In search of the sacred 'grail'... - page 5

New comment
 

Форекс это не подбрасывание монетки, поэтому "фундаментальные правила теории игр" здесь не работают,

And in general the notion of fundamental rules has little applicability to complex systems,

Well, yes, trading on the Forge is very often not a coin flip, but a sandwich.

About game theory, that's probably best left to Neutronych. I'm not very good at it. Somewhere I have heard that there is a minimax theorem (or principle). Well, if it is the rule thecore is talking about, then probably it is, as Foreh does not look like an antagonistic game. And even if the trader does have an antagonist, his intentions are not at all clear.

One last thing: somewhere I also saw out of the corner of my eye in the comments to another article by Golubovsky, that the system, as it becomes more complex, tends to be stochastic in the sense of control. So the fundamental deterministic rules may be disappearing, gradually changing into stochastic ones.

 
Mathemat >> :

Well, yes, trading on the Forge is very often not a coin flip, but a sandwich.

About game theory, that's probably best left to Neutronych. I'm not very good at it. Somewhere I have heard that there is a minimax theorem (or principle). Well, if it is the rule thecore is talking about, then probably it is, as Foreh does not look like an antagonistic game. And even if the trader does have an antagonist, his intentions are not at all clear.

One last thing: somewhere I also saw out of the corner of my eye in the comments to another article by Golubovsky, that the system, as it becomes more complex, tends to be stochastic in the sense of control. So the fundamental deterministic rules may be disappearing, gradually becoming stochastic.

Forex is like the weather. It is impossible to predict where the wind will blow (especially in windless weather - flat) (not hard, exactly impossible).

But there are seasonal fluctuations, for example buying silver in India in September-October due to national holidays.

 
thecore писал(а) >>

Forex is like the weather. It is impossible to predict where the wind will blow (especially when the weather is flat) (not hard, but impossible).

But there are seasonal fluctuations, such as buying silver in India in September-October due to national holidays.

I agree with the seasonal fluctuations. After all, there is always spring after winter, but it also snows in May :))

 
Figar0 писал(а) >>

Just in case you missed it, 'Who wants a strategy? Lots and free of charge)', there are already two solutions for this purpose...

Seen it. Didn't like it. And in general, I've grown cold to the standard indulgences.

 
thecore писал(а) >>

You didn't get it right, I got it right. Why shout about it? Especially a negative experience.

Maybe you misunderstood me. Everything worked for me, and forwards sometimes gave a positive result in the next 2-3 months, but in the period longer strategy did not give a positive result. That is why you are trying to optimise. And how do you select the best combinations of indicators out of the many combinations? By the balance after the test? By the drawdown? By expected payoff?

 

You know what I think the problem with finding an effective system is? That everyone looks too deeply into theory. They remember all the holy scientists and try to attribute to Forex all sorts of theories from other operas and so on. Perhaps, it makes sense, I cannot deny it, but I cannot listen to such criticism calmly either. How can I argue that "such" (I'm not specifically talking about mine, several systems have been mentioned here) system cannot work, and if it does, it is purely based on fitting history. I'll grant you that not everyone who asserted this had personally tried this system. Let's reason. As far as Forex market cannot be predicted by itself, it means that we cannot build any theories concerning it. Yes, there are moments when we can say with certainty - the trend will change...but it's seasonal, as we've been told. It is therefore impossible to come up with a strategy with 100% results, as well as claim that any other strategy proposed by the idea man will not work. I hope you get my point? I'm an adequate person and I cannot rely on the indicator that my system has given me for the day, willing as I am. Yes plus, yes fun. But not the system... and the entries were awkward on the pivots...

There was a suggestion to refuse indicators and work with bars. And I do.

 double sigyH1=iLowest (NULL, TF1,MODE_CLOSE,3,0);
   double sigyH4=iLowest (NULL, TF2,MODE_CLOSE,3,0);
   double sigyD1=iLowest (NULL, TF2,MODE_CLOSE,3,0);
   double sigxH1=iHighest(NULL, TF1,MODE_CLOSE,3,0);
   double sigxH4=iHighest(NULL, TF2,MODE_CLOSE,3,0);
   double sigxD1=iHighest(NULL, TF3,MODE_CLOSE,3,0);

     if ( sigyH1==1){ SigYTF1= I; SigXTF1=0;}
     if ( sigyH4==1){ SigYTF2= I; SigXTF2=0;}
     if ( sigyD1==1){ SigYTF3= I; SigXTF3=0;}
     if ( sigxH1==1){ SigYTF1=0; SigXTF1= I;}
     if ( sigxH4==1){ SigYTF2=0; SigXTF2= I;}
     if ( sigxD1==1){ SigYTF3=0; SigXTF3= I;}

But one cannot gather much information with only bars. This topic has discussed a lot. Forgive me for the comparison, but the second act of the quantum trading system about the third eye and vibration of space is about to begin.

