by series.  All EasyLanguage function are of three types.

  1. simple – like a Bollinger band calculation
  2. series – like we are talking about here
  3. auto-detect – the interpreter/compiler decides

The series function has a memory for the variables that are used within the function.  Take a look at this.

input: funcID(string),seed(numericSimple);

vars: count(0);
if barNumber = 1 then // on first bar seed count
	count = seed;
count = count+1;
print(d," ",funcID," ", count);
SeriesFunctionTest = count;
Count is class-like member

On the first bar of the function call – remember it will be called on each bar in the chart, the function variable count is assigned seed. Seed will be ignored on subsequent bars.   What makes this magical is that no matter how many times you call the function on the same bar it remembers the internal variables on somewhat of a hierarchical basis (each call remembers its own stuff.)  It like a class in that it gets instantiated on the very first call.  Meaning if you call it three times on the first bar of the data, you will have three distinct internal variable memories.  Take a look at my sandbox function driver and its output.

result = SeriesFunctionTest("Call #1",50);
result = SeriesFunctionTest("Call #2",5);
result = SeriesFunctionTest("Call #3",100);

//outPut

1170407.00 Call #1   51.00  //first bar  51 = seed + count + 1
1170407.00 Call #2    6.00  //first bar   6 = seed + count + 1
1170407.00 Call #3  101.00  //first bar 101 = seed + count + 1
	
1170410.00 Call #1   52.00 // second bar it remembered count was 51
1170410.00 Call #2    7.00 // second bar it remembered count was 6
1170410.00 Call #3  102.00 // second bar it remembered count was 101
	
1170411.00 Call #1   53.00 // you have a unique function values that
1170411.00 Call #2    8.00 // were instantiated on the first bar
1170411.00 Call #3  103.00 // of the test.
	
1170412.00 Call #1   54.00
1170412.00 Call #2    9.00
1170412.00 Call #3  104.00
Series functions rock - but they are resource hungry

Why is this important?

I have created a PLSimulator function that keeps track of the three agent’s performance.  I need the profit or loss to stick with each function and then also add or subtract from it.  This is a neat function.  Remember if you like this stuff buy my books at Amazon.com.

//ProfitLoss Simulator
Inputs:  signal(numericseries),tradePrice(numericSimple),orderType(numericSimple),useOte(Truefalse);
Vars:dmode(0),LEPrice(-99999),LXPrice(-99999),SEPrice(-99999),SXPrice(-99999);
vars: GProfit(0),OpenProfit(0);
vars: modTradePrice(0);

vars: printOutTrades(True);

modTradePrice = tradePrice;

if orderType = 1 then // stop order
begin
	if signal = 1 or (signal = 0 and signal[1] = - 1) then
		modTradePrice = maxList(open,modTradePrice);
	if signal = -1 or (signal = 0 and signal[1] = 1) then
		modTradePrice = minList(open,modTradePrice);
end;

if orderType = 2 then // limit order
begin
	if signal = 1 or (signal = 0 and signal[1] = - 1) then
		modTradePrice = minList(open,modTradePrice);
	if signal = -1 or (signal = 0 and signal[1] = 1) then
		modTradePrice = maxList(open,modTradePrice);
end;
if orderType = 3 then // market order
begin
	modTradePrice = open;
end;
		
If Signal[0]=1 And (Signal[1]=-1 Or Signal[1]=0) Then
begin
	LEPrice=modTradePrice; 
	SXPrice = -999999; 
	condition1 = false;
	If Signal[1]=-1 Then
	begin
    	SXPrice=modTradePrice;  
    	GProfit=(SEPrice-SXPrice)+GProfit;
    	condition1 = True;
    End;
    if not(condition1) then
    	if printOutTrades then Print(d," L:Entry ",LEPrice)
    else
    	if printOutTrades then Print(d," L:Entry ",LEPrice," ",(SEPrice-SXPrice)*bigPointValue:8:2," ",GProfit*bigPointValue:9:2);
End;
{('***********************************************}
If Signal[0]=-1 And (Signal[1]=1 Or Signal[1]=0) Then
begin
	SEPrice=modTradePrice;  
	LXPrice = 999999;
	condition1 = false;
    If Signal[1]=1 Then
    begin
    	condition1 = True;
      	LXPrice=modTradePrice;  
    	GProfit=(LXPrice-LEPrice)+GProfit;
    End;
    if not(condition1) then
    	if printOutTrades then Print(d," S:Entry ",SEPrice)
    else
    	if printOutTrades then Print(d," L:Exit ",LXPrice," ",(LXPrice-LEPrice)*bigPointValue:8:2," ",GProfit*bigPointValue:9:2);

End;
If Signal[0]=0 And Signal[1]=-1 Then
begin
	SXPrice = modTradePrice; 
    GProfit=(SEPrice-SXPrice)+GProfit;
    if printOutTrades then Print(d," S:Exit ",SXPrice," ",(SEPrice-SXPrice)*bigPointValue:8:2," ",GProfit*bigPointValue:9:2);

end;
If Signal[0]=0 And Signal[1]=1 Then
begin
	LXPrice = modTradePrice;
    GProfit=(LXPrice-LEPrice)+GProfit;
   if printOutTrades then Print(d," L:Exit ",LXPrice," ",(LXPrice-LEPrice)*bigPointValue:8:2," ",GProfit*bigPointValue:9:2);

end;

If Signal[1]=1 And useOte=True Then
begin
	OpenProfit=(Close[1]-LEPrice);
End;
If Signal[1]=-1 and useOte=True Then
begin
	OpenProfit=(SEPrice-Close[1]);
End;
If Signal[1]=0 Or useOte=False Then
begin
	OpenProfit=0;
End;

PLSimulator=(GProfit+OpenProfit)*bigpointvalue;
Simulate profit and loss and more importantly keep track of it

Feed tracker algorithm data into the function

Your information must be properly assigned to get this to work.  First, I show how to get the information into the function.  The function does all the work and returns the equity.  I then determine the best agent by looking at the ROC over the past thirty days of equity for each agent and pick the very best.  I then trade the very best.  This is a very quick application of the function.  I will have a more sophisticated function, something akin to Murray’s ECF but with much less overhead and more strategy templates.

