I programmed an EA called which measures the speed of change in pips and trades accordingly. my backtest-results are very promising and I’m testing it live right now. I believe the live test will be identical because of the way the EA is programmed: it only checks and adjusts every new 1H candle. A fixed stop is used of 40 pips. Profits are let grown to maximal potential. I attached the backtesting results. I used lotsize of 0.5, because the testingcapital is fixed 10000. As you can see the maximum total drawdown is 10% and the biggest risk/trade is 2%. The average risk per trade is 1.13%. In the testresults, NO compounding was used. (The compounding results are through the roof ). There is NO martiginale system used. MAX 1 buy or sell open at once.
I’m very intrigued, but would it be possible for us to test the EA ourselves? That way we can control data quality, length of the test, brokers used, etc.
Looks good. I see that you tested over a period of a year. That is more than I always do, but I tested mostly on M1, so sounds good to me. Edges may have a limited lifetime so you should not throw away a EA when the results are less profitable over a longer period of time, as long as the last period works…
Is the strategy fully automated? That means that you have a good working script cycle, nice. That can take some work.
Do you time measure pips by measuring seconds between pips, or by measuring pips over a bar in a lower timeframe? For example, count volume on M1 to see if it is speeding up. Would like to know your approach. I do the last mentioned one.
How do you determine direction of the trade? Is that trend based or based on the pips direction within the bar?
Nice work! Would like to see your code, but I can understand that you like to keep it to yourself…
Well, I’ve been busy programming and testing for the last days, but here’s a little feedback:
@Toekan: thx for the positive feedback. I know what you mean when you say: “…as long as the last period works” I don’t think any EA can work forever with the same parameters… you have to optimize the EA from time to time so that it works best for the last weeks/months.
This is indeed a fully automated EA and I am testing it live along with 7 other EA’s I programmed, all with different strategies, just to spread the risk. I have backtested the EA’s more in detail, using 1min history from 2007-2012 and the result are very comparable (with 90% data quality). Besides, if your EA only uses candle-open/close/high/low data to enter and exit trades, it makes no difference in using the 1m history data… I think ATC SHOP does not know this yet…
Now I’m testing it live on a demo-account, just to see if the EA acts the same as in the backtests. With only 2 weeks of live testing, it’s to early to judge the profits, but it seems that there is no difference between the live and the backtest on how the EA enters and exits trades… which is ofcourse good news
About this EA: in short what is does is measure the size/hour of the last X candles and when there’s a sudden increase in size/hour, it jumps in in that direction, setting it’s stop and limit relative to the average sizes of the candles. Like jumping on a train that is speeding up and getting off as it begins to slow down A breakout-strategy following price action… In theory a very simple strategy, but I don’t believe in very complicated EA’s that use hundreds of indicators. I think the most important part of the EA is the exit-strategy… not how you enter trades. I’m going to try programming an EA that enters trades randomly, and then handle the trades with a certain exit-strategy that still manages to be profitable in the long run. That would be the ultimate strategy I think
You are right about not posting my code I put alot of hours/weeks in the programming of my EA’s and testing them. If they should work, I’m thinking of selling them… and ofcourse using them myself. I hope you understand. Especialy since some other members, like ATC SHOP, rather make jokes about somebody’s good intentions and hard work, instead of seeing the bigger picture of why we all check this forum…
I agree. On the H1 it is also good to follow upcoming trades, so manual backup is possible. You just want to be sure that the signals are picked up and that it doesn’t enter a trade open/close loop… But got my safeguards now, you experience that once and make sure it wil happen never again.
Hmmm, I am also running 7 other EA’s and some in the sandbox. When you are on the run, the ideas keep popping up in your head, and you are not distracted by screen time while trading, so you can prog and prog… It should safe me time, but I maybe spending it on playing with the scripts and development…
My Cashcow script is also a simple one. I have complex scripts, but I am just very found of the simple one… it is consistent in returns and robust as can be…
Right you are, I am not posting mine either… But I think you know I truely asked it from a script technical point of view, but no worries. I have enough running at the moment, but it is always good to hone your prog skills by looking at other codes.
Ignore the nay-sayers… Most often they haven’t used scripts or have been beaten up by third party ones…
Well, if you like to chat a bout scripting… You know where to find me…
Hi, the initial investment has no influence on the EA’s performance. I’m testing it now live for 2 months and it does exactly the same as in the backtesting.
The live trades still match the backtest-trades.
With the basic, very cautious settings: Max drawdown = 4%, profit AVG is approximately 10%/year. I’ve also programmed an autolotsize function which enable to set your risk/trade. With this enabled and risk set to 2%: Max drawdown= 16%, yearly profit exceeds 70%.
I’m thinking of commercializing this EA, so I wonder if anyone knows how I can protect my EA? Or set it up so I can make in run on demo-accounts only? (as in “test for free before you buy”)
Its almost impossible to protect distributed code. Even if you move critical logic into a DLL using another language, it can be cracked by someone with the right tools. The best protection is to offer it as a web service, but that has some challenges as well.
As far as restricting it to demo, that is easy. Just use the IsDemo() MQL4 function. But again someone determined can beat that.
Very impressive results, can you give more details about the way it works? When I’ve tried implementing something along these lines myself it’s been less than profitable.
Look into mql lock for a way to protect your EA - or better yet, you could port it to MQL5 and then sell it on the official mql5.com market which will handle protecting the EA and locking it to one user’s account.
). There is NO martiginale system used. MAX 1 buy or sell open at once.
