Have You Ever Wondered If You Just Reversed the Logic?
You have been there before. What you thought was a great trading idea turns out to be a big flop. We have all developed these types of algorithms. Then it hits you, just flip the logic and in turn the equity curve. Hold your horses! First off you have to make sure it’s not just the execution costs that is hammering the equity curve into oblivion. When testing a fresh trading idea, it is best to keep execution costs to zero. This way if your idea is a good one, but is simply backward, then you have a chance of creating something good out of something bad. I was playing around with a mean reversion day trading algorithm (on the @ES.D – day session of the mini S&P 500) that created the following equity curve. Remember to read the disclaimer concerning hypothetical performance before proceeding reading the rest of this blog. It is located under the DISCLAIMER – REAMDE! tab. By reading the rest of this blog post it implies that you understand the limitations of hypothetical back testing and simulated analysis.
The pandemic created a strong mean reversion environment. In the initial stage of this research, I did not set the executions costs – they defaulted to zero. My idea was to buy below the open after the market moved down from the high of the day a certain percentage of price. Since I was going to be buying as the market was moving down, I was willing to use a wide stop to see if I could hold on to the falling knife. Short entries were just the opposite -sell short above the open after the market rallied a certain percentage of price. I wanted to enter on a hiccup. Once the market moved down a certain range from the high of the day, I wanted to enter on a stop at the high of the prior bar. I figured if the price penetrated the high of the prior five-minute bar in a down move, then it would signal an eventual rotation in the market. Again, I was just throwing pasta against the wall to see what would stick. I even came up with a really neat name for the algorithm the Rubber Band system – stretch just far enough and the market is bound to slam back. Well, there wasn’t any pasta sticking. Or was there? If I flipped the equity curve 180 degrees, then I would have a darned good strategy. All it would take is to reverse the signals, sell short when I was buying and buy when I was selling short. Instead of a mean reversion scheme, this would turn into a momentum-based strategy.
if c < todaysOpen and todaysOpen-c = maxOpenMinusClose and (maxCloseMinusOpen + maxOpenMinusClose)/c >= stretchPercent Then canBuy = True; if c > todaysOpen and c- todaysOpen = maxCloseMinusOpen and (maxCloseMinusOpen + maxOpenMinusClose)/c >= stretchPercent Then canShort = True;
Guts of the complete failure.
Here I measure the maximum distance from the highest close above the open and the lowest close below the open. The distance between the two points is the range between the highest and lowest closing price of the current day. If the close is less than today’s open, and the range between the extremes of the highest close and lowest close of the trading day is greater than stretchPercent, then an order directive to buy the next bar at the current bar’s high is issued. The order is alive until it is filled, or the day expires. Selling short uses the same calculations but requires the close of the current bar to be above the open. The stretchPercent was set to 1 percent and the protective stop was set to a wide $2,000. As you can see from the equity curve, this plan did not work except for the time span of the pandemic. Could you optimize the strategy and make it a winning system. Definitely. But the 1 percent and $2000 stop seemed very logical to me. Since we are comparing the range of the data to a fixed price of the data, then we don’t need to worry about the continuous contract distortion. Maybe we would have to, if the market price was straddling zero. Anyways, here is a strategy using the same entry technique, but reversed, with some intelligent trade filtering. I figured a profit objective might be beneficial, because the stop was hit several times during the original test.
$2K was hit often!Using some trade filtering and stop loss and profit objective on the reversal of the original strategy.
If you like the following code, make sure you check out my books at Amazon.com. This type of code is used the Hi-Res and Day-Trading editions of the Easing_Into_Easylanguage series.
if c < todaysOpen and todaysOpen-c = maxOpenMinusClose and (maxCloseMinusOpen + maxOpenMinusClose)/c >= stretchPercent Then canShort = True; if c > todaysOpen and c- todaysOpen = maxCloseMinusOpen and (maxCloseMinusOpen + maxOpenMinusClose)/c >= stretchPercent Then canBuy = True;
if canTrade and t >= calcTime(dontTradeBefore,dontTradeBeforeOffset) and t < calcTime(dontTradeAfter,dontTradeAfterOffset) and t < sessionEndTime(0,1) Then begin if shortsToday = 0 and canShort = True Then sellshort next bar at l stop; if buysToday = 0 and canBuy = True Then buy next bar at h stop; end;
Trade filtering was obtained by limiting the duration during the trading day that a trade could take place. It’s usually wise to wait a few minutes after the open and a few minutes prior to the close to issue trade directives. Also, range compression of the prior day seems to help in many cases. Or at least not range expansion. I only allow one long entry or one short or both during the trading day – two entries only! Read the code and let me know if you have any questions. This is a good framework for other areas of research. Limiting entries using the mp variable is a neat technique that you can use elsewhere.
And as always let me know if you see any bugs in the code. Like Donnie Knuth says, “Beware of bugs in the above code; I have only proved it correct, not tried it!”
To Rollover or not to Rollover – that is the Question!
In actuality that is only a portion of what we need to ask ourselves when dealing with rollovers and data. As you probably already know, continuous futures contracts are created by adjusting historic data in an attempt to eliminate the gap between the new and old contract. If you don’t adjust your data, then the gap causes erroneous disruptions in your historic back testing. A trade that bridges a rollover does not accurately reflect what really happened. So, we must use adjusted data. Two problems arise from using back adjusted data:
The date of the rollover can have a large impact on algorithm performance.
In real time trading, rollovers must take place, and this increases execution costs. If you trade a longer-term strategy, you may need to execute 12 additional spreads (exit old – enter new) in crude oil annually. If you levy a $50 charge (slip and commission) for each turn, then that is $600 in additional costs annually.
Several options can be used to trigger when to rollout of the expiring contract and roll into the new one. Some traders use consecutive days of higher volume and open interest in the new contract in relation to the expiring one to begin the process. This mechanism is also popular in creating continuous contract data. Other traders, and I was one of them, chose a particular trading day of the month based on the expiring date of the contract. Many futures contracts expire after a certain number of days have passed in the expiring month or the prior month of expiration. This mechanism, n-days into the expiring month, can also be used when creating continuous contracts.
Using TradeStation’s Default @CL
Using Wilder’s Parabolic SAR Algorithm
$50 levied for execution costs
Using Pinnacle Data’s Adjusted Data for CL
$50 levied for execution costs
The two equity curves are similar. However, TradeStation’s data produced nearly $60K in profit, whereas Pinnacle only produced north of $20K. I discovered in the early part of TradeStation’s @CL data, the values were derived strictly from the electronic markets. Up until about 2008 or so, I think the pits were still more active. Take a look at this graphic.
Nov-Dec 2004 – see how the pit data looks more valid
Pinnacle Data knits the pit session data with the purely electronic data during the transition period. However, this only accounts for the early trades in the back test. The parabolic indicator uses its prior output to calculate the next value. Any disruption in the differences in the relationships of historic highs and lows of the data will perpetuate through the output of subsequent parabolic calculations. Things start to work themselves out later in the back test. Indicators like the Parabolic are more sensitive to data differences.
What if you wanted to trade the Parabolic SAR real time?
This can be done, but results may look different than your back test. In trading crude oil, you should trade all 12 contracts. And you will need to execute roll spreads once a month, if you are in a position. This doesn’t answer the question on how to calculate the Parabolic once a rollover takes place. You have two options, 1.) you can use what Futures Truth coined as overlap data to calculate the new indicator reading. Basically, we would load the new contract data going back as far as we could and derive the indicator values off of the new data. This would work just fine with crude data, but many futures do not have a sufficient amount of overlap. In other words, the new contract would have very little history. And this data wasn’t sufficient to calculate accurate indicator values. 2.) Roll into the new contract and adjust the historical data of the old contracts by the discount between the two contracts that were rolled out of and into. This would alleviate the problems with the futures contract with very little history.
I like the second option best, because it works for all markets and even in crude oil you may not have overlap data to do the calculation. Even if you have enough data, it might not be all that good. If you can go back 100 days in the new contract, the data may deteriorate before you get those 100 days.
Data starts deteriorating well before 100 days back.
Wouldn’t it be great if we could test with this type of data? You know rollover and then adjust all the historic data that we need for our indicators using good data from the old contracts, but adjusted in terms of our new data? Well, we are in luck, I spent several hours doing this with my Python TS-18 software. I used Pinnacle Data‘s actual contract data – you get this when you buy their CLC database and derived the discount between the contracts. Pinnacle rolls at the same time each month in crude – it’s not a calendar date but a number of days into the prior expiration month. Take a look at this trade-by-trade listing.
Trade-by-trade with rollovers.
This shows rolling out of the old contract and into the new contract, and then backing up 100 days of old contract data and adjusting the data by the discounts as they occur. This could involve more than three adjustments historically. This rollover inspired data was then fed into the parabolic indicator. The indicator was restarted on each rollover going back N days. In other words, when a rollover was observed, the indicator was reset with the new rollover adjusted historic data by using a loop. The rollover mechanism called the parabolic calculation N times going back N days into our dynamically adjusted data. Once the loop finished, the new contract data was then fed into the calculation on day at a time. This occurred until another rollover was observed.
