Diablo - page 22

New comment
 

When observing Diablo, or when trading with an EA, after each direct order has closed in profit with all inverted orders of the same (relative to the initial price) direction, the Take Profit can be reduced by three corridor widths.

 
JonKatana:

Theoretically, there is one single price movement trajectory that can bring the balance to zero if the wrong lot is selected. It is a strictly one-way movement along the "dragon" trajectory, i.e. a pass to the second level, then a pullback to the first level, then a pass to the third level, then a pullback to the second level, etc. If the price follows strictly this trajectory, never deviating from it, sooner or later the deposit will be nullified. As soon as the price reverses or at least a couple of reverse level reversals (in one direction) occurs, the profit starts to increase inexorably.

Diablo itself is still profitable - it will take your deposit out in the plus at any protracted 'dragon', it will just take longer. But the wrong calculation of the lot may lead to the fact that the orders, to which the price has reached, there will be simply nothing to open. So we should not be greedy. This advice would also work for choosing the step between the orders.

Afternoon.

I've suddenly decided to do some basic theorizing and it's very important for understanding the work of grippers in particular.

I have modeled the price trajectories in Excel with the PCF generator: in fact I am modeling the system where the same channel width is always set for pending orders (for example, 120 points), if the price reaches the upper limit, mark it as "1", if it goes to the lower limit, mark it as "-1". Then we plot a random walk trajectory.

Example, some possible patterns:

At the first step, we have 0 - this is the point of reference. Then, at each step, the trajectory either goes up by one or down by one. So, we either catch the upper limit first, or the lower one. And so on.

I have limited my calculations to five steps.

2^5 = 32 - that is how many trajectory variants we have in total. Let's generate 320,000 trajectory instances (so there will be 10,000 instances for each unique trajectory) - this number is enough for statistical validity. And see if there is any variation in the probability of specific trajectories occurring.

In the figure, the pattern is a formalisation of the trajectory. For example, at the very bottom, the price goes down 5 times in a row; a little higher, the price goes up 5 times in a row.

And we see the uniform distribution. (The irregularities are caused by imperfections in the Excel PRNG.) It may seem strange to some, but this is the basic level of the theoretician. A series of coin tosses returns an equal probability result for any outcome in the series.

I decided to check it on quotes, made a simple Expert Advisor that breaks either a take or a stop. Runs only in long or short. The percentage of profitable trades is about 50% in any case (with a take equal to a stop, of course).

extern int TP=200;
extern int SL=200;

//+------------------------------------------------------------------+
//| expert start function                                            |
//+------------------------------------------------------------------+
int start()
  {
//----
    if (OrdersTotal()==0)
      
      {
        OrderSend(Symbol(),OP_SELL,0.1,Bid,3,Ask+SL*Point,Ask-TP*Point,"",777,0,Red);
        OrderSend(Symbol(),OP_BUY,0.1,Ask,3,Bid-SL*Point,Bid+TP*Point,"",777,0,Blue);
      }
    
    else
      
      {
   return(0);
      }
  }

On the minutiae history by opening price for 3 years, the result is similar to the model from excel:

Also a uniform distribution over 36,000 trajectory realisations.

The trajectory can be padded to 6, 7, ... steps. The result will be a uniform probability distribution.

Accordingly, when creating a gridiron, you should immediately understand that any trajectory, including the one that will knock out your deposit - with a large number of trials will occur as often as others.

 

And I would also add that if it is known, for example, that the end of the deposit will come after the 8th step of any one unique trajectory (for example, when price moves 8 times in one direction), then we can calculate the probability of this event and then estimate the frequency of occurrence. Having 8 steps, we obtain 2^8 = 256 unique trajectories. Since all trajectories are equally likely to happen, the probability of lethal = 1/256. Not too little, considering that there may be several thousands of trades in a year. And if the depo is drained after 6 steps, the probability of trajectory will be 1 / (2^6) = 1/64. In general, it is very risky.

In general, the topic is interesting, but experienced people have already written that any grider is a loser. This is probably very close to the truth.
 
alexeymosc:

And I would also add that if it is known, for example, that the end of the deposit will come after the 8th step of any one unique trajectory (for example, when the price moves 8 times in one direction), then we can calculate the probability of this event and then estimate the frequency of occurrence. Having 8 steps, we obtain 2^8 = 256 unique trajectories. Since all trajectories are equally likely to happen, the probability of lethal = 1/256. Not too little, considering that there may be several thousands of trades in a year. And if the deposit is sold after 6 steps, the probability of trajectory will be 1 / (2^6) = 1/64. Generally very risky.

In general, the topic is interesting, but experienced people have already written that any grider is losing. Perhaps it's very close to the truth.

That's right. But half of trajectories are profitable and half are loss-making. Diablo differs from all the others in that it has only ONE trajectory that will result in a loss. Of all possible trajectories. All the others are profitable!

For example, taking your calculations, when the deposit is zeroed after six steps, 1 time out of 64 cases we will lose an amount equivalent to the initial deposit. In the other 63 cases we will make a profit. Let's assume that the profit is equal to just one step. Taking 10 000 Rubles as an example, we will gain 63 x 10 000 = 630 000 Rubles in 64 Diablo trades while losing 6 x 10 000 = 60 000 Rubles once. Total net profit 630 000 - 60 000 = 570 000 roubles. And so every series of exhibitions.

