Finally after 3 months, today EA Pipscola_Y02 again suffered big losses. And once again this happens when the DST on / off is applied. In accordance with the schedule, began on March 13, 2011 the United States to apply the Daylight Saving Time (DST)and then to be followed by European countries on the last Monday in March.
I do not want to blame the time as the cause of this loss. Because even if I change the trading time 1 hour earlier, it's still could not prevent me from the loss. But at least I already have a conclusion about what to do when there is change in time zones.
Adjust Trading Time with the U.S. Time Zone
When the application of Daylight Saving Time (DST)ends, I have an opinion about what should be done. And I think that right now I have to do the same. We must adjust the time trading with the time zone used on the American market. Actually the logic is very simple, because we transact after closing the American market. It's only natural if we follow the time zone used by the U.S. market.
It is different if you run a trading system that deal only in the European Market. Either simultaneously with the Asian market or not, but before the U.S. markets open. In this case, you do not need to change trading time.
Finally, for the moment I change the trading time of Pipscola EA to 22-0 (GMT +1).
Get Back Your Money
Friday, March 18, 2011
Tuesday, March 8, 2011
Variations of Range Breakout Strategy
N-Bar Range Breakout
Range breakout is a very simple trading system that is very easy to learn for beginner traders. Observe the highest/lowest in 1 or more bars. Put a limit order a few pips away from that highest/lowest. This is the classic N-bar breakout strategy.
Session Range Breakout
This is just a variation of classic N-Bar range breakout strategy. The main difference of this system with N-Bar range breakout is the bar(s) that was observed to be static and associated with the time. Although the session in the forex is often associated with time zones (Asia, Europe and America), but in this case does not always mean that. Because they are static, the rules of this system : "Observe the highest/lowest of the HH:MM to HH:MM".
Opening Range Breakout
I think this system is still a variation of classic N-Bar range breakout. This system is an adaptation of the stock trading system called the Early Morning Breakout System.
This term is actually less suitable in Forex. This is because every Forex broker uses a different time zone.
Bar 0(GMT +1) Range Breakout
Now it's up to you want to call what this breakout kind of. Observations made on H1 timeframe. And probably only suitable for the USDJPY pair. I didn't test it yet on other pair.
For more details, I make an EA to test this system. I arranged to take buy position when price penetrates the high (bar 0 gmt +1) + 10 pips. And vice versa on the sell position. I do not use a limit order and let the EA work in stop and reversal mode. But I do not restrict to just open a trade position. If yesterday's EA took a buy position, and if today prices penetrate the high (bar 0 gmt +1)+10 pips, EA will add a new buy position.
Range breakout is a very simple trading system that is very easy to learn for beginner traders. Observe the highest/lowest in 1 or more bars. Put a limit order a few pips away from that highest/lowest. This is the classic N-bar breakout strategy.
Session Range Breakout
This is just a variation of classic N-Bar range breakout strategy. The main difference of this system with N-Bar range breakout is the bar(s) that was observed to be static and associated with the time. Although the session in the forex is often associated with time zones (Asia, Europe and America), but in this case does not always mean that. Because they are static, the rules of this system : "Observe the highest/lowest of the HH:MM to HH:MM".
Opening Range Breakout
I think this system is still a variation of classic N-Bar range breakout. This system is an adaptation of the stock trading system called the Early Morning Breakout System.
This term is actually less suitable in Forex. This is because every Forex broker uses a different time zone.
Bar 0(GMT +1) Range Breakout
Now it's up to you want to call what this breakout kind of. Observations made on H1 timeframe. And probably only suitable for the USDJPY pair. I didn't test it yet on other pair.
For more details, I make an EA to test this system. I arranged to take buy position when price penetrates the high (bar 0 gmt +1) + 10 pips. And vice versa on the sell position. I do not use a limit order and let the EA work in stop and reversal mode. But I do not restrict to just open a trade position. If yesterday's EA took a buy position, and if today prices penetrate the high (bar 0 gmt +1)+10 pips, EA will add a new buy position.
Tuesday, February 15, 2011
My Backtest is Running Slowly
I think there is something very disturbing in the development of EA Pipscola. EA is running very slow in backtest mode. There are several possibilities that be the cause. I read some suggestions for upgrading a PC that I use. Others suggested to optimize the program code on Indicator and Expert Advisor.
But the problems that I experienced only occur in the period 2009-2010. Weird is not it? For more details, as I do backtest using data from the period 2005-2008, the time needed to complete this process faster than the period 2009-2010. This is very annoying. To be honest until now I can not explain why this could occur. But from this incident I could learn a few things.