I would show the results of this testing but…
I got very diverse results when I recalculated the parabolic on each rollover with different loop counts. On the rollover, if I looped back 40 days to bring the parabolic up to speed, I got very similar results to TradeStation’s results using Pinnacle data. What the heck I will show that to you.
TS-18 Results with rollovers
The results are similar but the draw down of late concerns me. However, if I only go back 30 days to synchronize the parabolic on each rollover, the results are much better. This tells me that the parabolic is very sensitive to the data that is used for its calculations.
This post may have raised more questions than it answered. But I will try to get down to the differences between the different look back lengths on rollover and report any findings back here. However, you must keep rollovers in mind when dealing with trading futures, period.
Daily Bar Daytrade to 5 Minute Bars — Not So Easy!
Like in the previous post, I designed a simple optimization framework trying to determine the best day of the week to apply a long onlyvolatility based open range break out algorithm on natural gas. You might first ask why natural gas? Its a good market that has seasonal tendencies that can actually be day-traded – there is enough volatility and volume. Anyway, by simply slapping setExitOnClose at the end of your code turns your algorithm into one that exits every day at the settlement (fictional value that is derived via formula) price. You can always use LIB (look inside bar) to help determine the magnitude of intraday swings and there respective chronological order. But you are still getting out at the settlement price – which isn’t accurate. However, if you are trying to get a set of parameters that gets you into the ballpark, then this is a very acceptable approach. But you will always want to turn you daily bar strategy with LIB intro an intraday version for automation and accuracy purposes.
Day of Week Analysis On Intraday Data
When you optimize the day of week on daily bars, the first bar of the week is usually Monday. When testing with intraday bars on futures, the trading actually starts on Sunday evening. If your daily bar analysis reveals the best day of week is Wedneday, then you must inform TradeStation to take the trade from the opening on Tuesday night through the close on Wenesday. If you are executing a market order, then you just need to tell TradeStation to execute at Tuesday night’s open (1800 eastern standard time). The TRICK to allow trading throughout the 24 hour session (1800 to 1700) on just a particular day of the week is to capture the day of the week on the first bar of the trading session. Here is some bookkeeping you can take care of on the first bar of the trading session.
if t = sessionstartTime(0,1) + barInterval Then Begin
todaysOpen = o; openDOW = dayOfWeek(d); canBuy = False; if openDOW = daysOfWeekToTrade then canBuy = True; if canBuy and d[1] <> date of data2 then begin canBuy = False; print(d," turning off canBuy"); end;
if mp = 1 then curTradeDays = curTradeDays + 1; if mp = 0 then curTradeDays = 0; barsToday = 1; end;
Bookkeeping on the first bar of trading session
Here, I check to see if the first bar’s time stamp is equal to the sessionStarttime + barInterval. Remember TradeStation shows the close time of each bar as the time stamp. If adding a barInterval to the opening time results in a non-time, then use the calcTime function ( 1800 + 60 = 1860 – a non-time.) I first store the open of the session in my variable todaysOpen. Why not just use openD(0). Well openD(0) is great for sessions that don’t span midnight. If they span midnight then the openD returns the open of the12:00 am bar. Here is the output from August 9th with the open on August 8th at 1800. The two other values are midnight on August 8th and August 7th. So the openD, highD, lowD, and closeD functions involve the 24 hour clock and not the session.
On the first bar of the trading session I also capture the day of the week. The days of the week, using intraday data on a 24 hour session, range from Sunday to Thursday, instead of Monday thru Friday. CanBuy is turned on if the day of week equals the input provided by the user. When using daily bars and you optimize the day of the week, remember if you chose 2 or Tuesday you really mean Wednesday. If today is Tuesday, then buy next bar at… This will generate trades only on Wednesdays. When testing on intraday data you do not need to make a modification for the day of week. If the first bar of the trading session is Tuesday, then you will actually trade what is considered the Wednesday session that starts on Tuesday evening at 18:00.
What About Holidays?
Here is a quick trick to not trade on Holidays. The daily bar does not include preempted sessions in its database, whereas intraday data does. So, if at 18:00 the date of the prior bar does not match the date of the daily bar located in the Data2 slot, then you know that the prior day was preempted and you should not trade today. In other words, Data2 is missing. Remember, you only want to trade on days that are included in the daily bar database in your attempt to replicate the daily bar strategy.
date close moving average
1230703.00 data2 2.57382 2.70300
1230704.00 missing data2 2.57382 2.70300 <-- no trade
1230705.00 data2 2.57485 2.65100 <-- matches
How to count the daily bars since entry.
Here again you need to do some bookkeeping. If you are long on the first bar of the day, then you have been in the trade for at least a day. If you are long on two first bars of the trading session, then you have been long for two days. I use the variable, curTradeDays, to count the number of days we have been in the trade. If on the first bar of the trading session we are flat, then I reset curTradeDays to zero. Otherwise, I increment the variable and this only occurs once a day. You can optimize the daysInTrade input, but for this post we are just interested in getting out at the close on the day of entry.
Controlling one entry per day.
We only want one long entry per day in an attempt to replicate the daily bar system. There are several ways to do this, but comparing the current bar’s marketPosition to the prior bar’s value will inform you if a position has been transitioned from flat to long. If the prior bar’s position if flat and the current bar is now long, then we can turn our canBuy off or to false. If we get stopped out later and the market rallies again to our break out level, we will not execute at that level, because the trade directive will not be issued. Why not use the entriesToday function? Again midnight cause this function to reset. Say you go long at 22:00 and you do not want to enter another position until the tomorrow at 18:00. EntriesToday will forget you entered the long position at 22:00.
Cannot execute on the first bar of the day – does it matter?
Since this is an open range break out, we need to know what the open is prior to our break out calculation. Could the break out level be penetrated on the first 5 minute bar? Sure, but I bet it is rare. I tested this and it only occurred 4 or 5 times. In a back-test, if this occurs and the subsequent bar’s open is still above the break out level, TradeStation will convert the stop order to a market order and you will buy the open of the 2nd five minute bar. In real time trading, you must tell TradeStation to wait until the end of the first 5 minute bar before issuing the trade directive to replicate your back testing. If this is an issue, you can test with 5 minute bars, but execute on 1 minute bars. In other words, have your testing chart and an additional execution chart. Skipping the first 5 minute bar may be advantageous. Many times you will get a surge at the open and then the market falls back. By the time the first 5 minute bar concludes the market may be trading back below your break out level. You might bypass the bulge. But your still in the game. Here is how you can test the significance of this event.
if canBuy and h > todaysOpen + volAmt *volMult then Begin print(d," first bar penetration ",value67," ",value68," ",(c - (todaysOpen + volAmt *volMult))*bigPointValue ); value67 = value67 + 1; value68 = value68 + (c - (todaysOpen + volAmt *volMult))*bigPointValue; end;
Testing for penetration on first 5 minute bar of the day
By the end of the first bar of the day we know the open, high, low and close of the day thus far. We can test to see if the high would have penetrated the break out level and also calculate the profit or loss of the trade. In a back-test you will not miss this trade if the next bar’s open is still above the break out level. If the next bar’s open is below the break out level, then you may have missed a fake out break out. Again this is a rare event.
if t = sessionstartTime(0,1) + barInterval Then Begin todaysOpen = o; openDOW = dayOfWeek(d); canBuy = False; if openDOW = daysOfWeekToTrade then canBuy = True; if canBuy and d[1] <> date of data2 then begin canBuy = False; end;
if mp = 1 then curTradeDays =curTradeDays + 1; if mp = 0 then curTradeDays = 0; barsToday = 1;
volAmt = average(range of data2,volLen); movAvg = average(c,mavLen) of data2; if canBuy and h > todaysOpen + volAmt *volMult then Begin print(d," first bar penetration ",value67," ",value68," ",(c - (todaysOpen + volAmt *volMult))*bigPointValue ); value67 = value67 + 1; value68 = value68 + (c - (todaysOpen + volAmt *volMult))*bigPointValue; end; end;
if mp[1] = 0 and mp = 1 and canBuy then canBuy = False;
if canBuy and t <> sessionEndTime(0,1) Then Begin if barsToday > 1 and close of data2 > movAvg and range of data2 > volAmt * volComp then buy("DMIntraBuy") next bar at todaysOpen + volAmt *volMult stop; end; barsToday = barsToday + 1;
if daysInTrade = 0 then setExitOnClose;
sell("LLxit") next bar at lowest(l of data2,llLookBack) stop; if mp = 1 and daysInTrade > 0 then begin if curTradeDays = daysInTrade then sell("DaysXit") next bar at open; end;
Intrday system that replicates a daily bar day trader
We are using the setStopLoss and setProfitTarget functions in this code. But remember, if your continuous contract goes negative, then these functions will not work properly.
This type of programming is tricky, because you must use tricks to get EasyLanguage to do what you want it to do. You must experiment, debug, and program ideas to test the limitations of EasyLanguage to hone your craft.