sys1Equity = PLSimulator(sys1Signal,sys1TradePrice,1,True);
sys2Equity = PLSimulator(sys2Signal,sys2TradePrice,1,True);
sys3Equity = PLSimulator(sys3Signal,sys3TradePrice,1,True);


vars: multiAgent(0);

value1 = maxList(sys1Equity-sys1Equity[30],sys2Equity-sys2Equity[30],sys3Equity-sys3Equity[30]);
multiAgent = 1;
if sys2Equity-sys2Equity[30] = value1 then multiAgent = 2;
if sys3Equity-sys3Equity[30] = value1 then multiAgent = 3;

	
{print(d," ",sys1Equity-sys1Equity[30]," ",sys1Equity);
print(d," ",sys2Equity-sys2Equity[30]," ",sys2Equity);
print(d," ",sys3Equity-sys3Equity[30]," ",sys2Equity);}
print(d," MultiAgent ",multiAgent);

//system parameters
vars: movAvgLen(200),numDownDays(3),numDaysInTrade(2),stopLoss(5000);
Switch ( multiAgent )
Begin
	Case 1: 
		movAvgLen = 200;
		numDownDays = 2;
		numDaysInTrade = 15;
		stopLoss = 7500;
	Case 2: 
		movAvgLen = 140;
		numDownDays = 3;
		numDaysInTrade = 9;
		stopLoss = 2500;
	Case 3: 
		movAvgLen = 160;
		numDownDays = 3;
		numDaysInTrade = 15;
		stopLoss = 2500;
 End;
// Actual system execution
value1 = countIf(c < c[1],numDownDays);

if multiAgent <> multiAgent[1] then print(d," ---->multiagent trans ");

if c > average(c,movAvgLen) and value1 = numDownDays then
begin
	if multiAgent = 1 then buy("Sys1") next bar at open;
	if multiAgent = 2 then buy("Sys2") next bar at open;
	if multiAgent = 3 then buy("Sys3") next bar at open;
end;
if barsSinceEntry >= numDaysInTrade then
	sell next bar at open;
if marketPosition = 1 then sell next bar at entryPrice - stopLoss/bigPointValue stop;
Cool usage of a switch-case and agent determination

If this seems over your head…

Get one of my books are check out Jeff Swanson’s course.  EasyLanguage has so many little nuggets that can help you define your algorithm into an actionable strategy.  You will never know how your strategy will work until your program it (properly) and back test it.  And then potentially improve it with optimization.

Multi-Agent Results

 

 

EasyLanguage Tutorial, Python, TradingSimula18, Uncategorized

Trading Before and After Holidays

2 Comments

Morgan Stanley Research Idea from 2018

In May of 2018 Morgan Stanley presented a report from their research department.  The report was written by Tony Small and Matthew Hornbach.  The information presented validated the idea of seasonality in the markets.  The writers/researchers gathered all the data going back to 1984 on the ten-year futures and all the federal holidays going back over the same span of time.  They then analyzed either going long or short M days prior to the holiday and exiting N days before the holidays.  They also did the same thing after the holiday – entering M days after and exiting N days after.  The boundary was set to five days before and after.  This report was sent to me by one of my clients.  The conclusion of the article was that there seemed to be a seasonality component to the days prior and after a major holiday.  And that it seemed this component did indeed provide a technical advantage (on the surface.)  Here is their conclusion.

“Given the risk of data mining, investors should accept this analysis with some degree of caution,” Small and Hornbach said. “There is no guarantee that historical seasonality and past price performance will continue in the future.”

How would one test this idea on their own?

Tools needed:

  1. Python with NumPy installed.
  2. An AI chatbot such as ChatGPT.
  3. Access to TY futures data going back to 1984 (actual non-Panama adjusted data would be preferable.)
  4. A testing platform that allows future leak such as George’s own TradingSimula-18.  Excel would work too but would require a bunch of work.

Some programming knowledge.

  1. Create output for all of the federal holidays going back to January 1984.  This list must also include Good Fridays.  This list should have two fields – the date and the name of the holiday.
  2. TS-18 came in handy because I read the file that was created from step 1 and whenever I came across or spanned a holiday date, I created a loop that calculated the ROC for the prior five days and also for the five days that followed.   I created a nomenclature to quickly recognize the ROC span.
    • B54 – ROC between 5 days before and 4 days before – 1 day
    • B53 – ROC between 5 days before and 3 days before – 2 days
    • B52 – ROC between 5 days before and 2 days before – 3 days
    • B51 – ROC between 5 days before and 1 days before – 4 days
    • B43 – ROC between 4 days before and 3 days before – 1 day
    • B42 – ROC between 4 days before and 2 days before – 2 days
    • B41 – ROC between 4 days before and 1 days before – 3 days
    • B32 – ROC between 3 days before and 2 days before – 1 day
    • B31 – ROC between 3 days before and 1 days before – 2 days
    • B21 – ROC between 2 days before and 1 days before – 1 day
    • A12 – ROC between 1 days after and 2 days after – 1 day
    • A13 – ROC between 1 days after and 3 days after – 2 days
    • A14 – ROC between 1 days after and 4 days after – 3 days
    • A15 – ROC between 1 days after and 5 days after – 4 days
    • A23 – ROC between 2 days after and 3 days after – 1 day
    • A24 – ROC between 2 days after and 4 days after – 2 days
    • A25 – ROC between 2 days after and 5 days after – 3 days
    • A34 – ROC between 3 days after and 4 days after – 1 day
    • A35 – ROC between 3 days after and 5 days after – 2 days
    • A45 – ROC between 4 days after and 5 days after – 1 day
  3. This process created a file that looked like this:

    • Output of ROC with nested for loops
  4. With this data I summed up each ROC calculation (B54 thru A45).  From this you could pick an entry and exit day combination.  The best results were wrapped around Memorial Day.  Here was where I ran into a small but highly aggravating scenario.  Most of the holidays had similar numbers, but there were a few that had some really strange values.  I knew I couldn’t use regular adjusted futures data, because many times these values can go below zero and when doing division with these price adjustments, the values are simply not accurate.  So, I used continuous non adjusted data, but ran into a problem with the rollover gap.  Some of the holidays occurred really close to rollover dates.  The calculations that bridged a rollover was not accurate.  I then remembered Pinnacle data also includes Ratio Adjusted Data or RAD.  I applied the looping algorithm to this data, and it SEEMED to fix the problem.
  5. I was able to pick a B(n,m) or A(m,n) for each market.  If the magnitude of ROC was sufficiently negative, you could use the dates to short the market.

