In search of the sacred 'grail'... - page 6

 

http://forex.sunstation.com/pics/sm_1.gif

And I don't get it... where's the vibration? ))) It's not interesting. There's a guy who wrapped a roll so that you can't tell without a litre.)))

 

By the way, I'm still running my system in the demo. No optimiser yet, but still. My last deal was (now 16-55) at 04-15, and now I opened it again, i.e. I was out for 12 hours and went again. And all because the market was so weird. Run on EURUSD, GBPUSD, USDJPG, USD\CHF, AUD\USD, USD\CAD. Here's the updated chart. 8.61% drawdown.


 
thecore писал(а) >>

Do you think a 10-year forward test period is long enough?

Too long. I think a year is enough, and that's too long. The market is constantly changing and so are the signals.

thecore wrote >>

Before optimization, I set the drawdown level acceptable for me, for example 40%, and optimize it.

Optimization period is 5-10 years.

Then I run it outside the optimization zone and if the drawdown has not changed by more than 1/2, i.e. it is no more than 60%,

which is also acceptable for me, while maintaining a sufficient level of profit, e.g. 1/2 of the profit within the optimization zone

then I consider the Expert Advisor successful.

Nah, 40% is too much for optimisation. I would take about 15% maximum of $10,000 at 1.0 lot. I would take 2007-2008. And I would take 2008-2009 as forward (although it is a year too long, half a year), the drawdown should not be less than 20%. That is the result I would support with a sufficient amount of trades (depends on the tested period, well, for example 15-20 on М1 per month will suit me at a decent profit). But I think the main test of operability is forward pass on Blezneca Pair. It would be really cool.
 
Hoper23 >> :

http://forex.sunstation.com/pics/sm_1.gif

I don't get it... where are the vibrations? ))) It's not interesting. The dude wrapped a roll so you can't tell without a litre.))

Vibrations can be better viewed on

http://forex.sunstation.com/pics/spectrum_magickum.gif

 

Dear passengers, can anyone still tell me the actual code of the auto-optimiser? Because I'm already confused.

extern double Skillstart=1;//Процент шанса от 1% до 100% (оптимально 44%)
extern double Skillstep=1;//Процент шанса от 1% до 100% (оптимально 44%)
extern double Skillend=50;//Процент шанса от 1% до 100% (оптимально 44%)

extern double SkillMAXstart=51;//Процент ХОРОШЕГО шанса (оптимально 70%)
extern double SkillMAXstep=1;//Процент ХОРОШЕГО шанса (оптимально 70%)
extern double SkillMAXend=99;//Процент ХОРОШЕГО шанса (оптимально 70%)

extern double stKstart=5;//К период Стохастика
extern double stKstep=5;//К период Стохастика
extern double stKend=5;//К период Стохастика

extern double stPstart=3;//П период Стохастика
extern double stPstep=3;//П период Стохастика
extern double stPend=3;//П период Стохастика

extern double stDstart=3;//Д период Стохастика
extern double stDstep=3;//Д период Стохастика
extern double stDend=3;//Д период Стохастика

extern double Wstart=12;//ОсМа настройка
extern double Wstep=12;//ОсМа настройка
extern double Wend=12;//ОсМа настройка

extern double Hstart=26;//ОсМа настройка
extern double Hstep=26;//ОсМа настройка
extern double Hend=26;//ОсМа настройка

extern double Cstart=9;//ОсМа настройка
extern double Cstep=9;//ОсМа настройка
extern double Cend=9;//ОсМа настройка


extern double CCIstart=1;//НАстройка CCI
extern double CCIstep=1;//НАстройка CCI
extern double CCIend=50;//НАстройка CCI

extern double F_EMAstart=1;//Настройки МАСиДи
extern double F_EMAstep=1;//Настройки МАСиДи
extern double F_EMAend=50;//Настройки МАСиДи

extern double S_EMAstart=1;//Настройка МАСиДи
extern double S_EMAstep=1;//Настройка МАСиДи
extern double S_EMAend=50;//Настройка МАСиДи

extern double SMAstart=1;//Настройка МАСиДи
extern double SMAstep=1;//Настройка МАСиДи
extern double SMAend=50;//Настройка МАСиДи


extern double Sstart=0.1;
extern double Sstep=0.1;
extern double Send=1.5;

extern double Ostart=0.1;
extern double Ostep=0.1;
extern double Oend=1.5;

extern double Istart=0.1;
extern double Istep=0.1;
extern double Iend=1.5;

extern double Gstart=0.1;
extern double Gstep=0.1;
extern double Gend=1.5;

extern double Mstart=0.1;
extern double Mstep=0.1;
extern double Mend=1.5;

extern double CCstart=0.1;
extern double CCstep=0.1;
extern double CCend=1.5;
And it all needs to be optimised... Well, let's strain the grey matter in the skull...
 
