Channel Trading, Scalping With Keltner Channel

One of the most complicated thing that I feel when making Volatily Scalp system is anticipating a drastic decline of Average True Range (ATR) in record time. As a scalping system based on the trading channel, determining the length of the period Keltner Channel is very important. This is because the channel width of the Keltner Channel indicator can change over time depending on market volatility.
In the Channel trading, generally upper channel and lower channel act as overbought and oversold levels. Buy position is taken when the price hits or below the lower channel and sell position taken when price hit or above the upper channel. No matter with indicators that used to create the channel. As for an exit strategy there are many options but generally is the opposite of the entry signal itself.
Now let us look at the chart below. On this chart I put a Keltner Channel indicator (20.2). Red vertical line to mark the beginning of the transaction (at 21 GMT 1). I also placed an indicator CK_Speed (20.50) and ATR. (20).
It should VS_Eurocross can get sell signal as I describe in the chart above. Although the position of an exit may be slightly different. But what I want to show is how the system with the Keltner channel (20.2) miss a couple of profitable trades. Therefore, I draw 2 horizontal lines that act as a channel to clarify what I mean.
I've been plagued with this problem. And it seems impossible to know what is the optimal length of Keltner Channel Period at any time. But when I'm working on EA's Charlotte, I always strive to capture this pattern. I use the High / Low of the last 2 bars when the entry signal occurs as a place of exit. The results with the same chart as above, charlote EA is more aggressive.

Again a Commercial EA Sell Volatily Scalp Expert Advisor

A few days ago I received an email containing an attachment of a commercial expert advisor. Ea was named CaspianEA CHF. The sender suggests that the EA is very similar to VS_EuroCross. After watching the entire of Caspian code, then obviously it is a copy of VS_EuroCross.
I must admit it seller is smart to hide his cunning. Unlike the Pip Turbo EA, CaspianEA do some things to hide that EA is just an imitation.
Changing DLL File Name
The first thing done by CaspianEA to deceive buyers is to replace Volatily Dll file be CaspianEA.CHF name. This is not a difficult thing to change the name because I also recommend if you want to run vs_eurocross on several pairs in a terminal. But look at the function used, it is clear that the version used is version before d05. But fortunately they can not change the name of the function names are used. So they actually just change the file name. And so far, although I've sold some vs_eurocross source, but for the DLL source code itself, I never give it to someone.
Changing Volatily Period Variable with Caspian Factor
This is just a trick. In principle, this variable is the period of the Moving Average, Standard Deviation and Average True Range Indicator. CaspianEA using a smaller value such as 3. However, in the code, this value will be multiplied by 10. So to use volatily period = 20, then the value Caspian Factor must use 2.
Putting indicators that are not used in any trading decisions
This is exactly the trick that fooled many traders, especially for those who do not understand the MQL language. EA sellers will say they add some indicators. But in fact, their EA does not use these indicators in any trading decisions.

Variations of the Period, Entry and Exit

So far, all Volatily Scalp EA that I created is using the Moving Average as a target for the exit. In vs_eurchf and vs_eurocross I use 20 Period Moving Average while vs_eurgbp I use 50 Period Moving Average. This system will provide maximum results if the price moves only among the Upper / Lower channel and Moving Average ..
Actually, with 20 Period Moving average, we could impose the use of Upper / Lower channel as the exit. But unfortunately, it is very often if we wait for prices to penetrate Upper / lower channel, instead of profit that would come less. So when doing testing, I found an exit on the moving average is more profitable than the exit on the Upper / Lower of Keltner Channel.
10 Period Moving Average
This is slightly different if we use a shorter period, in this case 10 Period Moving Average. With a shorter period, this will increase the sensitivity of the moving average. This allows us to no longer put the Moving Average as a place of exit. However, the place for entry can be used as the exit for the opposite position.
I found the use of a shorter period would be very beneficial if used in the major pair. This is what I try to apply on USDCAD pair. Especially since I only use the shorter trading time (22-23) GMT +2, so I think it is wise to maximize the profit from each trade.
Combinations 10 and 20 Period Moving Average
Actually I was a bit hesitant to use this combination. For the first time I tried to use 2 Keltner Channels Indicator simultaneously. This provides 2 variations to place the entry.
The first variation, entry can be done if the price penetrates both Upper / Lower channel. To exit, we have 3 places, 10 Period Moving Average, Upper / Lower of Keltner Channel (10.1) and 20 period Moving Average.
The second variation, entry can be done if the price through the channel up / down Keltner Channel (10.1) while the distance between price and 20 Period Moving Averages is greater than 10 Period ATR. For the exit we can use as the first variation.
This method I use to pair USDCHF and USDJPY and You can see the back test results on my previous post.