ChatGPT and Python and Numpy to the rescue

I asked ChatGPT to provide python code to generate a list of all federal holidays from 1984.

from pandas.tseries.holiday import USFederalHolidayCalendar
from datetime import datetime,timedelta
import numpy as np
# Create a calendar for U.S. federal holidays
cal = USFederalHolidayCalendar()

# Get the holidays for the year 2014
holidays_2014 = cal.holidays(start='1984-01-01', end='2024-12-31').to_pydatetime()

federalHolidays = []
date_list = holidays_2014.astype(datetime).tolist()
for n in range(len(date_list)):
    dateString = f'{date_list[n]:%Y%m%d}'
    federalHolidays.append(int(dateString))
Create a list of federal holidays

This worked great, but I had to convert the NumPy datetime construct to a string.  I like to keep things simple.  You will see this in the code.  I started to get to work on the second part of the task when I discovered there were no dates for GOOD FRIDAYS – argh!  ChatGPT to rescue once again.

def good_friday_dates(start_year, end_year):
    good_friday_dates = []
    for year in range(start_year, end_year + 1):
        # Calculate the date of Easter Sunday for the given year
        easter_sunday = calculate_easter_date(year)

        # Good Friday is two days before Easter Sunday
        good_friday = easter_sunday - timedelta(days=2)

        # Add the Good Friday date to the list
        good_friday_dates.append(good_friday)

    return good_friday_dates

def calculate_easter_date(year):
    # Algorithm to calculate the date of Easter Sunday
    a = year % 19
    b = year // 100
    c = year % 100
    d = b // 4
    e = b % 4
    f = (b + 8) // 25
    g = (b - f + 1) // 3
    h = (19 * a + b - d - g + 15) % 30
    i = c // 4
    k = c % 4
    l = (32 + 2 * e + 2 * i - h - k) % 7
    m = (a + 11 * h + 22 * l) // 451
    month = (h + l - 7 * m + 114) // 31
    day = ((h + l - 7 * m + 114) % 31) + 1
    easter_date = datetime(year, month, day)
    return easter_date

# Generate Good Friday dates from 1984 to 2024
good_friday_dates_1984_to_2024 = good_friday_dates(1984, 2024)
Hope this code is right for Good Fridays

Good Friday, as Easter, is all over the map.  Tomorrow is Good Friday.  The code looks pretty gross, so I just accepted it.  It may not be 100% accurate, but if ChatGPT says so….  Once I had a list of all holidays, I was able to start cooking (MLK was added in 1986 and Juneteenth in 2021.)  Since Juneteenth is relatively new it did not offer a statistically significant number of observations.

Future Leak comes in Handy.

Every quant needs some software that will look through data and do some calculations.  I did this project with EasyLanguage, but it was a little difficult.  I had to do a two-pass process (well I did the same with Python too.)  I first created an indicator that imported the holiday dates and then outputted the close (or open of next bar) for close[5], close[4], close[3], close[2], close[1] and then waited 1, 2, 3, 4 and 5 bars after the date and printed out those respective closing prices.  I did not do this step in EasyLanguage.  I actually create EasyLanguage as my output.  Here is what it looked like:

holidayName[1]="MemorialDay";
fiveBeforeArr[1]=1010518;
fourBeforeArr[1]=1010521;
threeBeforeArr[1]=1010522;
twoBeforeArr[1]=1010523;
oneBeforeArr[1]=1010524;
oneAfterArr[1]=1010529;
twoAfterArr[1]=1010530;
threeAfterArr[1]=1010531;
fourAfterArr[1]=1010601;
fiveAfterArr[1]=1010604;
holidayName[2]="July4th";
fiveBeforeArr[2]=1010626;
fourBeforeArr[2]=1010627;
threeBeforeArr[2]=1010628;
twoBeforeArr[2]=1010629;
oneBeforeArr[2]=1010702;
oneAfterArr[2]=1010705;
twoAfterArr[2]=1010706;
threeAfterArr[2]=1010709;
fourAfterArr[2]=1010710;
fiveAfterArr[2]=1010711;
holidayName[3]="LaborDay";
fiveBeforeArr[3]=1010824;
fourBeforeArr[3]=1010827;
threeBeforeArr[3]=1010828;
twoBeforeArr[3]=1010829;
oneBeforeArr[3]=1010830;
oneAfterArr[3]=1010904;
twoAfterArr[3]=1010905;
threeAfterArr[3]=1010906;
fourAfterArr[3]=1010907;
fiveAfterArr[3]=1010910;
All key dates are stored in arrays.

Future leak allows you to peak into the future and this comes in super handy when doing research like this.  Here is some python code in TS-18 that does all these calculations in one sweep.


#   The date of the holiday and the holiday name are stored in lists
#   Each Holiday has its own date and 20 ROC calculations
#   These calculations are stored in lists as well
	tempDate = int(extraData[extraCnt][0]) #first field in the csv file
	whichHoliday = extraData[extraCnt][1] #second field in the csv file
	if date[D] == tempDate or (date[D] > tempDate and date[D1] < tempDate):
		foundJ = 99
		for j in range(12):
			if whichHoliday == listOfHolidayClass[j].holidayName:
				foundJ = j
				listOfHolidayClass[foundJ].holidayDates.append(tempDate);
				for j in range(-5,-1):  #Before Holiday
					for k in range(j+1,0):
						denom = close[D+j]
						listOfHolidayClass[foundJ].rocList.append((close[D+k] - close[D+j])/denom)
				for j in range(0,4):   #After Holiday - here is the future leak
					for k in range(j+1,5):
						denom = close[D+j]
						listOfHolidayClass[foundJ].rocList.append((close[D+k] - close[D+j])/denom)
				extraCnt = extraCnt + 1
Python code using future leak to calculate the 20 ROC from before and after holiday

