The most banal trading strategy - page 54

 
Maxim Kuznetsov:

is some new concept in trading that I am not yet familiar with :-)

I'll take that as a yes.

If I am not mistaken, it is called "safe haven".
 
Maxim Kuznetsov:

this is some new concept in trading, not yet known to me :-)

Kipish is an altered word, modified over time, the name of one of the old criminal trades - hippies. This business was born in Odessa and consisted of the following: a young pretty woman hippies would lure a victim-fraught to her flat allegedly for love-making.

Transferred to trading, a similar thing probably happens when victims are lured to enter in the direction of a false breakdown. The result is well known: after a reversal, the victims' stops are triggered).

 
Maxim Kuznetsov:

It is worth opening an economics textbook, a couple of books from 10-20 years ago (not poisoned by traders) "currency operations" and "banking";
Of course, your brain explodes at first sight, but the course becomes clearer and you do fewer stupid things.

recommend something.

 
khorosh:

Kipish is an altered word, changed over time, of the name of one of the old criminal trades - hippies. It was born in Odessa and consisted in the following: a young, pretty woman hippie would lure a victim-freak to a flat, supposedly to make love to her.

Transferred to trading, a similar thing probably happens when victims are lured to enter in the direction of a false breakdown. The result is well known: after a reversal, the victims' stops are triggered).

)))

This, Yes. A dubious event-misunderstanding.

PS Exit!

 

Continuing on with "THE MOST BANAL"

----

We take the last 3 events (no difference buy/sell, we don't distinguish open/close, just control points) from the "not-so-banal".

In theory, is it possible to build a derivative TS on this ?

And if so, would it be more profitable ( less risky) than the original one...

----

my guess is no. Except that the original never loses.

 
Maxim Kuznetsov:

Continuing on with "THE MOST BANAL"

----

We take the last 3 events (no difference buy/sell, we don't distinguish open/close, just control points) from the "not-so-banal".

In theory, is it possible to build a derivative TS on this ?

And if so, would it be more profitable ( less risky) than the original one...

----

my guess is no. Except that the original never loses.

but what about copying the signal or it is not a derivative TS?

Personally, I think you can.

 
Renat Akhtyamov:

what about signal copying or is it not a derivative of TC?

personally i think you can.

with full information - direct copying, without the slightest delay and in the same conditions, is just a duplicate. Not more profitable not more unprofitable.

But there is no information about the direction of the trade. buy or sell ?

in my suggestion, there are 2 more control points like this to decide...

---

or how many checkpoints do you need?

 
Maxim Kuznetsov:

with full information - direct copying, without the slightest delay and in the same conditions, is simply a duplicate. Not more profitable not more unprofitable.

But there is no information about the direction of the transaction. buy or sell ?

in my suggestion, there are 2 more control points like this to decide...

---

or how many control points do you need?

i.e. in essence you offered to train the TS on previous trades.

i suggest that the system can be implemented as a neuron.

i can use the number of previous trades as a parameter.

Well, that's interesting.

 
Renat Akhtyamov:

that is, you have essentially proposed to train the TS on previous trades.

I suggest that the system can be implemented as a neuron.

The system has a parameter - the number of previous trades on some conventional technology.

well, interesting.

even more interesting, if option 3 is "ignore".

In general, induction kind of says that this is impossible - the derivative of TC over the N control points will be no better than the original.

 
Maxim Kuznetsov:

even more interestingly, if option 3 is possible - "ignore".

Generally speaking, induction kind of says that it is impossible - the derivative of a TS over N control points will be no better than the original.

Just neural networks destroy such thoughts, as it is their way of working - not to do things as they turn out badly.

The only BUT. They, like all other TCs, are only able to view what has happened within a narrow framework.

That is they cannot divide: trend / flat.

;)