a trading strategy based on Elliott Wave Theory - page 258
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
Of course, before developing your ToR you must have estimated the average yield per transaction for the suggested trading strategy. Tell me, what is the expected profitability excluding brokerage commissions and for which currency pairs? Does it cover the spread?
Let me remind you that the sample should be representative.
Hi Sergei.
Until I reached my degree of incompetence, I was a DB architect and project manager - it is my professional skill, although I have already passed through it. I had the chance to create more "mature" systems for "mature" firms (for example a data warehouse for 120,000,000 facts is not a problem). All the components are, in general, already there (I will buy the missing computers, and it's not a problem) the only thing left is to develop the dll. Sergey, you also use MathCAD for your research and you know very well, that one "tightly written" page in this program can correspond to dozens, or even hundreds of pages in normal language. After evaluating everything, including the duration of calculations, I've come to the conclusion that I can't just switch completely to MT. It's a very long time.
As for the TS, it doesn't exist yet, but there is a forecasting subsystem consisting of 9 modules. It is being tested now. At the moment there are 273 forecasts, 41 of which I have found wrong. The appearance of forecasts is approximately the same as shown in the post "grasn 13.03.07 19:19", i.e. squares of different size (there were just dots in that post) that symbolize reversal areas and channels connecting them. And no more extra lines and curves, which I always get lost in when I look at published material.
Forecasting is on the clock (optimal ratio for anticipated deposit/possible losses and forecast accuracy) and forecast horizon can be from several days to a month or even more, let me remind you, it's not set in the system, it's determined by a specific time series. If the forecast is good the yield is good, of course without any real deal point in the reversal zone.
There are two main problems: the first one is improving accuracy of a forecast model and "catching" an optimal entry/exit point in the reversal zone. If the second problem is more or less clear, the first one requires explanation. The reversal zone can be at the same level, i.e. the channel change area is at the level of the reversal zone, but the calculation can be wrong in time. It turns out like with the weather. By the way, there is a good anecdote on this subject:
***
They decided in one city to shoot climatologists. They cause a lot of troubles: first they miss a flood or fail to warn us about drought; all in all they cause only losses and casualties. Drum roll, clear commands, "...toot", "...aim" and at the last moment a citizen runs out of the crowd of spectators and yells: "wait, don't shoot them, they can still do good!!!!". The Chief Justice asks with some hesitation "you think? Well, OK, what do you suggest?". The citizen, excited by this opportunity, makes a new suggestion "Let's hang them! Let them show the direction of the wind!"
***
The TC itself, as well as the process of controlling the compliance with the forecast, seem trivial to me and at the moment are of no particular interest.
Here is the situation, it is close and far from real trading at the same time, although I have already used it a couple of times, but it seems I have already bragged about it. :o)
to cooper123
For information. The calculating scheme you suggest I have already implemented. Although my aims were slightly different, namely to stay in the familiar software environment, but the philosophy is the same: divide and conquer.
I can share with you an Expert Advisor that will implement the exchange, with the purpose of joint testing and so on. Truth I haven't made dlls and made file swaps - a more universal thing, as it seems to me.
However, he sells his EAs but he is a good programmer and I don't need to work out anything. I have a feeling he will not have to work there. he also works with files.
since you have decided to pay anyway, it may be interesting.
if it is about the database. a database is good for smart selections and there are simple price series runs and nothing fancy. i also keep it in a file and the filling time is good, moreover it needs to be updated only once a week.
Good luck.
Thank you very much for the offer. Of course we need it! At least it will bring the comprehensive testing much closer. Admittedly, thought about such a scheme, at least for the first draft and it's quite possible that the exchange of files will work just as effectively. And cram data files into the database, easier than dealing with dll (and it's not about money, but speed, although it is possible that in the end I will stop at a dll).
You can email me the experts at grasn@rambler.ru or post them here, as you like (maybe someone will be just as interested).
about stuffing is probably a question of what to put, what to put where, the main question.
But as for the speed, I don't see any advantage of dll over files. Unless, again, you are used to it and want to do it that way. I get data, calculate it, look what's in the file, whether it's new data or not, and slip it into slipstream until the next minute. The whole calculation is within seconds (in simple strategies I have a year in minutes, I've made it in about 7 seconds, with linear regression in 3 months with a window of 100 bars in 3 minutes, and although I get the minute bars, I work higher than the 5-minute bars. What is the rush in such conditions?
Moreover, for me the dlls are fatal, I have to learn them. It is possible to use processes and pips, but it's not worth it.
...On the previous page IronBird provided a link to a post where UP on the forexclub forum publishes his mathematical proof that profitable trading is possible in forex. And (!) not due to violation of the Markov process, but just based on the assumption that it is a perfectly random, i.e. Markov process.