I then printed this out to a csv file and imported into Excel (see above) and found out the best combination of days for each holiday.  Now that I had the B(n,m) or the A(m,n) values, I need a process to tell TS-18 which day to buy/short and exit.  Here is some code that, again uses future leak, to print out the five dates before the holiday and the five dates after the holiday.

tempDate = int(extraData[extraCnt][0])
whichHoliday = extraData[extraCnt][1]
if date[D] == tempDate or (date[D] > tempDate and date[D1] < tempDate):
	for j in range(12):
		if whichHoliday == listOfHolidayClass[j].holidayName:
			foundJ = j
			print(date[D],end=',')
			print(whichHoliday,end=',')
			for j in range(-5,6):
				print(date[D+j],end=',')
				print("",end = '\n')
				extraCnt = extraCnt + 1

# Here is what the file looked like
19840103,NewYear,19831223,19831227,19831228,19831229,19831230,19840103,19840104,19840105,19840106,19840109,19840110,
19840221,PresidentsDay,19840213,19840214,19840215,19840216,19840217,19840221,19840222,19840223,19840224,19840227,19840228,
19840423,GoodFriday,19840413,19840416,19840417,19840418,19840419,19840423,19840424,19840425,19840426,19840427,19840430,
19840529,MemorialDay,19840521,19840522,19840523,19840524,19840525,19840529,19840530,19840531,19840601,19840604,19840605,
19840705,July4th,19840627,19840628,19840629,19840702,19840703,19840705,19840706,19840709,19840710,19840711,19840712,
19840904,LaborDay,19840827,19840828,19840829,19840830,19840831,19840904,19840905,19840906,19840907,19840910,19840911,
19841008,ColumbusDay,19841001,19841002,19841003,19841004,19841005,19841008,19841009,19841010,19841011,19841012,19841015,
19841112,VeteransDay,19841105,19841106,19841107,19841108,19841109,19841112,19841113,19841114,19841115,19841116,19841119,
19841123,Thanksgiving,19841115,19841116,19841119,19841120,19841121,19841123,19841126,19841127,19841128,19841129,19841130,
Create the dates that occur in the date B4 and after holiday

Now read the data in and pick how many days before to b/s and before to exit or after to b/s and to exit

holidayBuyDates = []
holidayXitDates = []
gfCnt = 0

#structure of holidayDatesData list - all strings
# 0.) Holiday Date
# 1.) Holiday Name
# 2.) 5-Before
# 3.) 4-Before
# 4.) 3-Before
# 5.) 2-Before
# 6.) 1-Before
# 7.) 0-Before/After
# 8.) 1-After
# 9.) 2-After
# 10.) 3-After
# 11.) 4-After
# 12.) 5-After
#Below I pick 1 day after to buy and 4 days after to exit [8,11]
for j in range(numHolidayDates):
    if holidayDatesData[j][1] == "MemorialDay":
        holidayBuyDates.append(int(holidayDatesData[j][8]))
        holidayXitDates.append(int(holidayDatesData[j][11]))
		
		

