My current EA

First words:

I am technically thinking - and because of that, I see the market as a chaotic system. You cannot predict chaotic systems so the only way to go is the way you handle a chaotic system: Use them with a well defined risk strategy.

The same as for Gambling. So, because I am a Software Developer I decided to try an EA which bases on this information. In the past I've developed (for fun) Bot's for Blackjack, Roulette and so on - that's why I started to try my tactics on the market.

Do not try to bend the spoon - that's impossible!

If you look for example on the EURUSD, D1 you will notice, that the price is now nearly the same like back in 1990. So basically you have a 50:50 chance for each order - just like with roulette.

Sadly it is hard and complex to write down what my EA is actually doing - let me try it this way: He goes for a 50:50 chance (using some levels just for not using random), places pending orders with tight S/L, do some stacking and uses a nice system of money managment. That's what it does.
No patterns at all - for the EA the market is just a random bunch of numbers.

So let's see my current backtesting results:

I test the EA with D1 Period, the EA places all pending orders at the start of each new day, with each order expires after 20 hrs. Leverage is 100:1, minimal lot size = 0.01 - so i can start with a 1000 $ deposit

I would love to hear comments, this screens are backtested using MT4 and I put up some screens in the next posts - be patient :-)

Backtesting for the last year, Starting Balance 1000$, fixed spread of 2pips

Profit: 3887$ (nearly 400%?)

Backtesting 2 years, Starting Balance 1000$, fixed spread of 2pips

Profit: $ 22.048,00

Same settings as above for the year of 2010

Profit: $ 17.003,00

So letting it run for the last 8 years (needs some time to finsh :-) ) - same settings as above

Profit: $ 170.067,00

my 3 cts.

1.too much DD
2.backtest in mt4 isuseless, just is useful to test the logic of the EA
3.model quality n/a is too poor, it should be, at least, 99%

Hi joan

I dont tell that this EA will be of any good, I am just testing my thouhts of some market conditions.

I am not a Pro in using MT4 Strategy tester, so i also dont know the exactly meaning of the modelling quality - can you tell me?

Beause of how this EA works internally it needs D1 Timescale and it will take months to test it forward - so I currently playing around with history data.

I will not sell this EA in the future, but maybe I'll describe it's tactic in more detail - the reason for this thread is simply, that I wanna hear some critics, some opinions - just people that kick me in my ass and tell me what the hell I missed or what I am doing wrong because this EA cannot work.

Please keep in mind, that my EA has far more lost trades than won trades, but still gains profit because of the internal managment

Here is USDJPY, 2010-01-01 to 2012-01-01

Profit: $ 5.340,08

Hi mr2...interesting EA of placing pending orders

i was looking for an EA to place pending order (based either on specific price or in terms of number pips) based on last week's closing price.

The pending order has both Take profit and Stop loss function.

Does your EA has such function? Mind sharing? I can't find an EA with such functionalities as i mentioned above. Thanks.

hi hewtwok

What exactly do you mean? My EA places several pending orders at each beginning of a day, all with predefined S/L and T/P

You can get the close prise of the last week, with "iClose" function with PERIOD_WEEK and shift 1 .. if it's that what you mean - look it up in mql documentation, it's the first place to go

Below backtested on USDJPY 1984-01-01 to 2012-01-01


Profit: $ 631.076,20

Hello Mr2, this looks good but I would be interested in a forward test in DEMO to begin with. If you like I could help with that - send me the details.

AndyG

hi mr2

thanks for reply.

however i am not a programmer

but thanks anyway.

My EA uses a capped progressive investment scheme - for savety reasons it is capped and the propability of a full loss goes nearly zero after some months (except end of world conditions with losses over a full year, propability nearly zero).

If I remove this safety cap the profit on USDEUR for the last year drops to just 1.535 (instead of 3.887) because of bad losses in the second half of the year.

Edit: SORRY, MISTAKE, the following values comes from a 1 YEAR period, it's the year 2004, jan-dec)

EURUSD (without cap) would be 236.004,00, starting with 1000 USD at 1.jan 2004, ending 31.dec 2004

I'll attach the Balance/Equity Graph for the 2004-year period without cap (with cap the balance raised almost linearly)

Hello Mr2.
I agree with you in the "concept" that the market is just a random bunch of numbers.