When Mining for Data, Make Sure You Have Accurate Data
Take a look at these results with no execution costs.
Results of Data Mining on Natural Gas
These results were derived by applying a simple algorithm to the natural gas futures for the past 15 or so years. I wanted to see if there was a day of the week that produced the best results with the following entry and exit techniques:
Long entries only
Open range break out using a fraction of the N-day average range.
Buy in the direction of the moving average.
Yesterdays close must be above the X-day moving average of closing prices.
Yesterday must have a wider range than a fractional multiple of the average range.
A fixed $stop and $profit used to exit trades.
Other exits
Exit at low of prior day.
Exit at the close of today – so a day trade.
Here is a list of the parameters that can be optimized:
daysOfWeekToTrade(1) : 1 – Monday, 2 – Tuesday…
mavLen(20): moving average calculation length.
volLen(10): moving average calculation length for range.
volMult(0.25): average range multiplier to determine break out/
volComp(.5): yesterday’s range must be greater than this percentage.
stopLoss(500): stop loss in $
profitTarget(1000): profit objective in $
daysInTrade(0): 0 get out at end of day.
llLookBack(2): pattern derived stop value – use the lowest low of N-days or the stop loss in $, whichever is closer.
if dayOfWeek(d) = daysOfWeekToTrade Then Begin if close > average(c,mavLen) and range > average(range,volLen)*volComp then buy("DMDailyBuy") next bar at open of tomorrow + average(range,volLen)*volMult stop; end;
if daysInTrade = 0 then setExitOnClose; sell("LLxit") next bar at lowest(l,llLookBack) stop; if marketPosition = 1 then begin if barsSinceEntry = daysInTrade then sell("DaysXit") next bar at open; end;
Why Is This Algorithm Wrong? It’s Not It’s the Data!
The algorithm, speaking in terms of logic, is accurate. The data is wrong. You cannot test the exit technology of this algorithm with just four data points per day – open, high, low, and close. If this simple algorithm cannot be tested, then what can you test on a daily bar basis accurately?
Long or short entry, but not both.
If the algorithm can enter both long and short, you need to know what occurred first. Doesn’t matter if you enter via stop or limit orders. Using a stop order for both, then you need to know if the high occurred first and got you long, and then later the low and got you short. Or just the opposite. You can test, with the benefit of hindsight, if only one of the orders would have been elected. If only one is elected, then you can proceed. If both then no-go. You must be able to use future leak to determine if only one of the orders were fillable. TS-18 allows future leak BTW. I say that like it’s a good thing!
L or S position from either a market or stop, and a profit objective.
If a long position is entered above the open, via a stop, and then the market moves even higher, you can get out on a sell limit for a profit.
Same goes for the short side, but you need to know the magnitude of the low price in relation to the open.
L or S position from either a market or limit and a protective stop.
If short position is entered above the open, via a limit order, and the market moves even higher, you can exit the short position at a loss via a buy stop order.
Same goes for the long side, but you need to know the magnitude of the high in relation to the open.
Long pyramid on multiple higher stop or lower limit orders, but not both.
The market opens and then sells off. You can go long on a limit order below the open, then you go long another unit below the original limit price and so on…
The market opens and rallies. You can go long on a stop order above the open, then you can go long another unit above the original stop price and so on…
Short pyramid on multiple lower stop or higher limit orders.
Same as long entries but in opposite direction.
Can’t you look at the relationships of the open to low and open to high and close to high and close to low and determine which occurred first, the high or the low of the day. You can, but there is no guarantee. And you can’t determine the magnitude of intraday day swings. Assume the market opens and immediately moves down and gets you short via a stop order. And from looking at a daily chart, it looks like the market never turned around, so you would assume you had a profit going into the close. But in fact, after the market opened and moved down, it rallied back to the open enough to trigger a protective stop you had working.
The entry technique of this strategy is perfectly fine. It is only buying in the direction of a breakout. The problems arise when you apply a profit objective and a stop loss and a pattern-based stop and a market on close order. After buying did the market pull back to get you stopped out before moving up to get you out at profit objective? Did the market move up and then down to the prior lowest low of N days and then back up to get you long? Or just the opposite? In futures, the settlement price is calculated with a formula. You are always exiting at the settlement price with an MOC or setExitOnClose directive (daily bar basis.)
No Biggie – I Will Just Test with LIB (Look Inside Bar. That should fix it, right?
It definitely will increase accuracy, because you can see the intraday movements that make up the daily bar. So, your entries and exits should be executed more accurately. But you are still getting out at the settlement price which does not exist. Here are the results from using LIB with 5-minute bar resolution.
You saw the beauty now you see the beast.
We Know the Weakness, But What Can We Do?
Test on a 5-minute bar as Data1 and daily as Data2. This concept is what my Hi-Res Edition of Easing Into EasyLanguage is all about. Here the results of using a higher resolution data and exiting on the last tick of the trading day – a known quantity.
As accurate as you can get – well with 5 minute bars that is.
These results validate the LIB results, right? Not 100%, but very close. Perhaps this run makes more money because the settlement price on the particular days that the system entered a trade was perhaps lower than the last tick. In other words, when exiting at the end of the day, the last tick was more often higher than the settlement price.
In my next post, I will go over the details of developing an intraday system to replicate (as close as possible) a simple daily bar-based day trading system. Like the one we have here.
A study between ice cream sales and crime rate demonstrated a high level of correlation. However, it would be illogical to assume that buying more ice cream leads to more crime. There are just too many other factors and variables involved to draw a conclusion. So, data mining with EasyLanguage may or may not lead to anything beneficial. One thing is you cannot hang your hat completely on this type of research. A reader of my books asked if there was evidence that pointed to the best time to enter and exit a day trade. Is it better to enter in the morning or in the afternoon or are there multiple trading windows throughout the day? I thought I would try to answer the question using TradeStation’s optimization capabilities.
Create a Search Space of Different Window Opening Times and Open Duration
My approach is just one of a few that can be used to help answer this question. To cut down on time and the size of this blog we will only look at day trading the @ES.D from the long side. The search space boundaries can be defined by when we open the trading window and how long we leave it open. These two variables will be defined by inputs so we can access the optimization engine. Here is how I did it with EasyLanguage code.
if t >= calcTime(openWindowTime,openWindowOffset) and t < calcTime(openWindowTime,openWindowOffset+windowDuration) Then Begin if entriesToday(d) = 0 and canBuy Then buy next bar at market; if entriesToday(d) = 0 and canShort Then sellshort next bar at market ; end;
if t = calcTime(openWindowTime,openWindowOffset+windowDuration) Then Begin if marketPosition = 1 then sell next bar at open; if marketPosition =-1 then buyToCover next bar at open; end; setExitOnClose;
Optimize when to open and how long to leave open
The openWindowTime input is the basis from where we open the trading window. We are working with the @ES.D with an open time of 9:30 AM eastern. The openWindowOffset will be incremented in minutes equivalent to the data resolution of the chart, five minutes. We will start by opening the window at 9:35 and leave it open for 60 minutes. The next iteration in the optimization loop will open the window at 9:40 and keep it open for 60 minutes as well. Here are the boundaries that I used to define our search space.
window opening times offset: 5 to 240 by 5 minutes
window opening duration: 60 to 240 by 5 minutes
Optimization Ranges for Window Open and Open Duration
A total of 1739 iterations will span our search space. The results state that waiting for twenty minutes before buying and then exiting 190 minutes later, worked best. But also entering 90 minutes after the open and exiting 4 hours later produced good results as well (no trade execution fee were utilized.) Initially I was going to limit entry to once per day, but then I thought it might be worthwhile to enter a fresh position, if the first one is stopped out, or pyramid if it hasn’t. I also thought, each entry should have its own protective stop amount. Would entering later require a small stop – isn’t most of the volatility, on average, expressed during the early part of the day.
Results of different opening and duration times.
Build a Strategy that Takes on a Secondary Trade as a New Position or One that is Pyramided.
This is not a simple strategy. It sounds simple and it requires just a few lines of code. But there is a trick in assigning each entry with its own exit. As you can see there is a potential for trade overlap. You can get long 20 minutes after the open and then add on 70 minutes (90 from the open) later. If the first position hasn’t been stopped out, then you will pyramid at the second trade entry. You have to tell TradeStation to allow this to happen.
Allow TradeStation to Pyramid up to 2 positions with different signals.
System Rules
Enter long 20 minutes after open
Enter long 90 minutes after open
Exit 1st entry 190 minutes later or at a fixed $ stop loss
Exit 2nd entry 240 minutes later or at a fixed $ stop loss
Make sure you are out at the end of the day
Sounds pretty simple, but if you want to use different stop values for each entry, then the water gets very muddy.
AvgEntryPrice versus EntryPrice
Assume you enter long and then you add on another long position. If you examine EntryPrice you will discover that it reflects the initial entry price only. The built-in variable AvgEntryPrice will be updated with the average price between the two entries. If you want to key off of the second entry price, then you will need to do a little math.
