![MQL5 - Language of trade strategies built-in the MetaTrader 5 client terminal](https://c.mql5.com/i/registerlandings/logo-2.png)
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
If a market is random then surely it is impossible to predict and therefor place trades on? If you mean random as in without patterns then I don't think the currency markets are random. Case in point the double top. How, if the markets are random could such a pattern emerge consistently? If you mean random ie it is impossible to determine at a future price then I would agree with you, but that isn't necessarily random. I would say that you could predict which way the market will move with a certain degree of accuracy.
Personally I believe that the markets follow patterns known as random walks found in an area of maths known as Stochastic Processes. Now from what I've learned so far (only a year and a half into a 4 year course) Markov chains are an important part of time-series forecasting and are used heavily in the forecasting of stock price data.
I doubt anyone trying to construct a neural network will be able to accurately predict the future price of a commodity, mearly tell whether the price will be higher or lower than current. However I am currently trying to do exactly that. I'm hoping to find a Unix neural network I can use at University on their Cray mainframe because it would be a waste not to use a supercomputer if you have access to one. Also I think that a more useful prediction would be of the (Open-Close) or (High-Low) range. I think it might be possible given enough data which is why I'm really hoping someone will have some software recommendations. As added incentive if anyone helps me out with software and using it I'll be happy to share any results I get with them. I'm at Edinburgh Uni and they have a cray XT4 which can run at 60 Teraflops (60 trillions calculations a second) and I really want to take advantage of it. So anyone who is serious about neural network forecasting message me.
as a fxdealer, profit is more important than truth.
maybe we should be more careful when using the word "truth", we don't need faith in market, we need profit. no one sell umbrella in sahara, we just need to find some orders to make profit. and the order(not truth) exists.
.............
Studies of different extrapolation methods showed that a simple EMA extrapolation is "good for" more than 90% for 1 bar prediction (in order to avoid any confusion with anybody, I repeat : it is a 1 (one) bar prediction)
.........................
mladenHi Mladen,
Even if it is only a 1 bar prediction, that is still a very bold statement.
Because that means that on a weekly chart that they can predict more then46 weeks out of 52 weeks the right direction ?....Or on a daily chart that they sit more then 229 days of the 255 days of trading in a year, in the right direction?..
regards...iGoR.
Hi Mladen,
Even if it is only a 1 bar prediction, that is still a very bold statement.
Because that means that on a weekly chart that they can predict more then46 weeks out of 52 weeks the right direction ?....Or on a daily chart that they sit more then 229 days of the 255 days of trading in a year, in the right direction?..
regards...iGoR.or it predicts where ma would go - if it goes up - 90% it will still to go up next bar (lag) untill slope 0 or it turns... : )))))
w/o 0 slope, if ma turning bars/continuation bars ratio 1/10 - your statement is true : )
...
Actually it is not such a bold a statement, but do not take my word for it...
Anyway, as a quick recall, a link to a site, one among other sites that are dedicated to forecasting (and are doing it on an "non amateur" level) can be found here ScienceDirect - International Journal of Forecasting, Volume 25, Issue 2, Pages 215-430 (April-June 2009) (they do that since 1985 and they are quite serious at it. If you go to volume 25 issue 1, there you can download articles and studies for free just in order to see what is it all about)
I post here one of the articles from that free issue (this article deals with ssa, not ema, but the result is almost as interesting
. Also needed to pack it in order to make it an attachment, so just unpack it to read it). Comparative study of more than 20 ways of forecasting (among which ema too) is not from that free issue, so sorry... All that stuff is not a such big news anyway (those "things" exist for quite some time, maybe the trouble is in us and our unwillingness to learn)
_______________________________________________________________________________________
PS: just "eyed" the "they sit more then 229 days of the 255 days of trading in a year" statement. That is very simplified view of forecasting. As usual we (as in you and me) do not agree on that point : I stated it more than once - distribution of losses and gains is not even and thus no one can not take a chunk of time series and say "here is where I will not loose" or "here I will loose". That not even the god can tell you (if I remember correctly there was a saying from Einstein that goes something like this : "God does not play dice". But that is a statement that I must disagree with too
He does when he is bored)
regards
mladen
Hi Mladen,
Even if it is only a 1 bar prediction, that is still a very bold statement.
Because that means that on a weekly chart that they can predict more then46 weeks out of 52 weeks the right direction ?....Or on a daily chart that they sit more then 229 days of the 255 days of trading in a year, in the right direction?..
regards...iGoR.hi
hi cockeyedcowboy
can you explain more how you determine this 3 cycles...?
thanks
lodol
Hello everyone
I am new here but have been trading for 18 months and 15 months fulltime.