#  Okay  Let's put in some logic to create a long position
if not(long) and canTrade and date[D] == holidayBuyDates[gfCnt]:
	price = close[D]
	tradeName = "Hol-A1Buy"
	posSize = 1
	tradeTicket.append([buy,tradeName,posSize,price,mkt])
#  Long Exits
if long and canTrade and date[D] == holidayXitDates[gfCnt]:
	price = longExit
	tradeName = "Hol-A4Xit"
	tradeTicket.append([exitLong,tradeName,curShares,price,mkt])
	gfCnt = gfCnt + 1
TradingSimual_-18 Code to execute buys and exits
No execution fees – executing on close of the day
Memorial Day Example -----------------------------------------
Sys Name : TS-18HolidayTest
Mkt Symb : TY Mkt Name : 10YNote
19840530     Hol-A1Buy  1 -35.90625       0.00       0.00
19840604     Hol-A4Xit  1 -34.12500    1731.25    1731.25
19850529     Hol-A1Buy  1 -18.43750       0.00       0.00
19850603     Hol-A4Xit  1 -16.96875    1418.75    3150.00
19860528     Hol-A1Buy  1  -0.40625       0.00       0.00
19860602     Hol-A4Xit  1  -2.87500   -2518.75     631.25
19870527     Hol-A1Buy  1  -1.09375       0.00       0.00
19870601     Hol-A4Xit  1  -0.28125     762.50    1393.75
19880601     Hol-A1Buy  1  -1.87500       0.00       0.00
19880606     Hol-A4Xit  1  -0.56250    1262.50    2656.25
19890531     Hol-A1Buy  1   2.68750       0.00       0.00
19890605     Hol-A4Xit  1   4.46875    1731.25    4387.50
19900530     Hol-A1Buy  1   2.50000       0.00       0.00
19900604     Hol-A4Xit  1   3.68750    1137.50    5525.00
19910529     Hol-A1Buy  1   6.59375       0.00       0.00
19910603     Hol-A4Xit  1   6.65625      12.50    5537.50
19920527     Hol-A1Buy  1  13.71875       0.00       0.00
19920601     Hol-A4Xit  1  14.50000     731.25    6268.75
19930602     Hol-A1Buy  1  28.21875       0.00       0.00
19930607     Hol-A4Xit  1  27.96875    -300.00    5968.75
19940601     Hol-A1Buy  1  24.62500       0.00       0.00
19940606     Hol-A4Xit  1  25.93750    1262.50    7231.25
19950531     Hol-A1Buy  1  33.34375       0.00       0.00
19950605     Hol-A4Xit  1  34.59375    1200.00    8431.25
19960529     Hol-A1Buy  1  32.09375       0.00       0.00
19960603     Hol-A4Xit  1  30.93750   -1206.25    7225.00
19970528     Hol-A1Buy  1  32.93750       0.00       0.00
19970602     Hol-A4Xit  1  33.75000     762.50    7987.50
19980527     Hol-A1Buy  1  40.00000       0.00       0.00
19980601     Hol-A4Xit  1  40.31250     262.50    8250.00
19990602     Hol-A1Buy  1  37.62500       0.00       0.00
19990607     Hol-A4Xit  1  37.81250     137.50    8387.50
20000531     Hol-A1Buy  1  36.10938       0.00       0.00
20000605     Hol-A4Xit  1  37.43750    1278.12    9665.62
20010530     Hol-A1Buy  1  42.92188       0.00       0.00
20010604     Hol-A4Xit  1  44.06250    1090.62   10756.25
20020529     Hol-A1Buy  1  50.14062       0.00       0.00
20020603     Hol-A4Xit  1  50.68750     496.88   11253.12
20030528     Hol-A1Buy  1  68.06250       0.00       0.00
20030602     Hol-A4Xit  1  68.34375     231.25   11484.38
20040602     Hol-A1Buy  1  64.43750       0.00       0.00
20040607     Hol-A4Xit  1  63.78125    -706.25   10778.12
20050601     Hol-A1Buy  1  72.32812       0.00       0.00
20050606     Hol-A4Xit  1  72.29688     -81.25   10696.88
20060531     Hol-A1Buy  1  65.46875       0.00       0.00
20060605     Hol-A4Xit  1  66.07812     559.38   11256.25
20070530     Hol-A1Buy  1  66.57812       0.00       0.00
20070604     Hol-A4Xit  1  66.06250    -565.62   10690.62
20080528     Hol-A1Buy  1  77.21875       0.00       0.00
20080602     Hol-A4Xit  1  76.10938   -1159.38    9531.25
20090527     Hol-A1Buy  1  87.12500       0.00       0.00
20090601     Hol-A4Xit  1  87.14062     -34.38    9496.88
20100602     Hol-A1Buy  1  95.48438       0.00       0.00
20100607     Hol-A4Xit  1  96.18750     653.12   10150.00
20110601     Hol-A1Buy  1 102.23438       0.00       0.00
20110606     Hol-A4Xit  1 103.12500     840.62   10990.62
20120530     Hol-A1Buy  1 115.32812       0.00       0.00
20120604     Hol-A4Xit  1 117.37500    1996.88   12987.50
20130529     Hol-A1Buy  1 115.51562       0.00       0.00
20130603     Hol-A4Xit  1 115.59375      28.12   13015.62
20140528     Hol-A1Buy  1 116.31250       0.00       0.00
20140602     Hol-A4Xit  1 116.56250     200.00   13215.62
20150527     Hol-A1Buy  1 121.20312       0.00       0.00
20150601     Hol-A4Xit  1 121.42188     168.75   13384.38
20160601     Hol-A1Buy  1 125.07812       0.00       0.00
20160606     Hol-A4Xit  1 126.50000    1371.88   14756.25
20170531     Hol-A1Buy  1 124.17188       0.00       0.00
20170605     Hol-A4Xit  1 124.65625     434.38   15190.62
20180530     Hol-A1Buy  1 120.35938       0.00       0.00
20180604     Hol-A4Xit  1 119.21875   -1190.62   14000.00
20190529     Hol-A1Buy  1 124.76562       0.00       0.00
20190603     Hol-A4Xit  1 125.76562     950.00   14950.00
20200527     Hol-A1Buy  1 137.43750       0.00       0.00
20200601     Hol-A4Xit  1 137.84375     356.25   15306.25
20210602     Hol-A1Buy  1 133.03125       0.00       0.00
20210607     Hol-A4Xit  1 133.42188     340.62   15646.88
20220601     Hol-A1Buy  1 122.48438       0.00       0.00
20220606     Hol-A4Xit  1 121.70312    -831.25   14815.62
20230531     Hol-A1Buy  1 115.40625       0.00       0.00
20230605     Hol-A4Xit  1 115.10938    -346.88   14468.75

Here are the results for the Morgan Stanley analysis in 2018

Best combos for entry and exit for each holiday.

My analysis covered the same period plus all of 2018, 2019, 2020, 2021, 2022, and 2023.  My combos were very similar in most cases, but the last few years moved the needle a little bit.  However, the results still showed a technical edge.

Complete EasyLanguage Systems, EasyLanguage, EasyLanguage equivalent to cool code, EasyLanguage Function, EasyLanguage Indicator, EasyLanguage Snippets, EasyLanguage Tutorial, EasyLanguage Tutorial, Free EasyLanguage, MultiCharts, Must Know EasyLanguage, Trend Following

How to Fix the Fixed Fractional Position Size

Leave a comment

The Fixed Fractional position sizing scheme is the most popular, so why does it need fixed?

Problems solved with Fixed Fractional:

  1. Efficient usage of trading capital
  2. Trade size normalization between different futures contracts
  3. Trade size normalization across different market environments

These are very good reasons why you should use positions sizing.  Problem #2 doesn’t apply if you are only trading one market.  This sounds idyllic, right?  It solves these two problems, but it introduces a rather bothersome side effect – huge draw down.  Well, huge draw down in absolute terms.  Draw downs when using a fixed fractional approach are proportional to the prior run up.  If you make a ton of money on a prior trade, then your position sizing reflects that big blip in the equity curve.  So, if you have a large loser following a large winner, the draw down will be a function of the run up.  In most cases, a winning trading system using fixed fractional position sizing will scale profit up as well as draw down.  A professional money manager will look at the profit to draw down ratio instead of the absolute draw down value.  The efficient use of capital will reflect a smaller position size after a draw down, so that is good right?  It is unless you believe in a Martingale betting algorithm – double up on losses and halve winners.  Are we just stuck with large “absolute” draw downs when using this size scheme?

Possible solutions to fixing Fixed Fractional (FF)

The first thing you can do is risk less than the industry standard 2% per trade.  Using 1% will cause equity to grow at a slower rate and also reduce the inevitable draw down.  But this doesn’t really solve the problem as we are giving up the upside.  And that might be okay with you.  Profit will increase and you are using an algorithm for size normalization.  In this blog I am going to propose a trading equity adjustment feature while using FF.  What if we act like money managers, and you should even if you are trading your own personal money, and at the end of the year or month we take a little off the table (theoretically – we are not removing funds from the account just from the position sizing calculation) that is if there is any extra on the table.  This way we are getting the benefit of FF while removing a portion of the compounding effect, which reduces our allocation for the next time interval.  How do you program such a thing?  Well first off let’s code up the FF scheme.

positionSize = round((tradingCapital * riskPerTrade) / (avgTrueRange(30)*bigPointValue),0);

Nothing new here.  Simply multiply tradingCapital by the riskPerTrade (1 or 2%) and then divide by a formula that defines current and inherent market risk.  This is where you can become very creative.  You could risk the distance between entry and exit if you know those values ahead of time or you can use a value in terms of the current market.  Here I have chosen the 30-day average true range.  This value gives a value that predicts the market movement into the future.  However, this value only gives the expected market movement for a short period of time into the future.  You could us a multiplier since you will probably remain in a trade for more than a few days – that is if you are trying to capture the trend.  In my experiment I just use one as my multiplier.