The actual link is http://forum.fxclub.org/showthread.php?t=22097&page=3
Yura, hello!
I read the post by my esteemed UP on the cited link once again. To be honest, I had no desire to comment on it. First of all, the author is right in many respects. Secondly, he confuses the Markovian process with a random - Wienerian, and this is not good. Besides, UP exploits as an estimate of "non-randomness" of time series (RT) samples the value of difference of the distribution function (DF) of the first differences from the exponential distribution. In fact it is an integral index and the expediency of its use is doubtful, of course. Thirdly, martingale belongs to a kind of money management, and is not a TS as such.
For such estimations it is more correct to use correlogram for the first BP difference. From its analysis one can draw a conclusion about the MARKING antipersistence of BP on small TF, and it is impossible to draw a conclusion about perspective of trading on large TF (like the daily). By the way, the author has correctly noted an interesting feature of BP with marked antipersistence: one can build a profitable trade opening randomly and setting TP and SL of a known value. There is no miracle here, it is possible to show strictly mathematically the possibility of stable marginal profit in this case. What is of interest is the estimation of the return per transaction. Unfortunately, on available instruments the profitability does not exceed 0.5 points per transaction. However, when combined with, for example, an autoregressive model, it allows to cover the existing spreads.
By the way, yesterday I tested the moving average based on Batterworth's LPF with the minimum possible phase delay. And what do you think? Trivial, rollback TS based on this TF has shown the average yield of 5-8 points per trade for most of the instruments. Pretty sure I messed up somewhere, but if not... then this is a breakthrough!
It would be interesting to look at this MA and compare its delay to the one I made.
Is this your code or someone else's ? Is it available somewhere ?
"having the lowest possible phase delay" - is this a mathematical result or implied from the currently available ones ?
to Neutron
I apologize that I got into the discussion, it's not like I was asked. :о) Unfortunately the distinguished UP gave a proof in general terms, I am not a statistical guru though I use it abundantly in my predictions and of course I may easily be wrong with my conclusions.
The way I see it, this integral indicator is essentially one of the criteria for determining a trend, based on the general emission analysis approaches for an exponential distribution. Criteria of this class work well enough and a proven local trend would be a good basis for "non-random" trading. But it is questionable for me to gracefully go to the choice of this particular distribution. It seems to me that the real results will be as different as "a steam locomotive from a bicycle".
"having the lowest possible phase delay" - is this a mathematical result or implied from the currently available ones ?
Yes, I wrote the script by myself.
Minimal phase delay (MF) means a theoretically possible minimum MF for a DF with a specified a priori steepness of the AFC cut-off edge and the amplitude of the beats in the passband. Excellent theoretical material on recursive filters can be found here:
https://www.mql5.com/en/forum
By the way, I managed to find a pair that shows H-volatility greater than 2 and allows to implement a profitable H-cougi strategy! The graph of the average earnings per trade (yield) as a function of the discreteness of the split is given below on the left and the dependence of the yield including the 7-point spread is on the right.
However, the question about the stability of the criterion for this instrument remains open...
Is this your code or someone else's ? Is it available somewhere ?
"having the lowest possible phase delay" - is this a mathematical result or implied from the currently available ones ?
Yes, I wrote the script myself.
The minimum phase delay (MF) implies the theoretically possible minimum MF for a DF at a predetermined steepness of the AFC cut-off edge and the amplitude of the beat in the passband. Excellent theoretical material on recursive filters can be found here:
https://www.mql5.com/en/forum
Sergey, I don't understand, do I have any opportunity to compare such МА with mine or not?
There must be an error in your link. The zip.gif is a file, not a web page. It's a little picture like that.
And there is no such page, let alone a splendid material. :-((
https://c.mql5.com/mql4/forum/2007/03/1.zip
As for comparison, I can directly post the code written in MathCad, MQL, screenshot from MT4 terminal or wait until you get acquainted with the given material yourself. What do you choose?
Igor Kim here has made an indicator for drawing regression channels.
It may be useful for manual trading.
http://forum.kimiv.ru/viewtopic.php?p=3170#3170
I drew a picture from it.
It may be me thinking too much but I do not understand what is a channel crossover.
It looks like a typical situation, but there is no channel crossover as such.
There are channels with different scales. I do not understand how a pivot zone is formed by crossing channels.
Maybe someone can explain this picture.
As for comparison, I can directly post the code written in MathCad, MQL, screenshot from MT4 terminal or wait until you get acquainted with the given material yourself. What do you choose?
Thank you, Serguei! I have watched it with great interest. It's a cool book with good practical application. And the maths there is quite appropriate. However, it will take me a lot of time to digest it all and then also to link it to the market.
That's why if you give me the right to choose I'll choose MQL. :-))
Here or directly to my mailbox - as you like.
Thank you in advance.