And anyway, how do you like the idea of creating an indicator based on a neural network? Uh-huh. It's supposed to be something interesting.

 
infinum13 >> :

Maybe you misunderstood me. Everything worked for me, and the forwards sometimes gave positive results for the next 2-3 months, but at a longer period the strategy did not give a positive result anymore.

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

That's why you are trying to auto-optimize.

I am not the author of this thread. I temporarily gave up auto-optimization.

And how would you select the best combinations of indicators from the variety of combinations? By the balance after the test? By the drawdown? By expected payoff?

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

The 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.

Sometimes, if the symbol's history is not too long, I use this variant of optimization and backtest: 

int start()
{
   if(IsTesting()==true)
   {
      if(IsOptimization()==true && DayOfWeek()&0xE==DayOfWeek())return;
      if(IsOptimization()==false && DayOfWeek()&0xE!=DayOfWeek())return;
   }
//код советника
//код советника
}

Instead of 

DayOfWeek()
it is possible to put another function, e.g,


Month()



or vice versa - a smaller one 

 Hour()



 
Hoper23 >> :

You know what I think the problem with finding an effective system is? That everyone looks too deeply into theory. They remember all the holy scientists and try to attribute to Forex all sorts of theories from other operas and so on. Perhaps, it makes sense, I cannot deny it, but I cannot listen to such criticism calmly either. How can I argue that "such" (I'm not specifically talking about mine, several systems have been mentioned here) system cannot work, and if it does, it is purely based on fitting history. I'll grant you that not everyone who asserted this had personally tried this system. Let's reason. As far as Forex market cannot be predicted by itself, it means that we cannot build any theories concerning it. Yes, there are moments when we can say with certainty - the trend will change...but it's seasonal, as we've been told. It is therefore impossible to come up with a strategy with 100% results, as well as claim that any other strategy proposed by the idea man will not work. I hope you get my point? I'm an adequate person and I cannot rely on the indicator that my system has given me for the day, willing as I am. Yes plus, yes fun. But not the system... and the entries were awkward on the pivots...

There was a suggestion to refuse indicators and work with bars. And I do.

But one cannot gather much information with only bars. This topic has discussed a lot. Forgive the comparison, but the second act of the quantum trading system about the third eye and vibrations of space begins.

And in general, how do you like the idea of creating an indicator based on a neural network? Uh-huh. I think it should be interesting.

Vibrations of space work very well by the way.


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


SZ


At the expense of bars you cannot gather information... What do the turkeys gather information on? What are they, the third eye?


In the end I got it from the indices too, but after analyzing everything very carefully, I realized that it was just a very good smooth chart.

 

I don't argue... It's smooth... well filtered... But you know, it's like making soup, tasting it, and something's missing... And you can't figure out what it is. And let's pour, pour, throw into this soup in the hope that some of it is OK.

Optimization, in my opinion, does not mean adjustment for the story or even worse - predicting the future. I believe that optimization (automatic) merely tells you about the state of the market N time ago as a result you can plan further market action. Suppose the market was stable and the probability percent has decreased, which results in more deals because this MEGA hits the right place even at 10%. But here the market has changed and became aggressive, we harden the conditions by percentage-filtering and here we have a MEGA at 60%, very carefully opens few deals, but with high quality. For me it is very logical. We analyse the market with a delay, so we don't really know what's going on with the market but we do it with history. In such moments of market transition from amorphous to aggressive state we suffer losses... small but losses... In practice, we only lose the drawdown, not the profit. And vice versa - we do not gain profit. But then we react to changes and a new tough policy of percents of probability of MEGA and everything is OK. We do not want to be greedy, as long as we do not take profit. As soon as the profit went uphill - all covered MEGA Greed and strangles ... strangles bitch so ....How it all ends, I think, do not need to explain to anyone.

 

"You can't gather information on a bar tab... but what do turkeys gather information on? What are they, a third eye?"

No, not a third eye. Obviously, turkeys feed on bars. Well, what else are they supposed to eat? You've got a turkey sitting there sharpening bar after bar. I have simply expressed my idea of an automated system. I am not refuting a new one, I am for a dialogue. But other than bars to feed the turkey. Not a turkey? Then who and what to feed? (for the smartest ones, a tick is a component of a bar) There is also an Expert Advisor that works on the news. But I have a hard time imagining such an automatic process. I tried it with my own EA - it had a lot of complicated stuff... I think it worked. But it did not show me anything smart on profit. The news - yes, it affects the movement. But not always where you program it to go. And looking at the feed is too slow, and before you know what's what, the trend is gone... So if anyone has an idea how to avoid bars - I am listening attentively.

1234567891011
New comment