Mashki and I. Captured by illusion... - page 3
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Давайте попробуем задействовать не статистику, а логику. Что такое разница между машкой и ценой открытия?
МА(0; 15) - Open[0] = (Close[0] + Close[1] + Close[2] + ... + Close[14] - 15*Open[0]) / 15
Каким образом эта величина может быть связана с трендом на нулевом баре?
It is strange that you have this question.
But if you involve "logic" instead of knowledge and experience. I guess it goes like this -
1) we think that the average bar closing price over the last 15 periods means something.
2) In the case of using this parameter in ATS, then if we are at 0 bar, the open has already happened. Decisions can be made.
3) if the current price changes, and it is included in the formula for calculating the average as Close[0] - due to its relatively low weight of 1/15, we will not notice insignificant fluctuations.
Excuses like "the pictures show this and that" are not acceptable. In the pictures, it looks like the crossed masks give good signals - but they don't.
This is not an excuse, it's a common way to illustrate your ideas. An additional task for minutes arises from studying the "picture" and the conclusions to it. ;)
All the more, no one prevents us from checking - the script is in place, so is the CD.
And we have not reached the signals. Where did you see them - I do not know.
There one can indeed see with the naked eye the bimodality of the distribution of distances between the old knee of the low zigzag and the top of the future one (FZZ). But this data should be read a little differently - because the frequency of the event is the lifetime of such a segment. FC- is indeed interesting. After completion of the first step of research its place and meaning I think will become clearer.
--- maybe I don't understand the meaning of the question... ?
I am more concerned about the problem of absence of sampling separation algorithm.
The hypothesis of difference should be tested.
So far we do not know how. :(
Это множество на классы не разбивается. И даже попытка сделать это противоречит самому смыслу идеи.
Какая аасоциация между этими картинками и состояниями рынка ?
Где собственно материал, имеющий статистическую представительность ?
Что вы хотите получить "разделяя на классы" эти 4 картинки, относящиеся к тому же, к совершенно разным объектам ?
These are IMHO exhaustive estimates of the sample so far. That's one.
2) If there will be an algorithm (however crude) for marking points before turning, (maybe even looking from the future), we would now like to notice differences in the behaviour of deviations. - Please feel free to outline it. it will allow us to move forward.
3) The pictures describe:
a) upper left corner - frequency distribution of MA(0; 15) - Open[0] = (Close[0] + Close[1] + Close[2] + ... + Close[14] - 15*Open[0]) / 15
b) right one is already for MA(0,100)...
c) lower left - distribution of bars with known distance of the last extremum at the Minute to an extremum (but in the future) at 5-minute (older TF :)
d) lower right - distance from the closing price to the nearest future extremum in the older TF.
4) If we manage to split the set, these four slices will be even more informative.
---
It makes sense to discuss statistical "representativeness" of the material only if there are suggestions concerning the necessary level of these representations. I have given all the parameters in the illustrations, including N, the sample size.
But in the discussed example with non-normal distribution there is not even a reference to the study period...
Let's introduce standards of reasoned debate.
Only in favour.
--------------
Your assertion about the impossibility of splitting the set, on the other hand, puts me in an awkward position.
And that - that even attempting to do so contradicts the very meaning of the idea.
I don't know what to think.
How to solve the "homework" problem?
;)
Let's try to use logic rather than statistics. What is the difference between the MA and the open price?
MA(0; 15) - Open[0] = (Close[0] + Close[1] + Close[2] + ... + Close[14] - 15*Open[0]) / 15
How can this value be related to the trend on the zero bar?
P.S. I'm not snide, I'm just trying to understand the choice of this particular value. The excuses like "the pictures show so-and-so" will not be accepted. It seems to me from the pictures that crossing of the MAs gives good signals, but it is not so.
МА(0; 15) - Open[0] = (Close[0] + Close[1] + Close[2] + ... + Close[14] - 15*Open[0]) / 15= (Close[0]-Open[0])/15 + ...+(Close[14]-Open[0])/15
This is the average value of the close price deviation from the open price on the zero bar.
Or if expressed in price increments and in case there are no gaps (Open[i]=Close[i+1])
Close[0]=Open[0]+x0
Close[1]=Open[0]=Open[1]+x1
Close[2]=Open[1]=Open[2]+x2=Open[0]-x1
Close[3]=Open[2]=Open[3]+x3=Open[0]-x1-x2
....
Close[14]=Open[13]=Open[14]+x14=Open[0]-x1-x2-..-x13
Then:
MA(0; 15) - Open[0] = (x0-13x1-12x2-...-x13)/15
If I'm not mistaken in calculations, the last increment in price (x0) has a very small value. The penultimate increment makes the main contribution, and then the weights decrease linearly.
If we take MA(0; 15) - Close[0], the picture changes. The last increment will have the highest value, and then the weights will linearly decrease. That's why if the MA is based on the indexes, we should subtract the last index, not the open one.
The analyzed residuals are in fact a weighted sum of price increments with linear decreasing weights
Если же взять МА(0; 15) - Close[0] то картина поменяется. Наибольшее значение будет иметь последнее приращение, а далее веса будут линейно убывать. Поэтому если машка по клозам строится, то и вычитать надо последний клоз, а не опен.
So Achilles will never catch up with the tortoise... ;)
Close[0] - will be known at the moment when it becomes Close[1].
So Achilles will never catch up with the tortoise... ;)
So on the story it doesn't matter, but in real life it's the same problem for calculating the mash itself by clozes.
If you don't want to deal with closis of unformed bar in real time, you can calculate it by ops and subtract ops.
Так на истории это неважно, а в реале эта же проблема и для вычисления самой машки по клозам.
Если в рил-тайме не охото иметь дело с клозом еще не сформировавшегося бара, то можно рассматривать машку по опенам и вычитать опен.
consider that Close[1] is subtracted. Close[0] for reaction to a very steep movement.
consider that Close[1] is subtracted. Close[0] for a response to a very steep movement.
However, in some cases the difference may be crucial (see calculation above).
Давайте попробуем задействовать не статистику, а логику. Что такое разница между машкой и ценой открытия?
This is a special case of MACD, i.e. MA(m)-MA(n), where n=1
MACD, in its turn, can be considered a special case of the linear combination of bars with the sum of combination ratios equal to zero.
And since the mask itself is a linear combination of quotes (with the sum of coefficients = 1), ... everything.
Slightly strange results of the plausibility check of the signals crossing the wagons (new to me, but I know there are coryphaei - they will correct me).
At the period from September 2008 till today (m5) EURUSD the following strategy should give a stat advantage.
1) We look for the distance modulo MaV_MaS. (200 и 15)
2) if it is less than 10 pips and the last top of 33 is higher than the current price by at least 25 pips - feel free to sell.
If the last 33 top is below the current price by at least 25 pips, then buy.
Stoplosses = 125 pips, TP = 600 pips ;) It is better to close in case of opposite signal.
---- this is not a joke.
Now it is worth to test the strategy on the tester. ;)