Using this formula and simple algebra we can arrive at ep2 using this formula: ep2 = 2*ap – ep1. Since we already know ep1 and ap, ep2 is easy to get to. We will need this information and also the functionality of from entry. You tie entries and exits together with the keywords from entry. Here are the entry and exit trade directives.
if time = calcTime(openTime,entryTime1Offset) then buy("1st buy") next bar at open;
if time = calcTime(openTime,entryTime2Offset) then buy("2nd buy") next bar at open;
if time = calcTime(openTime,entryTime1Offset + exitTime1Offset) then sell("1st exit") from entry("1st buy") next bar at open;
if time = calcTime(openTime,entryTime2Offset + exitTime2Offset) then sell("2nd exit") from entry("2nd buy") next bar at open;
if mp = 1 Then Begin value1 = avgEntryPrice; if currentShares = 2 then value1 = avgEntryPrice*2 - entryPrice; sell("1st loss") from entry("1st buy") next bar at entryPrice - stopLoss1/bigPointValue stop; sell("2nd loss") from entry("2nd buy") next bar at value1 - stopLoss2/bigPointValue stop; end;
if mp = 1 and t = openTime + barInterval then sell("oops") next bar at open;
Entry and Exit Directives Code
The trade entry directives are rather simple, but you must use the calcTime function to arrive at the correct entry and exit times. Here we are using the benchmark, openTime and the offsets of entryTime1Offset and entryTime2Offset. This function adds (or subtracts if the offset is negative) the offset to the benchmark. This takes care of when the trading windows open, but you must add the entry1TimeOffset to exit1TimeOffset to calculate the duration the trading window is to remain open. This goes for the second entry window as well.
Now let’s look at the exit directives. Notice how I exit the 1st buy entry with the code from entry (“1st buy”). This ties the entry and exit directives together. This is pretty much straightforward as well. The tricky part arrives when we try to apply different money management stops to each entry. Exiting from the 1st buy requires us to simply subtract the $ in terms of points from entryPrice. We must use our new equation to derive the 2nd entry price when two contracts are concurrent. But what if we get stopped out of the first position prior to entering the second position? Should we continue using the formula. No. We need to fall back to entryPrice or avgEntryPrice: when only one contract or unit is in play, these two variables are equal. We initially assign the variable value1 to the avgEntryPrice and only use our formula when currentShares = 2. This code will work a majority of the time. But take a look at this trade:
Didn’t have an intervening bar to update the 2nd entry price. The first entry price was used as the basis to calculate the stop loss!
This is an anomaly, but anomalies can add up. What happened is we added the second position and the market moved down very quickly – too quickly for the correct entry price to be updated. The stop out (2nd loss) was elected by using the 1st entry price, not the second. You can fix this with the following two solutions:
Increase data resolution and hope for an intervening bar
Force the second loss to occur on the subsequent trading bar after entry. This means you will not be stopped out on the bar of entry but will have to wait five minutes or whatever bar interval you are working with.
Fixed – just told TS to place the order at the close of the bar where we were filled!
OK – Now How Do We Make this a Viable Trading System
If you refer back to the optimization results you will notice that the average trade (before execution costs) was around $31. Keep in mind we were trading every day. This is just the beginning of your research – did we find a technical advantage? No. We just found out that you can enter and exit at different times of the trading day, and you can expect a positive outcome. Are there better times to enter and exit? YES. You can’t trade this approach without adding some technical analysis – a reason to enter based on observable patterns. This process is called FILTERING. Maybe you should only enter after range compression. Or after the market closed up on the prior day, or if the market was an NR4 (narrow range 4.) I have added all these filters so you can iterate across them all using the optimization engine. Take a look:
filter1 = True; filter2 = True;
if filtNum1 = 1 then filter1 = close of data2 > close[1] of data2; if filtNum1 = 2 Then filter1 = close of data2 < close[1] of data2; if filtNum1 = 3 Then filter1 = close of data2 > open of data2; if filtNum1 = 4 Then filter1 = close of data2 < open of data2; if filtNum1 = 5 Then filter1 = close of data2 > (h data2 + l data2 + c data2)/3; if filtNum1 = 6 Then filter1 = close of data2 < (h data2 + l data2 + c data2)/3; if filtNum1 = 7 Then filter1 = openD(0) > close data2; if filtNum1 = 8 Then filter1 = openD(0) < close data2;
if filtNum2 = 1 Then filter2 = trueRange data2 < avgTrueRange(10) data2; if filtNum2 = 2 Then filter2 = trueRange data2 > avgTrueRange(10) data2; if filtNum2 = 3 Then filter2 = range data2 = lowest(range data2,4); if filtNum2 = 4 Then filter2 = range data2 = highest(range data2,4);
Filter1 and Filter2 - filter1 looks for a pattern and filter2 seeks range compression/expansion.
Let’s Search – And Away We Go
I will optimize across the different patterns and range analysis and different $ stops for each entry (1st and 2nd.)
Optimize across patterns and volatility and protective stops for 1st and 2nd entries.
Best Total Profit
Trade when today’s open is greater than yesterday’s close and don’t worry about the volatility. Use $550 for the first entry and $600 for the second.
Best W:L Ratio and Respectable Avg. Trade
This curve was created by waiting for yesterday’s close to be below the prior day’s and yesterday being an NR4 (narrow range 4). And using a $500 protective stop for both the 1st and 2nd entries.
Did We Find the Holy Grail? Gosh No!
This post served two purposes. One, how to set up a framework for data mining and two, create code that can handle things that aren’t apparently obvious – entryPrice versus avgEntryPrice!
if filter1 and filter2 Then begin if time = calcTime(openTime,entryTime1Offset) then buy("1st buy") next bar at open;
if time = calcTime(openTime,entryTime2Offset) then buy("2nd buy") next bar at open; end;
Incorporating Filter1 and Filter2 in the Entry Logic
if currentbar>=1 then if oscVal>oscVal[1] then plot1(mavDiff,"+AO") else plot2(mavDiff,"-AO")
Williams Awesome Oscillator Source Code
And here is what it looks like:
Williams Awesome Oscillator
The code reveals a value that oscillates around 0. First calculate the difference between the 5-day moving average of the daily midPoint (H+ L)/2 and the 34-day moving average of the midPoint. A positive value informs us that the market is in a bullish stance whereas a negative represents a bearish tone. Basically, the positive value is simply stating the shorter-term moving average is above the longer term and vice versa. The second step in the indicator calculation is to subtract the 5-day moving average of the differences from the current difference. If the second calculation is greater than the prior day’s calculation, then plot the original calculation value as green (AO+). If it is less (A0-), then paint the first calculation red. The color signifies the momentum between the current and the five-day smoothed value.
Here I am using the very handy countIf function. This function will tell you how many times a Boolean comparison is true out of the last N days. Her I use the function twice, but I could have replaced the second function call with mavDn = 30 – mavUp. So, I am counting the number of occurrences of when the mavDiff is positive and negative over the past 30-days. I also count the number of times the oscVal is greater than the prior oscVal. In other words, I am counting the number of green bars. I create a ratio between green bars and 10. If there are six green bars, then the ratio equals 60% This indicates that the ratio of red bars would be 40%. Based on these readings you can create trade entry directives.
if canShort and mavUp > numBarsAbove and mavDiff > minDiffAmt and oscRatio >= obRatio then sellShort next bar at open;
if canBuy and mavDn > numBarsBelow and mavDiff < -1*minDiffAmt and oscRatio <= osRatio Then buy next bar at open;
Trade Directives
If the number of readings out of the last 30 days is greater than numBarsAbove and mavDiff is of a certain magnitude and the oscillator ratio is greater than buyOSCRatio, then you can go short on the next open. Here we are looking for the market to converge. When these conditions are met then I think the market is overbought. You can see how I set up the long entries. As you can see from the chart it does a pretty good job. Optimizing the parameters on the crude oil futures yielded this equity curve.
Too few trades!
Not bad, but not statistically significant either. One way to generate more trades is to install some trade management such as protective stop and profit objective.
Using wide stop and profit objective.
Using a wide protective stop and large profit objective tripled the number of trades. Don’t know if it is any better, but total performance was not derived from just a couple of trades. When you are working with a strategy like this and overlay trade management you will often run into this situation.
Exiting while conditions to short are still turned on!
Here we either get stopped out or take a profit and immediately reenter the market. This occurs when the conditions are still met to short when we exit a trade. The fix for this is to determine when an exit has occurred and force the entry trigger to toggle off. But you have to figure out how to turn the trigger back on. I reset the triggers based on the number of days since the triggers were turned off – a simple fix for this post. If you want to play with this strategy, you will probably need a better trigger reset.
I am using the setStopLoss and setProfitTarget functionality via their own strategies – Stop Loss and Profit Target. These functions allow exit on the same as entry, which can be useful. Since we are executing on the open of the bar, the market could definitely move either in the direction of the stop or the profit. Since we are using wide values, the probability of both would be minimal. So how do you determine when you have exited a trade. You could look the current bar’s marketPosition and compare it with the prior bar’s value, but this doesn’t work 100% of the time. We could be flat at yesterday’s close, enter long on today’s open and get stopped out during the day and yesterday’s marketPosition would be flat and today’s marketPosition would be flat as well. It would be as if nothing occurred when in fact it did.