Trading results, is there a dependency present?

This week begins with the loss derived from Eurgpb pair. But in the end these losses can be covered and still make a profit, although not as the previous week. The question whether any good results will be followed by the opposite result?
Apart from that, EA's performance was good overall, including new EA that run on USDCAD this week.
Determining Lot size
Along with the increasing number of pairs that use, the determination of risk for each pair and the EA becomes a problem. For the moment I tend to use fixed lot for EA that run on the major pairs. As for cross pair I use a fixed ratio. But from my observation, it is rarely a pair makes big profits for 2 successive days. Exceptions occur in 2 weeks ago in which a constant EURGBP pair produce a better profit than the other pair. But generally, every day the pair will produce more profit will be different.
Change lot size every day based on the performance of previous day might be an interesting idea. Since I tried VS_EurChf these characteristics have become my thought.
At the end, maybe we should learn about the z score, a statistical method to determine if there are dependencies or correlation among all our trades.

6% Profit During a Week

A week has passed and all my EA produces 6% profit from the Initial Balance. Of the total profit is 1 / 3 resulting from EURGBP pair. And pull it constant obtained every day for a week. Maybe if you run Scalping EA on several pairs at once, usually you will find that every day pair producing more profit will be different every day. And I think it was a good thing, because then diversification is working properly.
What about the EURCHF pair? I still think you can rely. It's just that maybe next week it my open time will shift from 19 to 20 GMT+2. This is based on the fact that often at the end of the UK session was already very low volatility and tend to increase until the closing of the American market. In the last week recorded frequently executed trading at 19 GMT +2, it gives a negative result. But I have not thought of to stop it because after all the great profit potential is still there at this hour.
For USDCHF and USDJPY pair I thought I would still use the trading time 22-23 GMT +2. Although the latest optimization shows that both pair can be traded (scalp) until 4 GMT+2. But I think in the medium term will be very volatile. But instead I will add a new pair of USDCAD. Recently I tried to add the VS_EUROCROSS Fractal indicator, and the results are good for USDCAD pair. Thus I will add this pair in my account next week.

A good performance in 3 days

A good start is in early September. Apart from the losses at the end of last month, eventually most of the loss can be offset by profits earned in the last 3 days. This can not be separated from the good performance of the 2 pair I have just used, the USDCHF and USDJPY.
3 days detailed statement
Also impressive performance was obtained from the EURGBP pair. So that it can be said that the worst results obtained from the EURCHF pair, although still positive in the end.
I give a special note to the EURGBP pair. This pair can be said to not give a meaningful contribution in the last month. But when I noticed lately, EURGBP pair is slightly more volatile than the EURCHF pair. This can be noticed before the closing session of the UK. Even the early Asian session that EURGBP volatility is still possible to get a few pips profit.
This will become more evident if we compare the daily chart for the both pair. For the EURGBP, its ATR (5) is above level 60, while EURCHF is still under level 60.
But maybe that interest is to compare the condition of the daily trend in the both pair. A question whether this is the difference? rd3etifapw
EurGbp Daily Chart
EurChf Daily Chart

Scalping And Volatility

I think August is the month that EA is not profitable for me. Right in the last week in August was EA I experienced considerable Drawdown. As a result the loss of 75% of the profits that have been obtained from the beginning of the month.
Perhaps this is the accumulation of the low volatility of current. When the market was in a low volatility, the possibility for a change (increasing) to the greater volatility. Actually I've noticed this situation since the beginning. As the UK towards the end of the session, the market does not show impulsive movement. Finally, Impulsive movements result in the end it started the American session. Fortunately the price corrected more than half, so that greater losses can be avoided.
Shock Trade
I personally hope this is a sign will be a change in Day-to-day volatility of the next. Although not in the short term, at least those changes may be seen in this month. Ideally for scalping system in the Asian session, at least ATR (20) are not under 6. And to achieve that level condition, at least daily range should be above 75 pips.