Capture and store the prior year/month NetProfit

When I come up with a trading idea I usually just jump in and program it.  I don’t usually take time to see if Easy Language already provides a solution for my problem.   Many hours have been used to reinvent the wheel, which isn’t always a bad thing.  So, I try to take time and search the functions to see if the wheel already exists.  This time it looks like I need to create the wheel.  I will show the code first and then explain afterward.

inputs: useAllocateYearly(True),useAllocateMonthly(False),initCapital(100000),removePerProfit(0.50),riskPerTrade(0.01);
vars: tradingCapital(0),prevNetProfit(0),tradingCapitalAdjustment(0);
vars: oLRSlope(0),oLRAngle(0),oLRIntercept(0), oLRValueRaw(0),mp(0);
arrays: yearProfit[250](0),snapShotNetProfit[250](0);vars: ypIndex(0);


once
begin
	tradingCapital = initCapital;
end;

if useAllocateYearly then
begin
	value1 = year(d);
	value2 = year(d[1]); 
end;

if useAllocateMonthly then //remember make sure your array is 12XNumYears
begin
	value1 = month(d);
	value2 = month(d[1]);
end;

if useAllocateYearly or useAllocateMonthly then
begin
	if value1 <> value2 then
	begin
		if ypIndex > 0 then
			yearProfit[ypIndex] = prevNetProfit - snapShotNetProfit[ypIndex-1]
		else
			yearProfit[ypIndex] = prevNetProfit;	
			
		snapShotNetProfit[ypIndex] = prevNetProfit;
		tradingCapitalAdjustment = yearProfit[ypIndex];	
		if yearProfit[ypIndex] > 0 then
			tradingCapitalAdjustment = yearProfit[ypIndex] * (1-removePerProfit);
		tradingCapital = tradingCapital + tradingCapitalAdjustment;
		print(d,",",netProfit,",",yearProfit[ypIndex],",",tradingCapitalAdjustment,",",tradingCapital);
		ypIndex +=1;
	end;
end
else
	tradingCapital = initCapital + netProfit;
Capture either the prior years or months net profit

I wanted to test the idea of profit retention on a monthly and yearly basis to see if it made a difference.  I also wanted to just use the vanilla version of FF.  The use of Arrays may not be necessary, but I didn’t know ahead of time.  When you program on the fly, which is also called “ad hoc” programming you create first and then refine later.  Many times, the “ad hoc” version turns out to be the best approach but may not be the most efficient.  Like writing a book, many times your code needs revisions.  When applying a study or strategy that uses dates to a chart, you don’t know exactly when the data starts so you always need to assume you are starting in the middle of a year.   If you are storing yearly data into an array, make sure you dimension you array sufficiently.  You will need 12X the number of years as the size you need to dimension your array if you want to store monthly data.

 


//250 element array will contain more than 20 years of monthly data
//You could increase these if you like just to be safe
arrays: yearProfit[250](0),snapShotNetProfit[250](0);
//Remember you dimension you arrray variable first and then
//Pass it a value that you want to initiate all the values
//in the array to equal
vars: ypIndex(0);
Dimension and Initiate Your Arrays

The first thing we need to do is capture the beginning of the year or month.  We can do this by using the year and month function.  If the current month or year value is not the same as the prior day’s month or year value, then we know we crossed the respective timeline boundary.  We are using two arrays, snapShotNetProfit and yearProfit (to save time I use this array to store monthlty values as well, when that option is chosen) and a single array index ypIndex.  If we have crossed the time boundary, the first thing we need to do is capture the EasyLanguage function NetProfit’s value.  NetProfit keeps track of the cumulative closed out trade profits and losses. Going forward in this description I am going to refer to a yearly reallocation.  If it’s the first year, the ypIndex will be zero, and in turn the first year’s profit will be the same as netProfit.  We store netProfit in the yearProfit array at the ypIndex location.  Since we are in a new year, we take a snapshot of netProfit and store it in the snapShotNetProfit array at the same ypIndex location.  You will notice I use the variable prevNetProfit in the code for netProfit.  Here is where the devil is in the details.  Since we are comparing today’s year value with yesterday’s year value and when they are different, we are already inside the new year, so we need to know yesterday’s netProfit.   Before you say it, you can’t pass netProfit a variable for prior values; you know like netProfit(1) or netProfit[1] – this is a function that has no historic values, but you can record the prior day’s value by using our own prevNetProfit  variable.  Now we need to calculate the tradingCapitalAdjustment.  The first thing we do is assign the variable the value held in  yearProfit[ypIndex].  We then test the yearProfit[ypIndex] value to see if it is positive.  If it is, then we multiply it by (1-removePerProfit).  If you want to take 75% of the prior year’s profit off the table, then you would multiply the prior year’s profit by 25%.  Let’s say you make $10,000 and you want to remove $7,500, then all you do is multiply $10,000 by 25%.  If the prior year’s netProfit is a loss, then this value flows directly through the code to the position sizing calculation (auto deallocation on a losing year).   If not, the adjusted profit portion of funds are deallocated in the position sizing equation.

The next time we encounter a new year, then we know this is the second year in our data stream, so we need to subtract last year’s snapshot of netProfit (or prevNetProfit) from the current netProfit.  This will give us the change in the yearly net profit.  We stuff this information into the yearProfit array.  The snapShotNetProfit is stuffed with the current prevNetProfit.  ypIndex is incremented every time we encounter a new year.  Notice how I increment the ypIndex – it is incremented after all the calculations in the new year.  The tradingCapitalAdjustment is then calculated with the latest information.