Take a look at this code and see if it makes sense to you.
if mp[1] = 1 and totalTrades > totTrades then canBuy = False;
if mp[1] = -1 and totalTrades > totTrades then canShort = False;
if mp[1] = 0 and totalTrades > totTrades then Begin if mavDiff[1] < 0 then canBuy = False; if mavDiff[1] > 0 then canShort = False; end;
totTrades = totalTrades;
Watch for a change in totalTrades.
If we were long yesterday and totalTrades (builtin keyword/function) increases above my own totTrades, then we know a trade was closed out – a long trade that is. A closed out short position is handled in the same manner. What about when yesterday’s position is flat and totalTrades increases. This means an entry and exit occurred on the current bar. You have to investigate whether the position was either long or short. I know I can only go long when mavDiff is less than zero and can only go short when mavDiff is greater than zero. So, all you need to do is investigate yesterday’s mavDiff to help you determine what position was entered and exited on the same day. After you determine if an exit occurred, you need to update totTrades with totalTrades. Once you determine an exit occurred you turn canBuy or canShort off. They can only be turned back on after N bars have transpired since they were turned off. I use my own barsSince function to help determine this.
if not(canBuy) Then if barsSince(canBuy=True,100,1,0) = 6 then canBuy = True; if not(canShort) Then if barsSince(canShort=True,100,1,0) = 6 then canShort = True;
if not(canBuy) Then if barsSince(canBuy=True,100,1,0) = numBarsTrigReset then canBuy = True; if not(canShort) Then if barsSince(canShort=True,100,1,0) = numBarsTrigReset then canShort = True;
if mp[1] = 1 and totalTrades > totTrades then canBuy = False;
if mp[1] = -1 and totalTrades > totTrades then canShort = False;
if mp[1] = 0 and totalTrades > totTrades then Begin if mavDiff[1] < 0 then canBuy = False; if mavDiff[1] > 0 then canShort = False; end;
if canShort and mavUp > numBarsAbove and mavDiff > minDiffAmt and oscRatio >= buyOSCRatio then sellShort next bar at open;
if canBuy and mavDn > numBarsBelow and mavDiff < -1*minDiffAmt and oscRatio <= shortOSRatio Then buy next bar at open;
The concept of Scope when it comes to a programming language can get very confusing very quickly. If you are new to programming and you think Scope is a mouthwash, no worries here. I will give a full explanation of Scope in terms of programming with the help of ChatGPT. Its response is italicized.
In Python, the scope of a variable refers to the region of the program where that variable can be accessed or referenced. There are two types of scopes in Python: global scope and local scope.
Here’s an example that demonstrates the concept of scope in Python:
# global variable x = 10
def my_func(): # local variable y = 5 print("x inside my_func:", x) # x is accessed from the global scope print("y inside my_func:", y) # y is accessed from the local scope
my_func()
print("x outside my_func:", x) # x is accessed from the global scope print("y outside my_func:") # y is not defined in the global scope, so this will raise a NameError
Example of Scope as demonstrated by Chat-GPT
In this example, we define a global variable x with a value of 10.
We then define a function my_func that has a local variable y with a value of 5. Inside the function, we print the values of x and y. The variable x is accessed from the global scope, which means it can be referenced from anywhere in the program.
The variable y, on the other hand, is only accessible within the function, which is the local scope. Outside the function, we print the value of x again, which is still accessible from the global scope. However, when we try to print the value of y, we get a NameError because y is not defined in the global scope.
Okay that was pretty formal. In simpler terms, a variable declared inside a function can only be seen inside that function, unless it is declared globally. Another thing about Scope, is that the variables have amnesia. Once you step out of the function the variable forgets what it was, and you can’t refer to its prior value upon return to the function.
Unless you are an EasyLanguage function. Scope is still constrained by an EasyLanguage function, because it is a separate piece of code held within the library of Analysis Techniques. Most of the time you can’t see what’s in the function unless you open it with the ELEditor. However, the variables that are defined inside the function do not suffer from amnesia. If you need to refer to a prior value of a locally declared variable, you can. This type of function is what EasyLanguage calls a Serial function. The only downside to this function is it slows processing down quite a bit.
Okay. To make a long story short I wanted to show the magic of EasyLanguage function that I have been working with on a project. This project includes some of Ehlers’ cycle analysis functions. The one I am going to discuss today is the HighRoof function – don’t worry I am not going to go into detail of what this function does. If you want to know just GOOGLE it or ask ChatGPT. I developed a strategy that used the function on the last 25 days of closing price data. I then turned around and fed the output of the first pass of the HighRoof function right back into the HighRoof function. Something similar to embedding functions.
doubleSmooth = average(average(c,20),20);
Sort of like a double smoothed moving average. After I did this, I started thinking does the function remember the data from its respective call? The first pass used closing price data, so its variables and their history should be in terms of price data. The second pass used the cyclical movements data that was output by the initial call to the HighRoof function. Everything turned out fine, the function remembered the correct data. Or seemed like it did. This is how you learn about any programming language – pull out your SandBox and do some testing. First off, here is my conversion of Ehlers’ HighRoof function in EasyLanguage.
This function requires just two inputs – the data (with a history) and a simple length or cut period. The first input is of type numericSeries and the second input is of type numericSimple. You will see the following line of code
This code prints out the last three historic values of the HighPass variable for each function call. I am calling the function twice for each bar of data in the Crude Oil futures continuous contract.
Starting at the top of the output you will see that on 1230206 the function was called twice with two different sets of data. As you can see the output of the first two lines is of a different magnitude. The first line is approximately an order or magnitude of 10 of the second line. If you go to lines 3 and 4 you will see the highPass[1] of lines 1 and 2 moves to highPass[2] and then onto highPass[3]. I think what happens internally is for every call on per bar basis, the variables for each function call are pushed into a queue in memory. The queue continues to grow for whatever length is necessary and then either maintained or truncated at some later time.
Why Is This So Cool?
In many languages the encapsulation of data with the function requires additional programming. The EasyLanguage function could be seen as an “object” like in object-oriented programming. You just don’t know you are doing it. EasyLanguage takes care of a lot of the behind-the-scenes data management. To do the same thing in Python you would need to create a class of Ehlers Roof that maintain historic data in class members and the calculations would be accomplished by a class method. In the case of calling the function twice, you would instantiate two classes from the template and each class would act independent of each other.
One last nugget of information. If you are going to be working with trigonometric functions such as Cosine, Sine or Tangent, make sure your arguments are in degrees not radians. In Python, you must use radians.
This pattern has been around for many years, and is still useful today in a day trading scheme. The pattern is quite simple: if today’s high exceeds yesterday’s high by a certain amount, then sell short as the market moves back through yesterday’s high. There are certain components of yesterday’s daily bar that are significant to day traders – the high, the low, the close and the day traders’ pivot. Yesterday’s high is considered a level of resistance and is often tested. Many times the market has just enough momentum to carry through this resistance level, but eventually peters out and then the bears will jump in and push the market down even more. The opposite is true when the bulls take over near the support level of yesterday’s low. Here is an example of Clear Out short and buy.
1st the high of yesterday is cleared out and then the low of yesterday.
How Do You Program this Simple Pattern?
The programming of this strategy is rather simple, if you are day trading. The key components are toggles that track the high and low of the day as the market penetrate the prior day’s high and low. Once the toggles are flipped on, then order directives can be placed. A max. trade stop loss can easily be installed via the SetStopLoss(500) function. You will also want to limit the number of entries, because in a congestive phase, this pattern could fire off multiple times. Once you intuitively program this, you will almost certainly run into an issue where a simple “trick” will bail you out. Remember the code does exactly what you tell it to do. Take a look at these trades.
When Back Testing TradeStation will convert stop orders to market orders.
On a Back Test, Stop Orders are Converted to Market Orders if Price Exceeds the Stop Level
In these example trades, the first trade is accurate as it buys yesterday’s low + one tick and then gets stopped out. Once a long is entered, the system logic requires the market to trade back below yesterday’s low before a long another entry is signaled at yesterday’s low. Here as you can see, the initial buy toggle is set to True and when a long position is entered the buy toggle is turned off. The market knee jerks back below yesterday’s low and stops out your long position. Since TradeStation’s paradigm is based on “next bar” execution, a long entry doesn’t occur as the wide bar crosses back up through yesterday’s low. This is a “bang-bang” situation as it happened very quickly. In a perfect world, you should have been quickly stopped out and re-entered back long at your price. However, the toggle isn’t turned back on until the low of the current bar falls a short distance below yesterday’s low. Since this toggle isn’t set before the market takes off, you don’t get your price. The toggle is eventually turned on and a buy stop order is issued and you can tell you get a ton of slippage. You actually buy the next bar’s open after the bar where the toggle was turned on. I dropped down to a one minute bar and still didn’t get the trade. A 10 second bar did generate the exit and re-entry at the correct levels, however. It did this because, the 10 second bar turned the toggle on in time for the stop order to be generated accurately.