 
JonKatana:

That's right. But with normal griders, about half of the trajectories are profitable and half are unprofitable. Diablo is different from all the others in that it has only ONE trajectory that will make a loss. Of all possible trajectories. All the others are profitable!

For example, taking your calculations, when the deposit goes to zero after six steps, 1 time out of 64 cases we will lose an amount equivalent to the initial deposit. In the other 63 cases we will make a profit. Let's assume that the profit is equal to just one step. Taking 10 000 Rubles as an example, we will gain 63 x 10 000 = 630 000 Rubles in 64 Diablo trades while losing 6 x 10 000 = 60 000 Rubles once. Total net profit 630 000 - 60 000 = 570 000 roubles. And so every series of exhibitions.

Well, about half of the trajectories that are lethal for normal griders, you are exaggerating. I have posted here on the forum grider, where, depending on the depo, the fatal outcome is also one - many times without reversals.

OK, next. 63 grids will give at least a doubling of the depo? If yes, then in the limit the trading system of "double-double-out" type will have MO equal to 0. If the depo is tripled for 63 withdrawals, it will already have a positive IR. If you answer this question accurately (reasonably, with calculations), honour and praise to you. I doubt there will be at least a doubling... Usually in a game like this the probability of doubling before draining doesn't exceed 50%.

 
alexeymosc:

You've got to be wrong about half of the trajectories being lethal for regular griders. I've posted a grider on the forum where, depending on the depo, the lethal is also one outcome - many times with no rollbacks.

OK, next. 63 grids will give at least double of depo? If yes, then in the limit the trading system of "double-double-out" type will have MO equal to 0. If the depo is tripled for 63 withdrawals, it will already have a positive IR. If you answer this question accurately (reasonably, with calculations), honour and praise to you. I doubt there will be at least a doubling... Usually in a game like this, the probability of doubling before draining doesn't exceed 50%.

I may have bent it - I haven't studied all the griders. How much 63 nets will give - I already wrote above, taking the minimum profit (one step, there is no less). There's a justification and calculation - read it carefully.
 
JonKatana:
I may have bent it - I haven't studied all the griders. How much 63 grids will give - I've already written above, taking the minimum profit (one step, no less). There's a justification and calculation - read it carefully.

It won't pass. You have taken everything too conventionally...

Let's say there is $1,000. The corridor is 0.00100 (100 pips). Let's say we trade with 0.1 lot. So, my dear, to double the deposit we need to take the corridor 100 times. Do you know how many steps will be enough to lose a deposit with such a lot to the margin call? And what is the probability of that happening? That's if you do the math, it all falls into place and you come out at the "about 50%" milestone. It's not exact, but it's what my fifth point is telling me.

 
alexeymosc:

It won't pass. You have taken everything too conventionally...

Let's say there is $1,000. The corridor is 0.00100 (100 pips). Let's say we trade with 0.1 lot. So, my dear, to double the deposit we need to take the corridor 100 times. Do you know how many steps will be enough to sink the deposit to a margin call with such a lot? And what is the probability of that happening? That's if you do the math, it all falls into place and you come out at the "about 50%" milestone. It's not exact, but my fifth point is telling me so.

Let's look at your example. Initial deposit $1000, corridor 10 real pips (or 100 five digits), lot 0.1. To double the initial deposit we need 100 profitable closes of one corridor size ($10 each).

Each time an unfavourable move we lose $10. So, to lose the entire thousand dollars, we need 100 consecutive closes in the red (without taking into account deposits). The probability of such an event according to your same formula is 1 / (2^100) = 1 / 1267650600228229401496703205376. That's a far cry from 50%.

 
JonKatana:

Let's look at your example. Initial deposit $1000, corridor 10 real pips (or 100 five-digit), lot 0.1. To double the initial deposit we need 100 profitable closes of one corridor size ($10 each).

Each time an unfavourable move we lose $10. So, to lose the entire thousand dollars, we need 100 consecutive closes in the red (without taking into account deposits). The probability of such an event according to your same formula is 1 / (2^100) = 1 / 1267650600228229401496703205376. That's a far cry from 50%.

Wait, you completely misunderstand me. When a fatal trajectory occurs, we lose ALL of our deposit. So we either double it or lose it all. Your calculations are completely out of place.
 
alexeymosc:
Wait, you completely misunderstood me. When a fatal trajectory occurs, we lose ALL the deposit. That is, we either double it or lose it all. Your calculations are completely out of place.

If the lot is constant, the only chance of losing the entire deposit is if this event occurs immediately, from the first Diablo exposure. If we do not withdraw money, with each closing in profit (of at least one step size) the chance of losing the deposit is halved.

In your example there are exactly 100 consecutive minus closes before we lose. What is unclear here? 1000 dollars, one close equals 10 dollars. 1000 / 10 = 100 closes. The lot you have defined yourself. And the probability of closing 100 consecutive negative steps, I calculated - 1 / 1267650600228229401496703205376.

1...1516171819202122232425262728
New comment