Dear passengers, can anyone still tell me the actual code of the auto-optimiser? Because I'm already confused.

I've written two dozen variables and I'm already confused.

In the optimizer code for 10 pages.

Why would you need it if you're already confused.

I already gave you an example.

on the second page.

 
infinum13 >> :

It's too long. I think a year is enough, and that's a bit long. The market is changing all the time, so are the signals, by the way.

And where have you seen a downtrend during 2007-2008?

How can your strategy be tested on the downtrend, if it has only seen the uptrend.

Naturally, it will lose on the period of 2008-2009.

Another thing if you suggested the period 2005-2007, oh, I wouldn't argue.

No, 40% is too much for optimization.

I wrote EXAMPLE.

It's for you it's too much, because for you 10.000$ is, probably, very big money (sorry, I'm not personally acquainted, don't be offended if I

exaggerated the wording a bit).

And for me it's two to three months earnings. And I can easily sacrifice and even 50% if the return is at 80-100% p.a.

I would take about 15% maximum of $10,000 at 1.0 lot. I would take 2007-2008. And 2008-2009 would be forward (although the year is too much, half a year), while the drawdown should not be less than 20%. That's this result I would believe, with a sufficient amount of trades (depends on tested period, well for example 15-20 on М1 per month will suit me with a decent profit).

The 15% drawdown out of 40% is not a problem. Increase the deposit by 2.67 times.

And keep the same level of lots.

And besides, it is very hard to make $1,000,000 with 15% drawdown for a period of 10 years.

It only works out to $100,000. And I don't need $100,000. I need $1,000,000.

But I think that the main test is the forward pass on the Pair of Blesnets. That would be really cool.

There's no such thing as twin pairs. All pairs have different volatility, different behaviour, because they reflect different markets.

But by adjusting a little the strategy parameters, we can achieve its usability on other pairs,

For example, EURUSD is well transferable to EURJPY or AUDUSD.

But it makes no sense to transfer it to EURAUD. It is completely different.

 
thecore >> :

You've got a lot of nerve.

I already gave it to you.

on the second page.

Well, I'm telling you, dear ZeCor, I looked it up, I was so amazed that I left for a day in the code world and somehow came back with a defeat. What has been posted there is a ready-made Expert Advisor, it's a horrible mess. I ripped out the optimization block and rewrote my code. In theory it should work - but it won't. I keep getting an error when calling the optimization block. The thing is that optimization is imbedded into all source code and locked to TP and SL. I tried to re-do it, but there are a lot of custom functions I'm like a pig in oranges with. I'm not a downer. I'm just confused in this code, I have too many variables to optimize (19), and there's a formula like

Combination = MathFloor((TPEnd-TPStart)/TPStep)*MathFloor((SLEnd-SLStart)/SLStep);
I can't figure out how to represent it with 19 variables. In short, I'm stumped with this code. I didn't say it doesn't work - it works fine in the original. But how do I apply it to myself?
 
Hoper23 >> :

Well, I tell you, dear ZeCor, I looked it up, I was so amazed that I went into the world of code for a day and somehow came back with a defeat. What has been posted there is a ready-made Expert Advisor, it's a horrible mess. I ripped out the optimization block and rewrote my code. In theory it should work - but it won't. I keep getting an error when calling the optimization block. The thing is that optimization is imbedded into all source code and locked to TP and SL. I tried to re-do it, but there are a lot of custom functions I'm like a pig in oranges with. I'm not a downer. I'm just confused in this code, I have too many variables that need to be optimized (19), and there's a formula like

I can't figure out how to represent it with 19 variables. Anyway, I'm stumped with this code. I didn't say it doesn't work - it works fine in the original code. But how do I apply it to myself?


In short, the programmer got tired and went for a beer.

Not to brag, but I just finished a project with over 200,000 lines of assembly language code

for the microcontroller, 10,000 lines of VHDL for the programmable logic matrix and 10,000 lines of Windows software

that the microcontroller communicates with and a couple of thousand lines of Windows driver code.

Plus the design and manufacture of two circuit boards.

I have been developing this project for the last year and a half.

So I'd better keep modestly silent about the fatigue.

 

But seriously, you have everything. You just need to sit tight for another week.

No one will do the job for you for free.

And for a fee, very few will.