Using a 10-second bar an accurate exit and entry were generated.
Okay – Can you rely on a 5 minute bar then?
Five minute bar data has been the staple of day trading systems for many years. However, if you want to test “bang-bang” algorithms you are probably better off dropping down to a N-seconds bar. However, this strategy as a whole is not “bang-bang” so with a little trick you can get more accurate entries and exits.
What’s the Trick?
In real-time trading, buy-stop orders below the market are rejected. So, the second and third trades that were presented would never have taken place. But, the backtest reflects the trades, and if you include execution costs, the performance might nudge you into not trading a possibly viable system. You can take advantage of the “next bar” paradigm by forcing the close of the current bar to be below a buy-stop price and above a sell short stop price. Does this trade look better? Again in a perfect world, you would have re-entered long on the wide bar that stopped us out. But I guarantee you a fast market condition was in effect. All a broker has to say to you when you complain about a fill is, “Sorry Dude! It was a fast market. Not held!” I can’t tell you how many times I requested a printout of fills over a few seconds from my brokers. It is like when a football coach tosses the RED FLAG. During the Pit Days you had a chance to get a fill cash adjustment because the broker was human and maybe he or she didn’t react quickly enough. But when electronic trade matching took over, an adjustment was highly unlikely. Heck, you sign off on this when you accept the terms of electronic trading. Fills are rarely made better.
The second and third trade don’t occur because you force the buy stop order to be valid.
How Does the Trick Affect Performance?
Here are the results over the past four months on different time frame resolutions.
10 Seconds Resolution.
10 seconds bar would be the most accurate if slippage is acceptable. And that is a big assumption on “bang-bang” days.
1 minute bar resolution.
The one minute bar is close but September is substantially off. Probably some “bang-bang” action.
5 minute bar resolution with “Trick”
This is close to the 10-second bar result. Fast market or “bang-bang” trades were reduced or eliminated with the “trick”.
5 minute bar resolution without “Trick.”
Surprisingly, the 5 minute bar without the “Trick” resembles the 10 seconds results. But we know this is not accurate as trades are fired off in a manner that goes against reality.
The two following table shows the impact of a $15 RT comm./slipp. per trade charge.
Without “Trick” and $15 RT
With “Trick” and $15 RT
Okay, Now That We Have That Figured Out How Do You Limit Trading After a Daily Max. Loss
Another concept I wanted to cover was limiting trades after a certain loss level was suffered on a daily basis. In the code, I only wanted to enter trades as long as the max. daily loss was less than or equal to $1,000 A fixed stop of $500 on a per trade basis was utilized as well. So, if you suffered two max. stop losses right off the bat ($1,000), you could still take one more trade. Now if you had a $500 winner and two $500 losses you could still take another trade.
Ouch! Two max losses, but still could take a third trade. Ouch again – stupid system.
Should I take that second trade? I just suffered three losses in a row. What to do? What to do? Damn straight you better that trade.
If you are going to trade a system, you better trade it systematically!
Now Onto the Code
//Illustrate trade stoppage after a certain loss has been //experienced and creating realistic stop orders.
if t = startTime then begin coBuy = False; coShort = False; beginOfDayProfit = netProfit; beginOfDayTotTrades = totalTrades; end;
canTrade = iff(t >=startTime and t < sess1EndTime,1,0);
if t >= startTime and h > highD(1) + clrOutBuyPer*(highD(1)-lowD(1)) then begin coShort = True; end;
if t >= startTime and l < lowD(1) - clrOutShortPer*(highD(1)-lowD(1)) then begin coBuy = True; end;
mp = marketPosition;
if canTrade = 1 and coShort and netProfit >= beginOfDayProfit - maxDailyLoss and c > highD(1) - minMove/priceScale then sellShort next bar at highD(1) - minMove/priceScale stop;
if mp = -1 then // toggle to turn off coShort - must wait for set up begin coShort = False; end;
if canTrade = 1 and coBuy and netProfit >= beginOfDayProfit - maxDailyLoss and c < lowD(1) + minMove/priceScale then buy next bar at lowD(1) + minMove/priceScale stop;
if mp = 1 then begin coBuy = False; end;
setStopLoss(500); setExitOnClose;
Strategy in its Entirety
You need to capture the NetProfit sometime during the day before trading commences. This block does just that.
if t = startTime then begin coBuy = False; coShort = False; beginOfDayProfit = netProfit; beginOfDayTotTrades = totalTrades; end;
Snippet that captures NetProfit at start of day
Now all you need to do is compare the current netProfit (EL keyword) to the beginOfDayProfit (user variable). If the current netProfit >= beginOfDayProfit – maxDailyLoss (notice I programmed greater than or equal to), then proceed with the next trade. The rest of the logic is pretty self explanatory, but to drive the point home, here is how I make sure a proper stop order is placed.
if canTrade = 1 and coShort and netProfit >= beginOfDayProfit - maxDailyLoss and c > highD(1) - minMove/priceScale then sellShort next bar at highD(1) - minMove/priceScale stop;
if mp = -1 then // toggle to turn off coShort - must wait for set up begin coShort = False; end;
Notice how I use the current bars Close - C and How I toggle coShort to False
If You Like This – Make Sure You Get My Hi-Res Edition of Easing Into EasyLanguage
This is a typical project I discuss in the second book in the Easing Into EasyLanguage Trilogy. I have held over the BLACK FRIDAY special, and it will stay in effect through December 31st. Hurry, and take advantage of the savings. If you see any mistakes, or just want to ask me a question, or have a comment, just shoot me an email.
Complete Strategy based on Sheldon Knight and William Brower Research
In my Easing Into EasyLanguage: Hi-Res Edition, I discuss the famous statistician and trader Sheldon Knight and his K-DATA Time Line. This time line enumerated each day of the year using the following nomenclature:
First Monday in December = 1stMonDec
Second Friday in April = 2ndFriApr
Third Wednesday in March = 3rdWedMar
This enumeration or encoding was used to determine if a certain week of the month and the day of week held any seasonality tendencies. If you trade index futures you are probably familiar with Triple Witching Days.
Four times a year, contracts for stock options, stock index options, and stock index futures all expire on the same day, resulting in much higher volumes and price volatility. While the stock market may seem foreign and complicated to many people, it is definitely not “witchy”, however, it does have what is known as “triple witching days.”
Triple witching, typically, occurs on the third Friday of the last month in the quarter. That means the third Friday in March, June, September, and December. In 2022, triple witching Friday are March 18, June 17, September 16, and December 16
Other days of certain months also carry significance. Some days, such as the first Friday of every month (employment situation), carry even more significance. In 1996, Bill Brower wrote an excellent article in Technical Analysis of Stocks and Commodities. The title of the article was The S&P 500 Seasonal Day Trade. In this article, Bill devised 8 very simple day trade patterns and then filtered them with the Day of Week in Month. Here are the eight patterns as he laid them out in the article.
Pattern 1: If tomorrow’s open minus 30 points is greater than today’s close, then buy at market.
Pattern 2: If tomorrow’s open plus 30 points is less than today’s close, then buy at market.
Pattern 3: If tomorrow’s open minus 30 points is greater than today’s close, then sell short at market.
Pattern 4: If tomorrow’s open plus 30 points is less than today’s close, then sell short at market.
Pattern 5: If tomorrow’s open plus 10 points is less than today’s low, then buy at market.
Pattern 6: If tomorrow’s open minus 20 points is greater than today’s high, then sell short at today’s close stop.
Pattern 7: If tomorrow’s open minus 40 points is greater than today’s close, then buy at today’s low limit.
Pattern 8: If tomorrow’s open plus 70 points is less than today’s close, then sell short at today’s high limit.
This article was written nearly 27 years ago when 30 points meant something in the S&P futures contract. The S&P was trading around the 600.00 level. Today the e-mini S&P 500 (big S&P replacement) is trading near 4000.00 and has been higher. So 30, 40 or 70 points doesn’t make sense. To bring the patterns up to date, I decided to use a percentage of ATR in place of a single point. If today’s range equals 112.00 handles or in terms of points 11200 and we use 5%, then the basis would equate to 11200 = 560 points or 5.6 handles. In the day of the article the range was around 6 handles or 600 points. So. I think using 1% or 5% of ATR could replace Bill’s point values. Bill’s syntax was a little different than the way I would have described the patterns. I would have used this language to describe Pattern1 – If tomorrow’s open is greater than today’s close plus 30 points, then buy at market – its easy to see we are looking for a gap open greater than 30 points here. Remember there is more than one way to program an idea. Let’s stick with Bills syntax.
10 points = 1 X (Mult) X ATR
20 points = 2 X (Mult) X ATR
30 points = 3 X (Mult) X ATR
40 points = 4 X (Mult) X ATR
50 points = 5 X (Mult) X ATR
70 points =7 X (Mult) X ATR
We can play around with the Mult to see if we can simulate similar levels back in 1996.
