Where is the line between fitting and actual patterns? - page 42

 
joo:

If it is possible to estimate the degree of correspondence of one phenomenon to another (it depends on which set of events is chosen as a phenomenon in TS), by correlation for example, then it is possible to express this degree of correspondence in % ratio (this is the option I presented). If it is not possible (either there is a response phenomenon, or not), then it only remains to count how many correspondences there were on Sample.

I see.

And in the above example can we say that 26.01.2011 k(Q:A)=50%?

And who would be the General Architect of TC?

 
lasso:

I see.

1) And in the above example can we say that 26.01.2011 k(Q:A)=50%?

2) And who would be the General Architect of TC?

1) I don't know. This is a practical implementation. It should be checked, if it works, it is possible (the definitions, at least, seem to be in order now).

2) You. The architect of the TC you just built. :)

 
I don't see any frequenters of the Pure Mathematics, Physics, Chemistry, etc.: problems for brain training, not related to trading in any way, maybe some of them will speak up? I would like to know the weaknesses of the approach of dividing all the TCs into two types, at least just at the level of inferences.
 
joo:
I don't see so many frequenters of Pure Math, Physics, Chemistry, etc.: problems for brain training, not related to trading, may be some of them will share their ideas? I would like to know the weaknesses of the approach of dividing all the TCs into two types, at least just at the level of inferences.

I don't know if I'm one of the regulars on this thread, but I'll say something just in case.

About the division into categories in general (any) - the idea is reasonable. Allows sorting out similarities and differences characteristic of the highlighted categories. In order to then already apply / not apply taking into account the identified properties. I do not write down my classifications explicitly, though I have been meaning to for a long time - I realize the essential benefit of doing it. And the obtained "document" - the classification table itself - may be useful for many things (I will not enumerate them).

About the division into, specifically, "unlimited-in-time"/"limited-in-time" - probably also reasonable to some extent. But by my broader classification, they both fall into subcategory 2 .2) systems with predetermined fixed stops. // See the excerpt from the classification below.

. 1) systems with continuous prediction // always have a prediction

. 2) systems with pulse forecasting // sometimes (at discrete moments) have a forecast .

. . 2.1) systems with position follow-up // position is closed by a signal to close, calculated in the course of the play when the position is already open .

. . 2.2) systems with predetermined fixed stops // of any nature (real/virtual, spatial/temporal)

. . . . 2.2.1) systems with predetermined fixed space stops // SL, TP, or pending stops/limits with a different opening volume .

. . . 2.2.2) systems with predetermined time stops // no comment, for it is clear from the name

. . . 2.2.3) a combination of 2.2.1 and 2.2.2

...............................

The disadvantage of both is the claim to know the best closing conditions already at the moment of opening. The advantage is the simplicity of the trading system.

Something like this.

 
MetaDriver: . 2) systems with impulse forecasting // sometimes (at discrete moments) have a forecast

Wow, so I'm not the only idiot like this. The fucking noosphere... I'll add a little: in the system whose outline I'm trying to draw now, the forecast should be very rare, "almost never". And the system is based on pseudoscience, i.e. statistics - and nothing else.

2 joo: To be honest, I haven't yet grasped the validity of exactly such a division as yours. I think.

 

http://forum.alpari.ru/showthread.php?t=38804&page=2

Ну, а если серьезно - в любом учебнике по трейдингу сказано, что главное - не вход, а выход. И это воистину так. Доля случайности в рыночных процессах настолько велика, что ТОЧНО настроенные индикаторы не дают никакого выигрыша в деньгах по сравнению с НЕТОЧНО настроенными Наоборот, излишняя фильтрация приводит к снижению общего количества сделок, то есть - к уменьшению прибыли, нисколько не увеличивая при этом профит-фактор.

another 40 pages of thoughts with no practical use ;)

As for the subject - try to optimize not the entry/exit parameters and SL/TR, but the deal time, not fixed, but between the moments of volatility narrowing - the thought has long been running in my head, but I do not know what I am doing all the time ))))))