Any "patterns" that we see in past data is irrelevant to the future, in the sense that the chance of that pattern to replicate in the exact manner is always 50/50 same as any other indicator.


Your Description how EA works, I would venture to say that a realistic results should be around Break-even less spreads.


The use of tight S/L, could be the reason your BT come as shown.


I agree with joancb in point 1. Too much “Max” dd. If that gets you when you first start, your account is gone.


Back Test in mt4 is not useless, you just need to import tick data to it and do your BT s.
That will show the real deal when using tight S/L, and producing 99% quality test.

.

Thanks for your post - if someone can provide me with apropriate tick data I would love to try my EA on it because the history which i got currently is not the best. I am also open for step by step instructions how to test it properly, just because I want to know, if my EA works or not or I am doing anything not right.

The only essential things are: a fixed Spread of 2pip, 4 digits and stopmargin of 3-5

But that's not the full point - because the EA mostly works with any random number feed. It would work in a casino too, but then I would get kicked out because they say "I steal their money".

In Forex I don't steal money from my broker, so noone should kick me out when using such strategies - hopefully.

the EA just invests a fraction of available capital - so it could last a very long time of losses. but it also needs some time to "produce" a good amount of money.

In the (capless) 2004-year example the EA handles 90 lots per trade in dec/2004 and it's still just a fraction of the balance so it could still handle a long run of losses.
Such losses happened in late 2011 (for some months) - the EA can hold such losses for nearly 1 year (no matter how much balance) before going dead or starting at initial deposit

Okay, because of your intents about the big DD i tried something out.

And yes, there is an additional risk of when to start the ea, because it can drop the balance down very hard at the beginning.

Tested this with a start in august.2010 the EA drops down the balance over some months and starts rising in dec. again.

But that's how it works - if it wont rises again, after a year of fails the balance is gone (1000 $).

I'll see that problem, but after further investigation the setups for such fails are very rare .. after 1 year of normal profit you will be able to lose 3 times of that .. it's still a risk :-)

Check this out, it seems complicated at the begining but when you are familiarized with it it is very easy...

http://eareview.net/tick-data

I would love to hear how your EA works,

Cheers!

I'll try that out, because I am much interested in the result with high quality tick data (even, when MT4 dont simulate it right).

Thx for the link

Ok, just fooling around and trying my EA on an H4 Basis .. it's not made for it, but i tried it anyway.

The risk of the mentioned bad luck when starting the EA (main reason is the minimum lot size of 0.01) is a bit greater, but that's what happened with my balance on H4 in the first week of this year:

Profit: 3.394,00 - out of a initial deposit of 1000 USD

I've tried the same for the period of 2011-01-01 until now and ended up with a profit of 135.000.

The Graph was quiet interesting, because until april the balance was droping constantly down to 25 $. This Drawn was followed by a huge up till the end of the period. Seems to work in bad cases also .. Lot size during the down was the minimum of 0.01 lot with 1:100 leverage could have been the end, but wasnt.
Cap was activated, so you see a linearly going line upside.

Tick history data is just the begining and you can get that from Dukascopy. After that, to make History data for mt4, made from Tick data is quite complicated.

It takes time to make. Once made, files are quite big in the vecinity of 6g per pair + around 2g per each time frame you want to BT. So..it Is not "just" a matter of someone providing it ready for mt4.

If you google birt 99% quality test, you will find all the instructions there and all the tech details and scripts for to make it yourself. If you are serious about testing your codes, you must take the time to master it. You will never regret the time spent on it.

.

I will try that out surely.. but maybe I just raise a demo account and let it run and see what happens :-)

Is it right, that MT4 does not use the real tick-data? So it is possible that some peaks are hidden during testing? I have no further information about this issue.

But I feel confident, that it's not really important because of the way my EA works .. it is not really dependent on perfect tick data .. there are only pending orders involved.

Sure there can be many things happen and there is still a hand of luck involved..

As it looks, my EA works on H4 also .. so i will made it ready for use in a demo account .. this way I got the best information if it works. I figured out, that my EA will need 2 weeks on H4 to make reasonable first profit .. let's try that :-) wish me luck ;-p

Best of luck mr2!

my fourth cent: forward testing is the best way to know if a system really works, and if it is in a Real account much better