Conquering Cyborg ~ Finally
Filed Under (EA System, Journal) by Casey on 13-10-2009
Tags: Backtest, Cyborg EA
It has been slow and difficult. Not difficult in the sense that backtesting was hard work. Difficult in the sense that I was really dragging my feet. Getting back to the momentum after about a month of non activeness is really difficult.
The law of momentum really applies here…
I have been dragging my feet last week, even with a renewed enthusiasm, I still could not get myself to really focus on my conquest on Cyborg.
The mind was strong but the body was weak… sighh… Nevertheless, I hope with these new found results. I might be able to meet my objectives.
I have over the past been backtesting Cyborg with a new idea. It may not be new to most of you but it was new to me. I previously did not reveal this idea because I wanted to ‘blackmail’ the cyborg developer… 
Why? because with such an interesting EA, he had never advice or even backtesting EURCHF and provide for his clients… Then again… I am not his client… but I have referred to him a few clients so I am not sure if his private forum ever mention this…
Anyhuuuu… I think it’s time I let you all know what I was doing with Cyborg. Oh.. for your information, I have been backtesting these on the very first version of Cyborg.
The idea I had used was the time factor, which I had mention before when I finally gave up on Best Forex Sniper. Cyborg has the parameter or algorithm to control the time to trade. With the experiences I have with Best Forex Sniper, I found out that if the trade were taken at a particular time period/frame/length… This strategy could prove successful.
So with that in mind, I set out to see what is the optimized time.
I have also in fact stated before, that volatility was a killer for this strategy, and usually the volatile period most of the time were during the currency trading hours and news hour.
So like scalping, the best time window to activate this EA was outside the country/continent trading hours of a particular currency pair… EURCHF would of course trade during asian hours and vice versa…
On top of that, I narrowed the window further by looking at the critical news period of both these currency pair. Forex Factory has one help of a calender site to do this… http://www.forexfactory.com/calendar.php
One flaw in this idea that I have recently discovered is that… most of the time, the trade does not close within the time window. So, if the trade prolongs outside this trading window, the it will be in the mercy of volatility. So I guess not all currency pair would be the best used of this strategy.
Below are two backtesting results on EURCHF. One was tested within the trading window while the other was open meaning trading window period was turned off. (Click on graph to view detail statement)
—————————————-
EURCHF(1999-2009)~Time period 1600 – 0500 hrs 
—————————————-
EURCHF(1999-2009)~Time period off
—————————————-
It is interesting to note, with that with out the time control, EURCHF has one instances where it was stopped out and but when I set the time window period off… that particular trade did not take place. I took a step further to identify what this happens… and no answers were found. So I took it as it was a bad trade… no 100% correct system.
Still… this was still itching in my mind… this test was from 1999 till 2009 and this drawdown took place in August 2008.
I understand that this currency pair has a very wide trading time window because EUR and CHF is practically housed under the same time zone. Other than that… I am ignorant about it’s characteristic…
Now here is the kicker… One of the many reason, I have been very quiet these past two weeks was also I was backtesting and re backtesting and testing again a few other currency pair which caught my attention…
I found two currency pair which does not seem to be affected by the time period window… and the result was a straight arrow to the sky of the graph…
Hmmm… shall I be a bastard again and hide my parameters? hehe… ok… just kidding.
Below are the results of two currency pair which I found with good results… with the time window turned off…
——————————–
EURGBP(1999-2009)~Time period off
——————————–
AUDUSD(1999-2009)~Time period off
——————————–
Here are the parameter which I had used for EURGBP and AUDUSD. Again… this is the first version of Cyborg that I was using. Not too sure if the new version would yield these sort of results
Here are the only parameter changed. All other were maintained to their default
mm=—Money Management—
UseMoneyManagement=0
mls=–Manual Lot Size If MM = false–
Lots=0.01000000
BrokerIsIBFX=0
BrokerPermitsFractionalLots=0
pom=–Primary Order Management–
TakeProfit=30
Slippage=5
TimeGrid=30
Grid=30
TP=30
MaxLevel=8
StopLoss=330
sm0=–Trading Hours – Brokers GMT Server Time–
sm2=UseTradingHours – 0 = false, 1 = true
UseTradingHours=0
As always… the next step is to go live with this…
great work on back testing
I had a play with your settings and managed to reproduce them
looking more closely it seems you achieved the outcome by using a wide stop (330 pips instead of the 90 pip default). That combined with a 6 level averaging down setting and a low number of pips per trade gave you the near straight line equity curve
I had a play with the settings which you may want to try, which is to limit cyborg to only 2 levels on averaging down, but to use a 4% based money management strategy. The curve is more bumpy, but it gets to a higher net equity by the end.
In terms of ideas, I wonder if there are a good set of settings for cyborg which dont have a wide stop and a large number of averaging down settings?
btw. I doubt that there is an answer to the above as I suspect the problem is not in the exit logic and the trade filters (e.g. time of day) which the EA gives you control over. I think there is something wrong with the entry logic as the EA sometimes tries to short its way into a price action that is clearly trending.
cyborg uses an stone age approach for the martingale, thats the reason why it fails on the long run.
only real adaptive martingale will make the race;-)
http://pcmax8multi.mt4live.com/
@Caprica, that is my dilemma… i was thinking about building on the success of my backtesting… optimize it to have higher yield and lower drawdown.
One thing good about cyborg is that, I can set the grid level. What is the grid level that you use on your setting?
The idea that I used on Cyborg was base on my experiences with BestForexSniper… using 30pip grid right up to the 8th level…
I do not mind testing out your setting as a basis and hopefully work out a better setting with improved yield and hopefully lower drawdown.
I do have two more currency with the same curve but they had very deep drawdown 48% and 63% respectively… due to the nature of the grid that I used was the reason for that… maybe with your settings I might improve on that…
@Onur: haha… i know u have the ‘best’ martingale EA… Again… care to share it with me…
i’ll be ur best friend… hehe
the static grid(pip step) is the problem of all martys. Find a way to adapt this grid distances to the momentum of the price movement and u will get ur “best” marty
BTW the live stat shows our 1st generation of multi pair EA. The equity curve of the second generation is a lot smoother yet. And the third we are working on, beats the s$$t out of the market… hehehe
short update: switched the public multi pair acc. for better analysis purpose to https://www.myfxbook.com/portfolio/pcmax8multi/3400
2. generation of multi pair EA:
http://www.myfxbook.com/members/pc8multifx/trend2multiok/3311
@Onur… good idea… any chance of sharing the indicators you used and the calculations you use to determine the grid distance?
Say… i got an idea for you… set up a manage account site… i can be your agent and get u customers to trade on a manage account basis… and together ve can rule ze world… muahahaha
Yeah, the idea rocks!
After taking a look at your live accounts i invite you to be my client… muahahaha