In search of the sacred 'grail'... - page 2

New comment
 
Hoper23 >> :

Article about auto-optimization, which looms as a bright spot on the site ... not that it does not work ... there are only 4 parameters that can be put on the optimum, and as it all works laxly.In general, maybe someone will prompt....

Try to look here (I haven't got round to it myself yet): - Autooptimization



 
That's what I "like" about such forums ... is that you have to register to download something ... It would be OK, but on a Bilain connection of 1-2kb / sec, it looks grim prospect ... but never mind.
 
I see. It's something like an expert. In principle, it works fast enough (only by opening prices, i.e. by bars). But the problem is that we have to add about 200 strings describing in details what and how the optimizer should optimize. The version with the library optimizer was easier... but the quality wasn't very good either. Ok, thanks to RID for the link, I'll be torturing the keyboard... For now there's only one indicator left...
 
The terver comes to mind... if one shooter hits the target with 40% probability, the second with 50% probability, the third with 25% probability... what is the probability that all shooters will hit the target at the same time...
 
Hoper23 писал(а) >>
I started with Stochastic - it's the fastest, then I diluted it with OsM filter...and then it was on to HighLow, Aligator, Mashki, CCI.
"Grale" doesn't smell here, it smells of ridiculousness and tinkering to "work" on a story. The approach of bumping indicators like this will achieve nothing. The profit of the system should be in the ideas, it should be reasonable at the design stage, and at the stage of construction - adjustment and tuning. It is possible to express everything or almost everything with a multi-period set of indicators, it's just a question of how to interpret them. In general, I think that 90% of indicators, and certainly almost all of the classical indicators, are a relic of manual trading, for automatic trading we need other tools. Well, this is your business, I will not interfere with my own ideas.) If you do not know what to do and what to expect, you cannot do it at all.)
 
Hoper23 >> :

Afternoon...or night, everyone. I'm working on a maniacal EA here. I'm not going to post it yet, it's just not finished yet, but I'll post it for sure...for FREE. I just need help. Question namba van - how to make auto-optimization ... say 8-16 napametrov? Article about auto-optimization, which looms large on the site ... not that does not work ... there to 4 parameters can be put on the optimum, and how dodgy everything works. In general, maybe someone will prompt Chago of their own developments? MY DEAR KIM, IT WOULD BE WORTH REFERRING TO YOU IN PARTICULAR. If I've got a good idea, I may try to draw some of them on the market but I don't have any good ones. Sometimes it overdraws, sometimes slows down, sometimes shows nothing worthwhile. I am not saying "give me a subomega indicator and I will change the world". The essence of my strategy in the Expert Advisor is the percentage ratio of a group signal of 8 indicators and a little filtered one. So I have indicators of standard collection and I want to experiment with individual indicators. In general, maybe the collective mind will win.

To help you understand, linear optimization of 8-16 parameters is 8-16 nested for loops, in which each

where each variable changes from minimum Min_n to maximum Max_n with the corresponding step Step_n.

Now imagine how many calls of the function optimized by such a linear optimizer should be performed:

(Max_1-Min_1)/Step_1 * (Max_2-Min_2)/Step_2 * * * (Max_16-Min_16)/Step_16.

A year ago I wrote such an optimizer for 10 parameters on mq4.

The optimization runs for about 10-20 minutes depending on Step_n.

The way out of this situation, as it seems to me, is as follows:

- switch to genetic, or other algorithm for fast optimization (difficult);

- search for xeon's posts, he wrote an optimization pack using MetaTrader built-in optimizer (strange);

-write optimizer in C or other fast programming language(complicated).

 
Hoper23 >> :

Afternoon...or night, everyone. I'm working on a maniacal EA here. I'm not going to post it yet, it's just not finished yet, but I'll post it for sure...for FREE. I just need help. Question nemba van - how to make auto-optimization ... say 8-16 napametrov? Article about auto-optimization, which looms large on the site ... not that does not work ... there to 4 parameters can be put on the optimum, and how dodgy everything works. In general, maybe someone will prompt Chago of their own developments? MY DEAR KIM, IT WOULD BE WORTH REFERRING TO YOU IN PARTICULAR. If I've got a good idea, I may try to draw some of them on the market but I don't have any good ones. Sometimes it overdraws, sometimes slows down, sometimes shows nothing worthwhile. I am not saying "give me a subomega indicator and I will change the world". The essence of my strategy in the Expert Advisor is the percentage ratio of a group signal of 8 indicators and a little filtered one. So I have indicators of standard collection and I want to experiment with individual indicators. In general, maybe the collective mind will win.