New Expert Advisor

I've made an EA. Still incomplete, but at least my backtest results do quite encouraging, at least for the last 1 year. EA is working on the USDCHF pair and just do trading at 22-23 hours GMT +2. Nevertheless my backtest results on USDJPY pair also gives good results although there is a greater drawdown.
New EA
And it turned out with little change in my vs_eurocross can also get a satisfactory result on the both pairs. I change volatility period to 10, and set a maximum trade per position to be 2 and maximum trades per bar to 1. As for the use of indicators ck_speed speed is limited to 1. (Yellow).
VS_eurocross backtest on usdchf
VS_EuroCross on USDJPY
Especially for USDJPY pair, my backtest results using vs_eurchf_d01 or pipturbo, by limiting the trading hours of time at 22-23 GMT +2 also gives good results.
Thus, at this point I added 4 EA and into my account using 3 difference EA.

Finding Best Time to Scalp Major Pair

I start my job again making trading system for major pair. This system is still working on a particular clock hour. For that I tried to dig up from some other popular EA, see the tendency at any hour they work.
I find generally brief EA is working on a range of time shorter than the EA is working on Cross Pair. This makes the number of trades executed may be very little. This coupled with the fact that they generally only work 4 days a week. The effect recovery factor will be a long time, when the EA suffering draw down.
Interestingly most of the EA is able to work in more than one pair at a time, either by using the same variable or different. Unfortunately one of the fundamental weaknesses that are owned by the Meta Trader is that we can not do backtest on more than one pair. So we can not determine whether it is more profitable or not. Diversification effects may not work properly as expected.
As a preliminary conclusion, I see that most of us will get a good backtest results if we restrict the EA to work on the 21-23 or 20-23 hours GMT +2. This is not foreign to me, because when I make VS_EurCHF, for the pair EURUSD, the profit factor increases if the trading time is limited to 20-1 GMT +2.
I tend to wish to make my EA is able to work as well as with the Cross pair. However, until now, I must admit, that forced to work like that cause suffering Account large draw down. Until now I was thinking to try some exit strategy. With a small amount of trade, should be made to maximize any existing trade instead of just scalp it.

Expert Advisor Performance Hourly

I think we should be grateful to Mt4stats which has developed a variety of features it has, so we get a better picture of how the performance of an expert advisor. The entire feature is currently not possible we will get in the Meta Trader Detailed Statement. The most concrete example is we can analyze the performance of expert advisor hourly. This is very useful for us to run the expert advisor who works only on at certain hours. Here is current performance on Alpari Russia demo server.

Pips vs Hours

If we consider the graph above, then we will have a picture of what hours on the expert advisors that produce more profit than the other hours. For plasticity at this time that most of the profit obtained on days 18 and 22 (19 and 23 GMT +2). Based on this information, we can see that the open hours for the best at this time are 19 and 23 or 19 and 22(GMT +2). But back at any time that this condition could be changed. However, this is the most close to the current market condition.

Keltner Channel and Bollinger Band

When I learn about Average True Range and Standard Deviation indicator, I find it interesting when put Keltner Channel and Bollinger Band together on a chart. As we know, that the Keltner Channel is made from a combination of Moving Average and Average True Range. And Bollinger Band is combination of Moving Average and Standard Deviation.
What's interesting my attention when Bollinger Band (20.2) is located inside the Keltner Channel. (20.1). This is a technical condition which is very appropriate to describe the market conditions are in the sideway. Moving Average is located right in the middle chart. Support and resistance are on the right Bollinger Band. Unfortunately, often in a situation like this, the price Range doesn’t allow us to make profitable trades. Especially with the wide spread condition plus the possibility of getting slippage when taking and liquidating a position.
However, the longer market in the situation likes this, and then will be going to the big new trend. Possibility of profit that we get, can be lost quickly when the breakout occurred.

Exploiting Low Volatility

At the decent time most of trades executed during 19-23 GMT+2 only. After that time often just one trade can be executed. With the possibility to set the minimum range we can exploit this low volatility. I change my set for 22-05 trading time. For the last 2 days, I don't use spread filter during this time. I set my min range to 1 and let my min profit to 2.
VS_EurChf Optimization
Actually optimizing VS_EurChf is more simple then vs_eurocross. Mostly you just need to optimize the trading time and stoploss. To get higher profit ratio then reduce maxtrade value. If you are optimizing on Mayor pair then you should always set TradeLowRangeDay to False as well tradeoutsideband. On the other hand, you set both variables to True on cross pair. rd3etifapw
Keep the robustness of your optimization. Every time you get satisfaction result on mayor pair, then test it on other mayor pair and it should be profitable. The result should be also profitable on cross pair.