Processor
Processor is the core of a computer. The speed and performance of a PC can be seen from the processor it uses. Processor that we usually use today are no longer using a core, but consists of several cores. Okay, it allows us to open several applications at once (multitasking). It's nice when we're optimizing an EA and at the same time still be able to enjoy High Definition video.
Maybe you think if I let Metatrader run alone then the optimization process will run faster. Unfortunately what you think only happens if my computer is still using single core processor. If my computer uses dual-core processor, then there is a unused core practically by Metatrader. EA only runs on a single thread. Even all the EA that are running on a Metatrader share the same thread.
Optimizing the code
In the end I found a way to increase the performance of EA Pipscola during backtest. Apparently myscreen function () which I use to display some information on the chart causes backtest process becomes very slow. So I modify the code to be executed only during live trading. You do this by adding the code
But the problems that I experienced only occur in the period 2009-2010. Weird is not it? For more details, as I do backtest using data from the period 2005-2008, the time needed to complete this process faster than the period 2009-2010. This is very annoying. To be honest until now I can not explain why this could occur. But from this incident I could learn a few things.
Processor
Processor is the core of a computer. The speed and performance of a PC can be seen from the processor it uses. Processor that we usually use today are no longer using a core, but consists of several cores. Okay, it allows us to open several applications at once (multitasking). It's nice when we're optimizing an EA and at the same time still be able to enjoy High Definition video.
Maybe you think if I let Metatrader run alone then the optimization process will run faster. Unfortunately what you think only happens if my computer is still using single core processor. If my computer uses dual-core processor, then there is a unused core practically by Metatrader. EA only runs on a single thread. Even all the EA that are running on a Metatrader share the same thread.
Optimizing the code
In the end I found a way to increase the performance of EA Pipscola during backtest. Apparently myscreen function () which I use to display some information on the chart causes backtest process becomes very slow. So I modify the code to be executed only during live trading. You do this by adding the code
if (! IsTesting ())just before calling
myscreen ();. So the complete code is now
if (! IsTesting ()) myscreen ();this code can be applied to all versions of Pipscola. This optimization can improve the performance of EA Pipscola more than 20% during backtest.
Sunday, January 2, 2011
PipsCola_Y02
Happy New Year 2011. Hopefully this year to be better than last year. I want to tell you that Pipscola Y version will soon be updated. Similarly, the version of D. But time will not be the same, because I still need a little time to complete the version of D. And I'll send this update to your mail box tonight.
Changes to the version of Y only occurs in an exit rule. In version PipsCola_Y01 exit only determined by the value of ATR (Average True Range) and MaxProfit. So in version PipsCola_Y02 just reinforces that EXIT must be done above and below the Moving Average on the time frame H1.
External Variable PipsCola_Y02
MaTP
To adapt the new exit strategy, I added a new External Variable. The default value of this (MaTP) parameter is 5.
If you want EA always close the all position above / below the MA, then you can set the parameters MaxTP in value as much as possible, say 100.
MinLevel
This parameter set a filter that is owned by PipsCola_Y02. In previous versions, this parameter is hidden. Now you can optimize it. The default value is 0.4.
VS_Pipscola_Y02 use charlotte3.dll library version 1.0.0.2. You have to rename the previous library to make it work on the same terminal.
Backtest result (Spread is 2.8, and some missing historical data in October 2010):
Changes to the version of Y only occurs in an exit rule. In version PipsCola_Y01 exit only determined by the value of ATR (Average True Range) and MaxProfit. So in version PipsCola_Y02 just reinforces that EXIT must be done above and below the Moving Average on the time frame H1.
External Variable PipsCola_Y02
MaTP
To adapt the new exit strategy, I added a new External Variable. The default value of this (MaTP) parameter is 5.
If you want EA always close the all position above / below the MA, then you can set the parameters MaxTP in value as much as possible, say 100.
MinLevel
This parameter set a filter that is owned by PipsCola_Y02. In previous versions, this parameter is hidden. Now you can optimize it. The default value is 0.4.
VS_Pipscola_Y02 use charlotte3.dll library version 1.0.0.2. You have to rename the previous library to make it work on the same terminal.
Backtest result (Spread is 2.8, and some missing historical data in October 2010):
MaTP 5, MaxProfit 10, January 2009 - October 2010
MaTP 5, MaxProfit 10, November 2010 - December 2010
MaTP 5, MaxProfit 100, January 2009 - October 2010
MaTP 5, MaxProfit 100, November 2010 - December 2010
MaTP 10, MaxProfit 100, January 2009 - October 2010
MaTP 10, MaxProfit 100, November 2010 - December 2010