Here is an example.

//Оптимизация
//Оптимизация
//Оптимизация
      if( Optimise!=0)
      {
         j=0;
         win_summ=0;
         L1= L1Start;
         while( L1<= L1End)
         {
            L2= L2Start;
            while( L2<= L2End)
            {
               L3= L3Start;
               while( L3<= L3End)
               {
                  L4= L4Start;
                  while( L4<= L4End)
                  {
                     P1=0;
                     while( P1<=1)
                     {
                        P2=0;
                        while( P2<=1)
                        {
                           P3=0;
                           while( P3<=1)
                           {
                              P4=0;
                              while( P4<=1)
                              {
                                 TradeBUYSELL= SELL;
                                 while( TradeBUYSELL<= BUY)
                                 {
                                    if( TradeBARStart>0 && TradeBARStop>0)
                                    {
                                       TradeBAR= TradeBARStart;
                                       while( TradeBAR<= TradeBARStop)
                                       {
//Оптимизатор
                                          Optimisator();
//Оптимизатор  
                                          j= j+10;
                                          TradeBAR++;
                                       }
                                    }else
                                    for( d=0; d< rd; d++)
                                    {
                                       TradeBAR= TradeBARSArray[ d];
//Оптимизатор
                                       Optimisator();
//Оптимизатор     
                                       j= j+10;
                                    }
                                    TradeBUYSELL++;
                                 }
                                 P4++;
                              }
                              P3++;
                           }
                           P2++;
                        }
                        P1++;
                     }
                     L4= L4+ L4Step;
                  }
                  L3= L3+ L3Step;
               }
               L2= L2+ L2Step;
            }
            L1= L1+ L1Step;
         }
         Print("_______Оптимизация закончена!");
      }
//Оптимизация
//Оптимизация
//Оптимизация
 

Figar0 писал(а) >>
... Вообще, мне кажется что 90% индикаторов... это пережиток ручной торговли, для автоматической нужны другие инструменты.

OK, I agree. Everyone knows that truth is born in an argument. In fact, there is truth in it. But then how do we give signals to the indicator, because nobody cancelled the If() system. Propose your concept of automatic systems.

 
thecore >> :

To help you understand, a linear optimization of 8-16 parameters is 8-16 nested for loops in which each

variable changes from minimum Min_n to maximum Max_n with the corresponding step Step_n.

Now imagine how many calls of the function optimized by such a linear optimizer should be performed:

(Max_1-Min_1)/Step_1 * (Max_2-Min_2)/Step_2 * * * (Max_16-Min_16)/Step_16.

A year ago I wrote such an optimizer for 10 parameters on mq4.

The optimization runs for about 10-20 minutes depending on Step_n.

The way out of this situation, as it seems to me, is as follows:

- switch to genetic, or other algorithm for fast optimization (difficult);

- search for xeon's posts, he has written the optimization package using the call of MetaTrader's built-in optimizer (strange);

-I want to write optimizer in C or any other programming language (difficult).

I understand it very well. This is exactly what is posted in Autooptimizer of foreign forum. The essence of optimisation is to follow the market. Optimisation is to do auto-optimisation every day.

Yes, and another thing, Figar0, comrade, and the system even in a fecal form works)))) Of course not as planned and I'm not shouting that you're a fool and I'm smart, I just want to say that the strategy is working, but it's only a preliminary test on the demo for today. I started with 50$, lot 0.01 at 60% and 0.02 at 75% on 5 pairs

 
Hoper23 писал(а) >>

"omitting" indicators, Figar0 offers you, Hoper23, to work directly with bars, i.e. directly with price information, without intermediaries

123456789...11
New comment