Trading probability - page 5

New comment
 

Did you all buy the 2010 maths labs :)))).... phew, did you download them? :)

 
SProgrammer >>:

Мат лаб то 2010 все уже купили :)))).... тфу-ты, скачали? :)

I use MathCad 6.0 Professional Edition from 1995. 5Mb in the archive, works fine without installation.

 
I take it that the swaps have been neglected...
;)
 
getch >>:

В результатах "Задачи о разорении" не получается интуитивно понять некоторый момент.
Получается, что при вероятности выигрыша в сделке > 50%, вероятность разорить другого игрока растет с количеством денег у того и у другого. Даже когда у них их одинаковое количество.

The more money you have, the less likely you are to go bust.

Example:

You play against a player who has the same amount of money at the start.
If you are lucky 60% of the time (you win a coin), then with an initial deposit of 1 coin there is a 60% chance of ruining your opponent, and with an initial deposit of 10 coins there is a 98% chance.

 
getch >>:

Чем больше денег, тем вероятность разориться меньше.

Пример:

Вы играете с игроком, у которого столько же денег в начале.
Если вам везет в 60% случаев (выигрываете монету), то при начальном депо 1 монета вероятность разорить соперника равна 60%, а при начальном депо в 10 монет - 98%.

Then the expression "grow profits and cut losses" is fundamentally wrong: cut profits and stop losses.

 
Urain писал(а) >>

Then the expression "grow profits cut losses" is fundamentally wrong: cut profits and stop losses.


This refers to money management - such as the optimum f, etc. It means the choice of how much to put in each transaction (what part of the deposit). But all these theories are designed to ensure that the probabilities of winning and losing are constants and unchangeable, which in reality is unattainable.

 
Do you want an example of over-sitting (and martin at the same time) from the classics? Accidentally came across (didn't remember this fragment) in Ilf and Petrov. I'll delete it later - it's long. But what language!
"About three years ago, when, for the first time since the revolution, honey
subjects accepting life insurance reappeared, Varfolomeich decided to enrich himself
at the expense of Gosstrakh. He insured his grandmother of one hundred and two years, a venerable
woman whose age made the entire Gusische proud, for one thousand roubles. The ancient
woman was afflicted with many senile illnesses. So Varfolomeich had to pay high insurance premiums
. Varfolomeich's calculation
was simple and correct. The old woman could not live long. Bartholomewitch's calculations
ich said that she would not live even a year, for a year would have to pay
sixty roubles of insurance money, and 940 roubles would be a profit
almost guaranteed.
But the old woman did not die. For the one hundred and third year she lived quite happily
. Disgruntled, Bartholomew renewed his insurance for the second year.

In her 104th year, the old woman had recovered considerably - she had developed an appetite and the index finger of her right hand, twisted by gout
for ten years, unclenched. Varfolomeich was dismayed to see that, having spent one hundred and twenty
rubles on his grandmother, he had not received a penny of interest on the capital.
Grandma
didn't want to die: she was capricious, demanded coffee and one summer she even crawled out to the Place de la Commune Paris to listen to a newfangled fiction -
music radio. Varfolomeich hoped that the musical flight would finish off
the old woman, who indeed fell ill and lay in bed for three days,
sneezing every minute. But the body prevailed. The old woman got up and demanded pussy
la. I had to pay the insurance money for the third time. The situation became
unbearable. The old woman had to die and yet she didn't. The thousand-
ruble mirage was melting away, the deadline was running out, the insurance had to be renewed.
Disbelief gripped Varfolomeich. The damned old woman might have lived another
twenty years."
But then the MM triggered in Bartholomewicz's mind:
"Better to lose, he decided, one hundred and eighty roubles than two hundred and forty,
three hundred, three hundred and sixty, four hundred and twenty or maybe even che-
hundred and eighty, not to mention the interest on the capital."
 
Avals >>:


это к управлению капиталом относится - типа оптимальная f и т.д. Т.е. выбор сколько ставить в каждой сделке. Но все эти теории рассчитаны на то что вероятности выигрыша и проигрыша есть константы и неизменны, что в реальности недостижимо


You are probably right that the author was referring to reinvestment, but I decided to check the statement literally.
Well, if there is a formula, why not put the parameters into it, maybe something will come out. Here is the result :
As I wrote before the tested EA makes a random entry using this formula
void RANDOMIS()
{//+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+
 TYPE=-1;
 if(rBars==Bars)return;
 int mr0=MathRand()%2;
 int mr1=MathRand()%2;
 int mr2=MathRand()%2;
 int mr3=MathRand()%2;
 if(mr0==0 && mr1==0 && mr2==0)
   {if(mr3==0)TYPE=0;
    else TYPE=1;
   }
return;
}//+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+
i.e. not only the direction is random, but the opening time is also chosen by the RMS with a normal distribution.
then let's set the TP/SL 100/100 and two options for approaching the levels to the market
kTP - takeprofit approach speed, kSL - stop loss approach speed
respectively kTP/kSL 1/0.5 and 0.5/1
Thus, the test results, the pass - the averaging of 100 random measurements at one period of quotes at one TF all the same
1/0,5
-8718
-8315
-9369
-8205
-7748
total average -8471
0,5/1
-10954
-9968
-10991
-10372
-11919
total average -10840
Conclusion Shifting the probability of reaching the level towards takeprofit we get a stable gain before the opposite variant.
 

Gentlemen, do you even read the forum? I gave you the code to run in the tester as far back as March 28th - https://www.mql5.com/ru/forum/124836/page20 

extern int       tp=25;
extern int       sl=25;
extern int       mins=60;
int init(){
   return(0);
}
int deinit(){
   return(0);
}
static int r=0;
static datetime st;
int start()
{
   if ( Time[0] != st ){
      st=Time[0];
      
      r--;
      
      if ( r <= 0 ){
         double d = MathRand()/32767.;
         r = d * mins*60;
         
         if (  MathRand() > 32767/2 ) 
            OrderSend(Symbol(),OP_SELL, 0.1, Bid, 0, Ask+sl*Point, Ask-tp*Point );
         else
            OrderSend(Symbol(),OP_BUY, 0.1, Ask, 0, Bid-sl*Point, Bid+tp*Point );
      }
   }
}
And even more than a year ago - gave you the indicator - https://www.mql5.com/ru/forum/113106

And all are investigating something here - and then get offended - that I'm not sharing ...
And fzuke ... :)
 

I even formulated the CONCLUSION - https://www.mql5.com/ru/forum/124836/page13

*** правило которое работает всегда - да элементарно - :) ..... - Итак правило - При случайном входе ( покупка или продажа и время ) вероятность срабатывания стопа или тейка будет ПРОПОРЦИОНАЛЬНА их размеру. То есть если TP = 20 а SL = 20 то веротность закрытия в прибыли будет равна верятности закрытию с убытком. Не зависимо от тренда и валютной пары и времени в истории. Ну а если TP = 2* SL то вероятность убытка будет в два раза выше. :) Доказывается через интегральную функию или также называемую Гаусовым интегралом, применяется как раз для расчета того какая вероятность будет. :) И это будет работать даже с учетом того что на рынке у нас так называемое устойчивое распределение. :) или лучшек называть его по имени великого Леви. :)

123456789101112...15
New comment
Reason: