From theory to practice - page 924
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
Still on topic, is it possible to squeeze something worthwhile out of the price increases or not? It seems to me that the question is not closed yet.
from two weeks.
I have the suspicion that the nonsense described in this thread works, but only and exclusively for certain trading periods, the so-called "cycles". I remember, in this branch Demko was an expert of trade cycles.
I.e. you cannot choose the size of a sliding window on a ball - it has to correspond to a certain cyclicality of the market.
I found it:
Forum on trading, automated trading systems and testing trading strategies
From theory to practice
Nikolay Demko, 2018.08.30 19:15
Different authors write differently because they all have one thing in common, they all take some kind of periodicity, a reasonable economic periodicity.
I can list them all: trading sessions, the first periodicity, it is repeated daily.
Next, there is a weekly periodicity: every Friday a lot of people open their positions, on Monday. On Wednesday it is the day of double swap.
Then it is monthly. It is understandable, every firm makes monthly reports.
Then quarterly, then seasonal summer-winter.
Annual. Five-year. 12 years.
60-year. Kondratieff cycles.
Added to this is the periodicity of economic indicators. They are not released periodically in time, but with calendar conditions (such as the last Thursday of the month, it may move between periods of 3 to 5 weeks, although on average - 4) but everyone knows the date of their release in advance.
The daily cycle of volatility is the most pronounced, I wrote you about the window = day 300 pages ago.
I remember that. I do use it now - it's theoretically sound, no argument there. But it doesn't feel right... I can't explain it in words yet.
I have the suspicion that the nonsense described in this thread works, but only and exclusively for certain trading periods, the so-called "cycles". In this branch the expert of trade cycles was Demko.
I.e. you can't choose the size of a sliding window on a ball - it has to correspond to a certain cyclicality of the market.
Found it:
finally :-)
The day or two is still partly dictated by the interbank, but there are also futures which are withdrawn after sharp moves and so on and so forth.
and different instruments are subject to different cycles. All of them have weekly cycles, Euro, Pound and Franc (going down) have daily cycles.
On the Jew, you can trace cycles up to 30 minutes, which is why some people say it is "frantic" :-)
I remember that. I do use it now - it's theoretically sound, no argument there. But it doesn't feel right... I can't explain it in words yet.
The easiest way is to take a wave as a window, it is a channel that you can safely analyse.
You are ignoring the waves. That is your mistake.
It is easiest to take a wave as a window, it is the channel you can safely analyse.
You are ignoring the waves. That is your mistake.
I don't know how to pick a sliding window and whether it should be one. And whether they are equal probability, or whether, as Automat thinks, it's the older TFs that take precedence, i.e. some giant windows...
In the past, it was as clear as day to me - the window should be such that a pseudo-Poisson flow of quotes is observed in it - the day is ideal for this.
However, my infamous EURJPY trade showed that this was not the case. Oddly enough, on tests the best results are shown by sliding window models = 2xTF. I.e. = 8 hours(2xH4), 2 days(2xD1), 2 weeks etc.
I don't know how to explain it... There are some cycles in the market, some structure (as Wisard_2018 has written about more than once), but I can't understand it, much less justify it mathematically.
And without that understanding, it's all screwed! Let the other uncles now think and tell you all about it like an interrogation.
I have the suspicion that the nonsense described in this thread works, but only and exclusively for certain trading periods, the so-called "cycles". I remember, in this branch Demko is an expert of trade cycles.
I.e. you can't choose the size of a sliding window on a ball - it has to correspond to a certain cyclicality of the market.
I found it:
Let me say straight away that I have tried it on minutes solely within increments:
1. Build several increments simultaneously with different periods starting from 1440 minutes and open trade at simultaneous breakdown of a channel on different periods.
2. Thin out the series of increments, leaving only those increments greater than three standard deviations in the sample.
3. To thin the series of increments leaving only those that are less than the standard deviation of the sample.
4. Select a dynamic window starting at 1440 minutes so that the asymmetry would be +-1 point and the least kurtosis.
This is not theoretical, but practically tested with unsuccessful results. And no asymmetries and excesses improve the situation.
ps - There are still options!
I have the suspicion that the nonsense laid out in this thread works, but only and exclusively for certain trading periods, the so-called "cycles".
Is it so hard to understand that you are trying to combine completely different things. In the picture below the blue line is the difference between the opening price of M1 EUR/USD and the simple МА of period 201 plotted against it. The red line is actually a downward shifted MA for beauty purposes. You try to trade on the blue line, and if the red line is moving in a narrow range comparable to the width of the blue channel, it seems to you that everything is OK. But it's just random luck. If the range of change of the red line was comparable in amplitude to the range of the blue line, then you could trade that way.
Well, I wouldn't be so categorical. I, for example, have never lost a single account. Another issue is that I have never had any profit - all the time expectation of profit = 0. I cannot overcome this 0, but I know how - I need to find non-randomness in BP, which is 98% random (I agree with this figure given by Comrade Che).
So, to find that 2%, you need insight, not brute force of options. Knowledge + genius, that's what it takes! And if my references to kurtosis and asymmetry don't help, you have to dig in other directions: ACF, entropy, etc.
But nobody's going to do that here, alas... Am I all-powerful to do everything for everybody?!