// atrMult will be a small percentage like 0.01 or 0.05 atrVal = avgTrueRange(atrLen) * atrMult;
//original patterns //use IFF function to either returne a 1 or a 0 //1 pattern is true or 0 it is false
The Day of Week In A Month is represented by a two digit number. The first digit is the week rank and the second number is day of the week. I thought this to be very clever, so I decided to program it. I approached it from a couple of different angles and I actually coded an encoding method that included the week rank, day of week, and month (1stWedJan) in my Hi-Res Edition. Bill’s version didn’t need to be as sophisticated and since I decided to use TradeStation’s optimization capabilities I didn’t need to create a data structure to store any data. Take a look at the code and see if it makes a little bit of sense.
newMonth = False; newMonth = dayOfMonth(d of tomorrow) < dayOfMonth(d of today); atrVal = avgTrueRange(atrLen) * atrMult; if newMonth then begin startTrading = True; monCnt = 0; tueCnt = 0; wedCnt = 0; thuCnt = 0; friCnt = 0; weekCnt = 1; end;
if not(newMonth) and dayOfWeek(d of tomorrow) < dayOfWeek(d of today) then weekCnt +=1;
dayOfWeekInMonth = weekCnt * 10 + dayOfWeek(d of tomorrow);
Simple formula to week rank and DOW
NewMonth is set to false on every bar. If tomorrow’s day of month is less than today’s day of month, then we know we have a new month and newMonth is set to true. If we have a new month, then several things take place: reinitialize the code that counts the number Mondays, Tuesdays, Wednesdays, Thursdays and Fridays to 0 (not used for this application but can be used later,) and set the week count weekCnt to 1. If its not a new month and the day of week of tomorrow is less than the day of the week today (Monday = 1 and Friday = 5, if tomorrow is less than today (1 < 5)) then we must have a new week on tomorrow’s bar. To encode the day of week in month as a two digit number is quite easy – just multiply the week rank (or count) by 10 and add the day of week (1-Monday, 2-Tuesday,…) So the third Wednesday would be equal to 3X10+3 or 33.
Use Optimization to Step Through 8 Patterns and 25 Day of Week in Month Enumerations
Stepping through the 8 patterns is a no brainer. However, stepping through the 25 possible DowInAMonth codes or enumerations is another story. Many times you can use an equation based on the iterative process of going from 1 to 25. I played around with this using the modulus function, but decided to use the Switch-Case construct instead. This is a perfect example of replacing math with computer code. Check this out.
switch(dowInMonthInc) begin case 1 to 5: value2 = mod(dowInMonthInc,6); value3 = 10; case 6 to 10: value2 = mod(dowInMonthInc-5,6); value3 = 20; case 11 to 15: value2 = mod(dowInMonthInc-10,6); value3 = 30; case 16 to 20: value2 = mod(dowInMonthInc-15,6); value3 = 40; case 21 to 25: value2 = mod(dowInMonthInc-20,6); value3 = 50; end;
Switch-Case to Step across 25 Enumerations
Here we are switching on the input (dowInMonthInc). Remember this value will go from 1 to 25 in increments of 1. What is really neat about EasyLanguage’s implementation of the Switch-Case is that it can handle ranges. If the dowInMonthInc turns out to be 4 it will fall within the first case block (case 1 to 5). Here we know that if this value is less than 6, then we are in the first week so I set the first number in the two digit dayOfWeekInMonth representation to 1. This is accomplished by setting value3 to 10. Now you need to extract the day of the week from the 1 to 25 loop. If the dowInMonthInc is less than 6, then all you need to do is use the modulus function and the value 6.
mod(1,6) = 1
mod(2,6) = 2
mod(3,6) = 3
This works great when the increment value is less than 6. Remember:
1 –> 11 (first Monday)
2 –> 12 (first Tuesday)
3 –> 13 (first Wednesday)
…
…
6 –> 21 (second Monday)
7 –> 22 (second Tuesday).
So, you have to get a little creative with your code. Assume the iterative value is 8. We need to get 8 to equal 23 (second Wednesday). This value falls into the second case, so Value3 = 20 the second week of the month. That is easy enough. Now we need to extract the day of week – remember this is just one solution, I guarantee there are are many.
mod(dowInMonthInc – 5, 6) – does it work?
value2 = mod(8-5,6) = 3 -> value3 = value1 + value2 -> value3 = 23. It worked. Do you see the pattern below.
case 6 to 10 – mod(dowInMonthInc – 5, 6)
case 11 to 15 – mod(dowInMonthInc – 10, 6)
case 16 to 20- mod(dowInMonthInc – 15, 6)
case 21 to25 – mod(dowInMonthInc – 20, 6)
Save Optimization Report as Text and Open with Excel
Here are the settings that I used to create the following report. If you do the math that is a total of 200 iterations.
Seasonal Day Trader Settings
I opened the Optimization Report and saved as text. Excel had no problem opening it.
Optimization results output to Excel and cleaned up.
I created the third column by translating the second column into our week of month and day of week vernacular. These results were applied to 20 years of ES.D (day session data.) The best result was Pattern #3 applied to the third Friday of the month (35.) Remember the 15th DowInMonthInc equals the third (3) Friday (5). The top patterns predominately occurred on a Thursday or Friday.
switch(dowInMonthInc) begin case 1 to 5: value2 = mod(dowInMonthInc,6); value3 = 10; case 6 to 10: value2 = mod(dowInMonthInc-5,6); value3 = 20; case 11 to 15: value2 = mod(dowInMonthInc-10,6); value3 = 30; case 16 to 20: value2 = mod(dowInMonthInc-15,6); value3 = 40; case 21 to 25: value2 = mod(dowInMonthInc-20,6); value3 = 50; end;
if value1 = dayOfWeekInMonth then begin if patternNum = 1 and patt1 = 1 then buy("Patt1") next bar at open; if patternNum = 2 and patt2 = 1 then buy("Patt2") next bar at open; if patternNum = 3 and patt3 = 1 then sellShort("Patt3") next bar at open; if patternNum = 4 and patt4 = 1 then sellShort("Patt4") next bar at open; if patternNum = 5 and patt5 = 1 then buy("Patt5") next bar at low limit; if patternNum = 6 and patt6 = 1 then sellShort("Patt6") next bar at close stop; if patternNum = 7 and patt7 = 1 then buy("Patt7") next bar at low limit; if patternNum = 8 and patt8 = 1 then sellShort("Patt8") next bar at high stop; end;
setExitOnClose;
The Full Monty of the ES-Seasonal-Day Trade
I think this could provide a means to much more in-depth analysis. I think the Patterns could be changed up. I would like to thank William (Bill) Brower for his excellent article, The S&P Seasonal Day Trade in Stocks and Commodities, August 1996 Issue, V.14:7 (333-337). The article is copyright by Technical Analysis Inc. For those not familiar with Stocks and Commodities check them out at https://store.traders.com/
Please email me with any questions or anything I just got plain wrong. George
When I worked at Futures Truth, we tested everything with our Excalibur software. This software used individual contract data and loaded the entire history (well, the part we maintained) of each contract into memory and executed rollovers at a certain time of the month. Excalibur had its limitations as certain futures contracts had very short histories and rollover dates had to be predetermined – in other words, they were undynamic. Over the years, we fixed the short history problem by creating a dynamic continuous contract going back in time for the number of days required for a calculation. We also fixed the database with more appropriate rollover frequency and dates. So in the end, the software simulated what I had expected from trading real futures contracts. This software was originally written in Fortran and for the Macintosh. It also had limitations on portfolio analysis as it worked its way across the portfolio, one complete market at a time. Even with all these limitations, I truly thought that the returns more closely mirrored what a trader might see in real time. Today, there aren’t many, if any, simulation platforms that test on individual contracts. The main reasons for this are the complexity of the software, and the database management. However, if you are willing to do the work, you can get close to testing on individual contract data with EasyLanguage.
Step 1 – Get the rollover dates
This is critical as the dates will be used to roll out of one contract and into another. In this post, I will test a simple strategy on the crude futures. I picked crude because it rolls every month. Some data vendors use a specific date to roll contracts, such as Pinnacle data. In real time trading, I did this as well. We had a calendar for each month, and we would mark the rollover dates for all markets traded at the beginning of each month. Crude was rolled on the 11th or 12th of the prior month to expiration. So, if we were trading the September 2022 contract, we would roll on August 11th. A single order (rollover spread) was placed to sell (if long) the September contract and buy the October contract at the market simultaneously. Sometimes we would leg into the rollover by executing two separate orders – in hopes of getting better execution. I have never been able to find a historic database of when TradeStation performs its rollovers. When you use the default @CL symbol, you allow TradeStation to use a formula to determine the best time to perform a rollover. This was probably based on volume and open interest. TradeStation does allow you to pick several different rollover triggers when using their continuous data.
You can choose type of trigger – (3) Dynamic or (4) Time based.