 
IgorM:

http://forum.alpari.ru/showthread.php?t=38804&page=2

another 40 pages of thoughts with no practical use ;)

As for the subject - try to optimize not the entry/exit parameters and SL/TR, but the deal time, not fixed, but between the moments of volatility narrowing - the thought has long been running in my head, but I do not know what I am doing all the time ))))))

Igor, you're cheating... You've already done a narrowing of the volatility channel and it worked quite well... :)
 
artmedia70:
Igor, you are cheating... You've already done a narrowing of the volatility channel and it worked quite well... :)


...Ya!

with all these searches you can't remember what you've found and why else you need to look - sorry, but then for the point: the forum is evil!

)))

SZS: Artem, thanks for some intellectual stimulus, now I have no time to sing your praises (at home early traditional vanity))) ), but in order not to forget, I've drawn your halo on your avatar with a marker! )))))))))))))))))))))))

 
IgorM:

http://forum.alpari.ru/showthread.php?t=38804&page=2

another 40 pages of thoughts with no practical use ;)

As for the subject - try to optimize not the entry/exit parameters and SL/TR, but the deal time, not fixed, but between the moments of volatility narrowing - the thought has long been running in my head, but I do not know what I am doing all the time ))))))

I don't know to whom this post was addressed. But the boldface indicates that it belongs to the second type. The lifetime of a trade (according to the second type) does not have to be fixed (it only has to be of a finite length and can be optimised), it can also be a function of the volatility value.
 
MetaDriver:

I don't know if I'm one of the regulars on this thread, but I'll say something just in case.

About dividing into categories in general (into any categories) - the idea is sensible. Allows sorting out similarities and differences specific to the categories highlighted. In order to then already apply / not apply taking into account the identified properties. I do not write down my classifications explicitly, though I have been meaning to for a long time - I am aware of the usefulness of such an activity. And the obtained "document" - the classification table itself - may be useful for many things (I will not enumerate them).

About the division into, specifically, "unlimited-in-time"/"limited-in-time" - probably also reasonable to some extent. But by my broader classification, they both fall into subcategory 2 .2) systems with predetermined fixed stops. // See the excerpt from the classification below.

. 1) systems with continuous prediction // always have a prediction

. 2) systems with pulse forecasting // sometimes (at discrete moments) have a forecast .

. . 2.1) systems with position tracking // position is closed by a signal to close, calculated in the course of the play when the position is already open .

. . 2.2) systems with predetermined fixed stops // of any nature (real/virtual, spatial/temporal)

. . . . 2.2.1) systems with predetermined fixed spatial stops // SL, TP, or pending stops/limits with a different opening volume .

. . . 2.2.2) systems with predetermined time stops // no comment, for it is clear from the name

. . . 2.2.3) a combination of 2.2.1 and 2.2.2

...............................

The disadvantage of both is the claim to know the best closing conditions already at the moment of opening. The advantage is the simplicity of the trading system.

Something like this.

I see. It would be interesting to know to what extent these types of TS, in your opinion, are prone to fitting.


Mathemat:

....

2 joo: Honestly, haven't really grasped the validity of just such a division as yours yet. I think.

I will not give my calculations to clearly substantiate it - I do not possess them. I will not give mathematical calculations to prove it clearly - I don't have them. As to the subject - I've tried to understand how a boundary between adjustment and optimization can be found (it hasn't been done until now), what this boundary may depend on, which TS are inclined to adjust and which aren't. It is an attempt to understand what is a "Regularity" and how to look for it and use it.

Summing up, I want to say that during these last few days a huge amount of work on understanding processes in the market has been done, perhaps more than the previous 2 years. Much remains untold, but not because I am sorry to share, but because it is at the level of feelings and to say what would be understandable to others is impossible. But, unequivocally, the way is right, at least now it is clear which way to go (and then the spring will show, where to shit...).