Damned Martin - page 34

New comment
 
Maxim Kuznetsov:
By the way about martins, K->1+sqrt(1/2) (K tends from above to the specified limit) , provides an optimum of profitability/time_to_breakdown. K=1.73 is suitable for most cases

Man, you've clearly expressed something interesting, only too briefly. I didn't get it at all, unfortunately.

 
K is the martingale coefficient, in this case expressed to "lot multiplication"(present+1). At K=2 the lot is multiplied by half in the wish "to get all for 1 trade", but the number of possible multiplications is, to put it mildly, small (the depo is very much present). At lower coefficients. The "steps to death" is bigger (or you can take a bigger lot), and the expectation of ruin is bigger, but you are in drawdown longer. When K is roughly as indicated, most algos with martingale on board gain more max balance until the inevitable stop out.
 
Maxim Kuznetsov:
K is the martingale coefficient, in this case expressed to "lot multiplication"(present+1). At K=2 the lot is multiplied by half in wish to "get all for 1 trade", but the number of possible multiplications is small to say the least (the depo is very much present). At lower coefficients. The "steps to death" is bigger (or you can take a bigger lot), and the expectation of ruin is bigger, but you are in drawdown longer. When K is roughly as indicated, most algos with martingale on board gain more max balance until the inevitable stop out.

Some will say that optimization may show a different appropriate value on the N-th period of the graph

 
Ivan Butko:

Some will say that optimisation may show a different appropriate value at the N-th period of the graph

Optimisation by an MM "optimiser" is beyond good and evil :-)
 

It is not possible to beat back losses with one order with a lot multiplied by K, but with several orders without lot multiplication, each of which opens only when there is a signal. One of my latest EAs is based on this principle.

Test since 15.08.2016. Maximum drawdown is less than 10%. Tester optimizer was not used.


 
khorosh:

It is not possible to beat back losses with one order with a lot multiplied by K, but with several orders without lot multiplication, each of which opens only when there is a signal. One of my latest EAs is based on this principle.

Test since 15.08.2016. Maximum drawdown is less than 10%. Tester optimizer was not used.



Why I still don't see it in the market?

 
khorosh:

It is not possible to beat back losses with one order with a lot multiplied by K, but with several orders without lot multiplication, each of which opens only when there is a signal. One of my latest EAs is based on this principle.

Test since 15.08.2016. Maximum drawdown is less than 10%. Tester optimizer was not used.


no difference. If the volume in the market follows the real drawdown (including locks) - we are facing a martingale.

 
Ivan Butko:

Why am I still not seeing it in the marketplace?

Why sell grails?)))
Maxim Kuznetsov:

it makes no difference. If the volume in the market follows the real drawdown (including locks) - we are facing a martingale.

Yes, I'm martingale. But the variant using an increased lot at once, with the help of which one tries to cover losses at once, has one drawback: this order with an increased lot may be wrong and you will have to increase the lot again; the total lot quickly reaches unacceptable values. In my example, after two wrong orders a symmetrical lot is formed. This allows you not to hurry to open the 3rd order and enter only if there is a good signal.
 
khorosh:
Why sell grails?))Yes martingale. But in the variant where an increased lot is used at once, with the help of which one tries to cover the loss immediately, there is a disadvantage: this order with an increased lot may be wrong and one will have to increase the lot again, and the total lot quickly reaches unacceptable values. And in my variant after two erroneous orders only a symmetrical lot is formed.

A well-masked LaBoucher ?

 
khorosh:
Why sell grails?))

Mm... maybe a signal?

1...2728293031323334353637
New comment