Here is a table of how the tradingCapital and profit retention adjustment are calculated.  A yearly profit adjustment only takes place after a profitable year.  A losing year passes without adjustment.

All tests were carried out on the trend friendly crude oil futures with no execution costs from 2006 thru 2/28/2204.

See how money is removed from the allocation model after winning years.

Here are some optimization tests with 75% profit retention on yearly and monthly intervals.

Yearly First-

Yearly retention with no stop loss or break-even levels.

Now Monthly-

Monthly retention with no stop loss or break-even levels.

What if we didn’t reallocate on any specific interval?

Huge drawdowns with very little change in total profit.

Add some trade management into the mix.

Here we optimize a protective stop and a break-even level to see if we can juice the results.  All the trade management is on a position basis.

200K with No Reallocating with $14k and $8.5 stop/break-even levels [ranked by Profit Factor]
No position sizing and no reallocation with $5K and $10K stop/break-even levels

200K and monthly reallocating with $7K and $5.5K stop/break-even levels [BEST PROFIT FACTOR]
200K and monthly reallocating with $7K and $8.5K stop/break-even levels [2ND BEST PROFIT FACTOR]

Are we really using our capital in the most efficient manner?

If we retain profit, should we remove some of the loss form the position sizing engine as well.  All the tests I performed retained profit from the position size calculations.  I let the loss go full bore into the calculation.  This is a very risk averse approach.  Why don’t I retain 25% of losses and deduct that amount from the yearly loss and feed that into the position sizing engine.  This will be less risk averse – let’s see what it does.

Not as good.  But I could spend a week working different permutations and optimization sessions.

Are you wondering what Trend Following System I was using as the foundation of this strategy?  I used EasyLanguage’s Linear Regression function to create buy and short levels.  Here is the very simple code.

Value1 = LinearReg (Close, 60, 1, oLRSlope, oLRAngle, oLRIntercept, oLRValueRaw);

Value2 = oLRSlope;

Value3 = oLRAngle;

Value4 = oLRIntercept;

Value5 = oLRValueRaw;


//Basically I am buying/shorting on the change of the linear regression slope
//Also I have a volatility filter but it really isn't used
If value2 >=0 and value2[1] < 0 and avgTrueRange(30)*bigPointValue < 10000 then 
	buy positionSize contracts next bar at market;
If value2 <=0 and value2[1] > 0 and avgTrueRange(30)*bigPointValue < 10000 then 
	sellShort positionSize contracts next bar at market;

mp = marketPosition;

//I also have incorporated a 3XATR(30) disaster stop
if mp = 1 and c <= entryPrice - 3 * avgTrueRange(30) then sell next bar at market;
if mp = -1 and c >= entryPrice + 3 * avgTrueRange(30) then buyToCover next bar at market

If you want to see some more Trend Following models and their codes in EasyLanguage and TS-18 Python check out my TrendFollowing Guide and Kit.

…

Complete EasyLanguage Systems, EasyLanguage, EasyLanguage equivalent to cool code, EasyLanguage Function, EasyLanguage Snippets, EasyLanguage Tutorial, Free EasyLanguage, MultiCharts, Must Know, Must Know EasyLanguage

How To Test and Optimize Turn of the Month Seasonality

Leave a comment

Historical evidence suggests a potential seasonal pattern around the end of the month in the markets.

If you have been involved with the markets for even a short period of time, you have heard about this trade.  Buy N days prior to the end of the month and then exit M days after the end of the month.  This is a simple test to perform if you have a way to determine the N and the M in the algorithm.  You could always buy on the 24th of the month, but the 24th of the month may not equal N days prior to the end of the month. 

Simple approach that doesn’t always work – buy the 24th of the month and exit the 5th of the following month.

if dayOfMonth(d) = 24 then buy next bar at open;

if marketPosition = 1 and dayOfMonth(d) = 5 then sell next bar at open;

Before we get into a little better coding of this algorithm, let’s see the numbers.  The first graph is trading one contract of the ES futures once a month – no execution fees were applied.  The same goes for the US bond futures chart that follows.  Before reading further please read this.

CFTC-required risk disclosure for hypothetical results:

Hypothetical performance results have many inherent limitations, some of which are described below. No representation is being made that any account will or is likely to achieve profits or losses similar to those shown. in fact, there are frequently sharp differences between hypothetical performance results and the actual results subsequently achieved by any particular trading program.

One of the limitations of hypothetical performance results is that they are generally prepared with the benefit of hindsight. In addition, hypothetical trading does not involve financial risk, and no hypothetical trading record can completely account for the impact of financial risk in actual trading. For example, the ability to withstand losses or to adhere to a particular trading program in spite of trading losses are material points which can also adversely affect actual trading results. There are numerous other factors related to the markets in general or to the implementation of any specific trading program which cannot be fully accounted for in the preparation of hypothetical performance results and all of which can adversely affect actual trading results.
Buying N days before EOM and Selling M days after EOM
Ditto!

No Pain No Gain

Looking into the maw of draw down and seeing the jagged and long teeth.

Draw down as a percentage of account value.
Ditto2

The bonds had more frequent draw down but not so deep.  These teeth can cause a lot of pain.

Well, George what is the N and M?

I should have done M days before and N days after to maintain alphabetic order, but here you go.

ES: N = 6 AND M = 6

US: N =10 AND M = 1

How did you do this?

Some testing platforms have built-in seasonality tools, but, and I could be wrong I didn’t find what I needed in the TradeStation function library.  So, I built my own.