How to Choose a Broker

I run my demo again today rd3etifapw. I decide to use NordMarkets to run VS_EuroCross (FixedTP). I use the same setting with my previous forward test on Tadawul. There is a significant difference in EurGbp Spread at NordMarkets. It offer 2 fixed pip spread.
It doesn't mean i give a recomendation to use this broker to run my EA. Its only cause I see that seem it price movement close to real feed.
However if you are looking for a good broker to run my EA, then just look their liquidity in the Asian Session. Compare several broker real feed, your first attention to find the highest true range (20 Period ATR). Mostly you will find the broker with 5 digits has higher range then 4 digit broker. Next look their spread. Its good if you can find a fixed spread broker that their spread is less then 5.
ECN Broker
It seem a hard one to find low fixed spread broker and looking ECN broker is the best one. ECN offer floating spread (low) according to market liquidity and mostly give another transaction cost. Thus we pay for commission and spread. If you decide to use ECN broker then make sure their commission is less then 1 pip value per trade size. This is important because my EA calculates the profit in pip. I ever find a broker with really good feed and execution. It has higher liquidity then others broker during trading time. But I forget to calculate their stupid commission. Its equal to 5 pip! Sick. My EA can make 7 positive transactions, but all become negative, because it just 4 pips profit per trade.
Requote
Requote is another aspect that you must think before using a broker. Unfortunately we must trade in real to know it. There is no other way to know how good their execution except to test it directly.

I Stop My Test

I already stop my forward test on TadawulFX. The data feed on EurGbp is very difference with actual price and unacceptable. However it show us the useful of showing our target. But if you want to compare between your demo on other broker with Tadawul, then here the login 999333047 and password bxr3kqr of demo account.
I run 3 vs_eurocross on this demo account, 2 on EURCHF pair and the rest on EURGBP. Since last week I use the FixedTP version.
Set 1:EURCHF
OpenHour=19;
CloseHour=23;
Risk=25;
UseAtrMod=true;
MyOwnTarget=5;
VisibleTP=True;

Set 2:EURCHF
OpenHour=22;
CloseHour=05;
Risk=25;
UseAtrMod=true;
MyOwnTarget=5;
VisibleTP=True;

set 3: EURGBP
OpenHour=19;
CloseHour=23;
Risk=25;
UseAtrMod=true;
MyOwnTarget=5;
VisibleTP=True;

Others setting is default for all sets. Trading hour is GMT+2.

A week before updating

As I have stated before that the next version (D05) will be unlimited one, and may be in the short period there is no more any change after that. The version will be release as usual on 27 July or before 1s August 2009.
The FixedTP version is the base of D05 version. Also as I stated before about possibility to use new Indicator, but seem i will not use it the next version. Right now i just think to add news filter because this task is always appear every month.
About Selling My Source Code
I think there is a misunderstanding regarding this issue. It seem i will force you to buy (donate) to get the next version. No, absolutely its a wrong perception. In fact I will release D05 without time expiration as usual in previous version.
What I sell is my VS_EUROCROSS without using activex file. I give you a choice to use unexpired VS_EuroCross with activex version, or to buy the source vs_eurocross without using dll. GPL license will guarantee you can modify my source and resell it, but you have also give your source to your buyer. Term Free is the freedom to get and modify the source code. Other wise you can modify and never sharing your modification if you don't want to share your source.
Anyway if you have another option then you can also contact me to see if we can make agreement.

Trade in the right time and the right size

Since d04 I make forward test with 2 instances of VS_Eurocross on EUR/CHF pair with difference time trading set. It is just another way to optimize my system instead of changing any possible variable. I still try to build a version that can run with several time set at once with difference risk, Money Management and maximum trade. What the lack of Meta Trader platform that we can perform a back test with more then 1 system in single account. So we must combine all system in a single EA to perform such test.
Perhaps by doing this test, it can help us to decide when to put our risk at maximum value, and when we must trade with low risk. Putting risk on the right size is not easy task. Although there are some money management strategy but not every system can match with those formulas.
I post the trades of the trading from last night and seem if trading time 19-05 is worse in this case, it hold trade longer and get less pip.
Open Hour 19
Open Hour 22
On both picture you see also why I am not recomending to run more then 1 expert at a terminal. You can see the spike, once expert can get 2 trades while the other just a single trade. No matter how good your PC and the realibility of your internet connectivity, all expert advisor that run in a terminal, is executed in a single Thread. You can search in google what the term of Thread meaning in programming. But what I know that if an EA is executing its order, then the other's expert on the same terminal must wait to execute their order.