I am getting ahead of myself, because we can simply use the default @CL data to derive the rollover dates (almost.) Crude oil is one of those weird markets where LTD (last trade days) occurs before FND (first notice day.) Most markets will give you a notice before they back up a huge truck and dump a 1000 barrels of oil at your front door. With crude you have to be Johnny on the spot! Rollover is just a headache when trading futures, but it can be very expensive headache if you don’t get out in time. Some markets are cash settled so rollover isn’t that important, but others result in delivery of the commodity. Most clearing firms will help you unwind an expired contract for a small fee (well relatively small.) In the good old days your full service broker would give you heads up. They would call you and say, “George you have to get out of that Sept. crude pronto!” Some firms would automatically liquidate the offending contract on your behalf – which sounds nice but it could cost you. Over my 30 year history of trading futures I was caught a few times in the delivery process. You can determine these FND and LTD from the CME website. Here is the expiration description for crude futures.
Trading terminates 3 business day before the 25th calendar day of the month prior to the contract month. If the 25th calendar day is not a business day, trading terminates 4 business days before the 25th calendar day of the month prior to the contract month.
You can look this up on your favorite broker’s website or the handy calendars they send out at Christmas. Based on this description, the Sept. 2022 Crude contract would expire on August 20th and here’s why
August 25 is Tuesday
August 24 is Monday- DAY1
August 21 is Friday – DAY2
August 20 is Thursday – DAY3
This is the beauty of a well oiled machine or exchange. The FND will occur exactly as described. All you need to do is get all the calendars for the past ten years and find the 25th of the month and count back three business days. Or if the 25 falls on a weekend count back four business days. Boy that would be chore, would it not? Luckily, we can have the data and an EasyLanguage script do this for us. Take a look at this code and see if it makes any sense to you.
Case "@CL": If dayOfMonth(date) = 25 and firstMonthPrint = false then begin print(date[3]+19000000:8:0); firstMonthPrint = true; end; If(dayOfMonth(date[1]) < 25 and dayOfMonth(date) > 25 ) and firstMonthPrint = false then begin print(date[4]+19000000:8:0); firstMonthPrint = true; end;
Code to printout all the FND of crude oil.
I have created a tool to print out the FND or LTD of any commodity futures by examining the date. In this example, I am using a Switch-Case to determine what logic is applied to the chart symbol. If the chart symbol is @CL, I look to see if the 25th of the month exists and if it does, I print the date 3 days prior out. If today’s day of month is greater than 25 and the prior day’s day of month is less than 25, I know the 25th occurred on a weekend and I must print out the date four bars prior. These dates are FN dates and cannot be used as is to simulate a rollover. You had best be out before the FND to prevent the delivery process. Pinnacle Date rolls the crude on the 11th day of the prior month for its crude continuous contracts. I aimed for this day of the month with my logic. If the FND normally fell on the 22nd of the month, then I should back up either 9 or 10 business days to get near the 11th of the month. Also I wanted to use the output directly in an EasyLanguage strategy so I modified my output to be exact EasyLanguage.
Case "@CL": If dayOfMonth(date) = 25 and firstMonthPrint = false then begin value1 = value1 + 1; print("rollArr[",value1:1:0,"]=",date[9]+19000000:8:0,";"); firstMonthPrint = true; end; If(dayOfMonth(date[1]) < 25 and dayOfMonth(date) > 25 ) and firstMonthPrint = false then begin value1 = value1 + 1; print("rollArr[",value1:1:0,"]=",date[10]+19000000:8:0,";"); // print(date[4]+19000000:8:0); firstMonthPrint = true; end;
Code to print our 9 or 10 bars prior to FND in actual EasyLanguage
Now. that I had the theoretical rollover dates for my analysis I had to make sure the data that I was going to use matched up exactly. As you saw before, you can pick the rollover date for your chart data. And you can also determine the discount to add or subtract to all prior data points based on the difference between the closing prices at the rollover point. I played around with the number of days prior to FND and selected non adjusted for the smoothing of prior data.
Actual data I simulated rollovers with.
How did I determine 8 days Prior to First Notice Date? I plotted different data using a different number of days prior and determined 8 provided a sweet spot between the old and new contract data’s open interest. Can you see the rollover points in the following chart? Ignore the trades – these were a beta test.
The Open Interest Valley is the rollover date.
The dates where the open interest creates a valley aligned very closely with the dates I printed out using my FND date finder function. To be safe, I compared the dates and fixed my array data to match the chart exactly. Here are two rollover trades – now these are correct.
Using an adjusted continuous contract you would not see these trades.
This post turned out to be a little longer than I thought, so I will post the results of using an adjusted continuous contract with no rollovers, and the results using non-adjusted concatenated contracts with rollovers. The strategy will be a simple 40/20 bar Donchian entry/exit. You maybe surprised by the results – stay tuned.
Backtesting with [Trade Station,Python,AmiBroker, Excel]. Intended for informational and educational purposes only!
Get All Four Books in the Easing Into EasyLanguage Series - The Day Trade Edition is now Available!
Announcement – A Day Trade Edition will be added to my Easing Into EasyLanguage Series this year! This edition will be the fourth installment and will utilize concepts discussed in the Hi-Res and Advanced Topics editions. I will show how to develop and program algorithms that will enter after the open of the day and get out before the market closes. Hence, no overnight exposure. Most examples will be carried out on the mini Dow, Nasdaq, S&P500 and Russel. The programming skills that you will learn can be carried to any market that provides enough bang for the buck to day trade. Look for this edition later this year. But get geared up for it by reading the first three editions in the series now. Get your favorite QUANT the books they need!
The Cover of my latest book. The first in the series.
Hello to All! The Easing Into EasyLanguage Series is now complete with the publication of the Advanced Topics Edition. This series includes three educational editions. Start out with the Foundation Edition. It is designed for the new user of EasyLanguage or for those you would like to have a refresher course. There are 13 tutorials ranging from creating Strategies to PaintBars. Learn how to create your own functions or apply stops and profit objectives. Ever wanted to know how to find an inside day that is also a Narrow Range 7 (NR7?) Now you can, and the best part is you get over 4 HOURS OF VIDEO INSTRUCTION – one for each tutorial. All source code is available too, and if you have TradeStation, so are the workspaces. Plus you can always email George for any questions. george.p.pruitt@gmail.com.
Hi-Res Edition Cover
This book is for those that have read the Foundation Edition or have some experience working with EasyLanguage and the various functions that help make a trading decision. This book’s audience will be those programmers that want to take an idea, that requires an observation of intraday market movements to make a trading decision, and program it accurately. If you have programmed daily bar systems, and you want to drill down and add some components that require additional market information (like what came first – the high or the low), then you have come to the right place. If you want to buy and sell short in the same day and use trade management principles such as profit targets and stop losses then The Hi-Res Edition is the book you need. There are two paradigms that EasyLanguage covers: daily and intraday bar programming. It’s the same language, but the move from daily to intraday programming can be quite difficult. Learn all the essentials and shortcuts with this edition. 5 HOURS OF VIDEO INSTRUCTION in this Hi-Res edition – one for each tutorial. All source code is available too, and if you have TradeStation, so are the workspaces. Plus you can always email George for any questions. george.p.pruitt@gmail.com.
Advanced Topics Cover
Advanced Topics (AT) could cover a vast amount of ideas and concepts and be the length of “War and Peace” on steroids. Since this book is part of the series, I wanted to cover a handful of concepts that included the follow programming constructs. Arrays and their manipulation. Buffers (fixed length arrays) and the tools to maintain buffer elements with formulas for extraction and sorting. Finite State Machines using the switch-case construct and range based case values. Using original text graphic objects and retrieving and analyzing their properties to determine X and Y coordinate values of text location. Seasonality: The Ruggiero/Barna Universal Seasonal and the Sheldon Knight Seasonal methods. In AT, you will also find an introduction to EasyLanguage’s Project Concept and the steps to create one by adding/deleting component files. TradeStation now provides access to fundamental data such as Commitment of Traders – learn how to convert the Net Change indicator into a strategy utilizing the FundValue functionality. If you wanted to find out how to merge multiple time frames into a single indicator, you are in luck! Create a MTF indicator for yourself.
Get Day Trading Edition Today!
Day Trading (DT) – This is a surprise installment in my Easing into EasyLanguage Series, as I had only intended on three books. However, I think it will fit well with the other books. Daytrading is a very popular approach as overnight risk is eliminated. Don’t worry there is plenty of risk during the day too! However, it can be very difficult to accurately program a trading idea on higher resolution data such as five- or one-minute bars. Like my other books, there is no “Holy Grail” included. And if you are looking for a book that gets in and out of a trade in a few seconds, this is not the one for you. I discourage trading more than a handful of trades per day – this is best left up to the professionals. But, if you want to learn about volatility-based break outs, pyramiding, scaling out, zone-based trading, accurate trade accounting and having a peek at algorithms that once ruled the systematic daytrading industry, then this is the book for you. A beginner might have a little difficulty in following along with the tutorials. If you have read the first two books (Foundation and Hi-Res) in this series, you are good to go. Or if you have some experience working with EasyLanguage and minute data, you will be OK as well.
Pick up your copies today – e-Book or paperback format – at Amazon.com