A TradingDaysLeftInMonth function had to be created.  This function is a broad swipe at attempting to determining this value.  It’s not very smart because it doesn’t take HOLIDAYS into consideration.  But for a quick analysis it is fine.  How does one design such a function?  First off, what do we know to help provide information that might be useful?  We know how many days are in each month (again this function isn’t smart enough to take into consideration leap years) and we know what day of the week each trading day belongs to.  We have this function DayOfWeek(Date) already in EasyLanguage.  And we know the DayOfMonth(Date) (built-in too!) With these three tidbits of information, we should be able to come up with a useful function.   Not to mention a little programming knowledge.  I was working on a Python project when I was thinking of this function, so I decided to prototype it there.  No worries, the algorithm can be easily translated to EasyLanguage. And yes, I could have used my concept of a Sandbox to prototype in EasyLanguage as well.  Remember a sandbox is a playground where you can quickly test a snippet of code.  Using the ONCE keyword, you can quickly throw some generic EasyLanguage together sans trade directives and mate it to a chart and get to the nuts and bolts quickly.  I personally like having an indicator and a strategy sandbox.  Here is a generic snippet of code where we assume the day of month is the 16th and it is a Monday ( 2 – 1 for Sunday thru 7 for Saturday) and there are 31 days in whatever month.

currentDayOfWeek = 2;
currentDayOfMonth = 16;
loopDOW = currentDayOfWeek;
daysInMonth = 31
#create the calender for the remaining month
tdToEOM=0; #total days to EOM
for j in range(currentDayOfMonth,daysInMonth+1):
    if loopDOW != 1 and loopDOW != 7:
       tdToEOM +=1;
    print(j," ",loopDOW," ",tdToEOM)
    loopDOW +=1;
    if loopDOW > 7: loopDOW = 1; #start back on Monday
Create a synthetic calendar from the current day of month

I just absolutely love the simplicity of Python.  When I am prototyping for EasyLanguage, I put a semicolon at the end of each line.  Python doesn’t care.  Here is the output from this snippet of code.

Cur>Day    DOWDay  DaysLeftAccum.
-----------------------------------
16           2       1  Monday
17           3       2  Tuesday
18           4       3  Wednesday
19           5       4  Thursday
20           6       5  Friday
21           7       5  Saturday
22           1       5  Sunday
23           2       6  Monday
24           3       7  Tuesday
25           4       8  Wednesday
26           5       9  Thursday
27           6      10  Friday
28           7      10  Saturday
28           1      10  Sunday
30           2      11  Monday
31           3      12  Tuesday

On Monday 16th there were 12 Trading Days Left In Month Inclusive
Output of Python Snippet - use in EZLang.

I start out with the current day of the month, 16 and loop through the rest of the days of the month.  Whenever I encounter a Sunday (1) or a Saturday (7) I do not increment tdToEOM, else I do increment.  

Here is how the function works on a chart.  Remember in TradeStation I am placing a market order for the NEXT BAR.

Counting the days until the EOM

This snippet of code is the heart of the function, but you must make in generic for any day of any month.  Here it is in EasyLanguage – you will see the similarity between the Python snippet and its corresponding EasyLanguage.

array: monthDays[12](0);

monthDays[1] = 31;
monthDays[2] = 28;
monthDays[3] = 31;
monthDays[4] = 30;
monthDays[5] = 31;
monthDays[6] = 30;
monthDays[7] = 31;
monthDays[8] = 31;
monthDays[9] = 30;
monthDays[10] = 31;
monthDays[11] = 30;
monthDays[12] = 31;

vars: curDayOfMonth(0),curDayOfWeek(0),loopDOW(0),tdToEOM(0),j(0);

curDayOfWeek = dayOfWeek(d);
curDayOfMonth = dayOfMonth(d);

{Python prototype
tdToEOM=0;
for j in range(currentDayOfMonth,daysInMonth+1):
    if loopDOW != 1 and loopDOW != 7:
       tdToEOM +=1;
    print(j," ",loopDOW," ",tdToEOM)
    loopDOW +=1;
    if loopDOW > 7: loopDOW = 1;
}

loopDOW = curDayOfWeek+1; 
tdToEOM=0;

for j = curDayOfMonth to monthDays[month(d)]
begin
    if loopDOW <> 1 and loopDOW <> 7 then
       tdToEOM +=1; // tdToEOM = tdToEOM + 1;
    loopDOW +=1;
    if loopDOW > 7 then loopDOW = 1;
end;
TradingDaysLeftInMonth = tdToEOM;
EasyLanguage Function : TradingDaysLeftInMonth

I used arrays to store the number of days in each month.  You might find a better method.  Once I get the day of the month and the day of the week I get to work.  EasyLanguage uses a 0 for Sunday so to be compliant with the Python function I add a 1 to it.  I then loop from the current day of month through monthDays[month(d)].  Remember month(d) returns the month number [1…12].  A perfect index into my array.  That is all there is to it.  The code is simple, but the concept requires a little thinking.  Okay, now that we have the tools for data mining, let’s do some.  I did this by creating the following strategy (the same strategy that create the original equity curves.)

inputs: numDaysBeforeEOM(8),numDaysAfterEOM(10),movingAvgLen(100);
inputs: stopLossAmount(1500),profitTargetAmount(4000);

vars: TDLM(0),TDIM(0);

TDLM = tradingDaysLeftInMonth;
TDIM = tradingDayOfMonth;

if c >= average(c,movingAvgLen) and TDLM = numDaysBeforeEOM then
begin
	buy("Buy B4 EOM") next bar at open;	
end;

if marketPosition = 1 and barsSinceEntry > 3 then
begin
	if TDIM = numDaysAfterEOM then 
	begin
		sell("Sell TDOM") next bar at open;	
	end;
end;
setStopLoss(stopLossAmount);
setProfitTarget(profitTargetAmount);
EasyLanguage function driver in form of Strategy

A complete strategy has trade management and an entry and an exit.  In this case, I added an additional feature – a trend detector in the form of a longer-term moving average.  Let’s see if we can improve the trading system.  Thank goodness for Genetic Optimization.  Here is the @ES market.

Get your Pick ready to mine!

Smoothed the equity curve – took the big draw down out.

Genetically MODIFIED – Data Mining at its best!

Here are the parameters:

Did not like the moving average. Wide stop and wide profit objective. Days to EOM and after EOM stayed the same.

Bond System:

Bond market results.

If you like this type of programming check out my books at Amazon.com.  I have books on Python and of course EasyLanguage.  I quickly assembled a YouTube video discussing this post here.

Conclusion – there is something here, no doubt.  But it can be a risky proposition.  It definitely could provide fodder for the basis of a more complete trading system.

George’s Amazon Page

 

Backtesting with [Trade Station,Python,AmiBroker, Excel]. Intended for informational and educational purposes only!