2 Hours Delay on Weekly Opening

I still don’t know how to react about difference opening time for each broker. Since my work using Alpari historical and I never test it with other Broker’s data. So far on forward test in Tadawul give good result if the expert trade in the first hour, but I don’t know how the performance in the past. There is no special exit strategy for any trade during this time. It’s fine if our trade become profitable during this hour too. How if it doesn’t make profit? This is important because the EA will enter to non trading time, and the EA never find this case in the back test.
Managing Your Risk
Although it is not covered in my EA, but truly there is a difference in potential gain in the first trade. Normally after trending market in American season the first trade tend to be above 2nd deviation Keltner Channel or even more. It is not always happened at 20 (GMT), it may be later. But the point is our potential profit will be higher. You can make a back test, and you will find that the higher profit almost in the first trade.
And if we look to back test result, during 20-22 (GMT) trading time is the most profitable area. This is not meaning that we never getting loss during this time, but this is because our profit during this time is bigger then after 22 GMT. It is because the ATR value is still high during this time.
What an idea that may we apply in this situation is optimizing our Money Management. We can set our MM to give more risk during bigger profit time or trade, and start reducing our risk after that.

Back test Problem

So far the common problem that rises when doing back test is digit difference between brokers. But the fact that I never find that problem when running in mode even real account. However I realize that the closing function in the expert is not normalized as it must. Therefore I make an update of MQL to solve this problem. But again if you make backtesting then historical data should come from the related broker or the backtesting become meaningless when using different broker with different data.
Although I give a code “4digits” in the file’s name, but it does not mean it can’t work with 5 digit broker. The difference between previous update, that this update normalize all closing function with broker’s digit. Therefore I am still keeping the previous update, if current update can fix digit problem then I will delete the previous one.

Correlation Indicator

Another popular method to scalp cross pair is using correlation technique. I get some suggestion to use correlation indicator to improve my EA performance. But first I have to make correlation indicator that match with my system.
And I get idea to make correlation indicator based on market volatility. The main principle of this indicator to compare the mayor pairs volatility. And therefore the formula that I use is so simple. Take the difference between current price and Moving Average of Median Price and then divide by current ATR.
The Indicator has 2 input variables for Mayor Pair that will be measured. By default it will measure the EURUSD and USDCHF pair. These pair should form a negative correlation that if EURUSD showing a positive number then USDCHF should show negative number. This is will be formed by two lines that make us easy seeing the correlation.
The histogram is the difference (absolute value) between both lines. I have tested to change CK_Speed indicator in VS_EURCHF, and showing better result, however the result on VS_EUROCROSS is worst then using CK_Speed indicator.
VSCorrelation

Create Your Own Exit Strategy

As any other non stop reversal systems, creating exit strategies become a hardest part in developing trading system. Since I include the MQL4 source code in VS_Eurocross_d02 version, then you can analyze its exit logic. And it is possible to change all exit logic with your own. If you are doing back test; take a look to your journal at the end of back test. You will see a message like "case1:10","case2:10". Each represents the logic code and the total of that exit logic executed during back test period. And you can optimize it by trying to remove it one by one to see the different.
You can have different strategies, but here what I did:
I separate my exit strategies by its working time.
Its mean that I have exit logic that work during trading time, after trading time and work all around the time. If you have exit logic, try to put them on the correct working time.
Create exit strategies related to Keltner channel position
Think if price across the moving average, can it touch next channel? It’s no problem if we make profit at this point, but how if we are holding a loosing trade?
For me, it’s all a guessing game, but using one ore more indicators we can make accurate decision.
Using Profit Accumulation
You will find that sometimes I let each open position to get its target and sometimes I will accumulate the profit from all open position.
Bar counting method
I have assuming that if several bar has been passed and the open trade does not hit it’s maximum target, then try to reduce the target related how much bar it has been passed.