:) You mean something like on the picture (that is the continuation of the "fooling around" demo, and that there 91 is the current stream)
_________________________________________________________
Since these things are not meant to be fooling around:
As of traders (talking about institutional traders, bankers, not the individual ones now) and how do they "screw up" : assuming that their knowledge is adequate (which, sincerely, I do not think is the case) I really do not know where does the greed takes over and leads to what we are witnessing today. I mean, how much is enough? And where does the "I want more" stops? Because, when it takes over (the greed), the risks that are taken are enormous and very fast they find themselves doing what Madoff did (btw: a simple question : why do they pronounce his name as if it was Madeoff and not Madoff?) Do not full ourselves - US government "help" for banks is exactly that : covering those stupid ways of risk management (a nice way to say greed) and more or less those banks did exactly the same thing as Madoff
As I said : since the distribution of gains and loses is not even, no one can predict what the next one would be (it would take to predict the outcome of the prediction, and that I dont think even Wolfram can not do) Hence, trading without money management is simply mad (no, not made:))
Steps in that way are done everyday (as far as I know you too did one of very good versions of money management). No system can be done based only on the middle component (the entry itself) without the starting component (how much can I risk) and the end component (how much is enough to earn) "Knowing the future" 90% does not change anything in that.
regards
mladen
Hi Mladen,
Thanks for your links and attachements.
I am very busy at this moment and hope to find some time to read it.
As you say maybe "we" do not understand the word forecasting or predicting.
But what I understood was that they have a 90% reliability of predicting the next bar.
In normal trading terms I do not look to that as taking a chunk out of time series but predicting ONLY the next bar.
One repeats that every new bar. If that predicting has a 90% reliability then that means if I open a trade on monday morning and close it on friday evening whatever size that the losses and profits would be on how they would be distributed then it should be clear to "we"(us) that predicting each bar with a 90% reliability will lead to a huge profit even used without S/L or T/P.
So there needs to be some sort of a catch in that "simple" statement because "we" all know that if this would be possible....then ......
regards....iGoR....................
PS: just "eyed" the "they sit more then 229 days of the 255 days of trading in a year" statement. That is very simplified view of forecasting. As usual we (as in you and me) do not agree on that point : I stated it more than once - distribution of losses and gains is not even and thus no one can not take a chunk of time series and say "here is where I will not loose" or "here I will loose". That not even the god can tell you (if I remember correctly there was a saying from Einstein that goes something like this : "God does not play dice". But that is a statement that I must disagree with too
He does when he is bored)
regards
mladenHi Mladen,
Thanks for your links and attachements.
I am very busy at this moment and hope to find some time to read it.
As you say maybe "we" do not understand the word forecasting or predicting.
But what I understood was that they have a 90% reliability of predicting the next bar.
In normal trading terms I do not look to that as taking a chunk out of time series but predicting ONLY the next bar.
One repeats that every new bar. If that predicting has a 90% reliability on a weekly chart then that means if I open a trade on monday morning and close it on friday evening whatever size that the losses and profits would be on how they would be distributed then it should be clear to "we"(us) that predicting each bar with a 90% reliability will lead to a huge profit even used without S/L or T/P.
So there needs to be some sort of a catch in that "simple" statement because "we" all know that if this would be possible....then ......
regards....iGoR
.........................
As I said : since the distribution of gains and loses is not even, no one can predict what the next one would be (it would take to predict the outcome of the prediction, and that I dont think even Wolfram can not do) Hence, trading without money management is simply mad (no, not made:))
Steps in that way are done everyday (as far as I know you too did one of very good versions of money management). No system can be done based only on the middle component (the entry itself) without the starting component (how much can I risk) and the end component (how much is enough to earn) "Knowing the future" 90% does not change anything in that.
regards
mladenHi Mladen,
My english is not my native language so to put things straight so we would understand each other:
This is what you wrote:
The predicting part (and some of the reasons why I think trends do exist) :
Studies of different extrapolation methods showed that a simple EMA extrapolation is "good for" more than 90% for 1 bar prediction (in order to avoid any confusion with anybody, I repeat : it is a 1 (one) bar prediction)
I understand by what you wrote that "they" can predict with a 90% probability what the outcome of the next bar (ONLY the next bar) will be. Meaning a down bar or an up bar.
That means if they look to a weekly chart they can say with a 90% reliability that the next weekly bar will be a down bar or an up bar.
That means that one takes in a position accordingly the moment that the market opens on sunday evening and close that position the friday evening.
No S/L and no T/P are placed. As said only opening postion and closing positions.
But now I read that you say that no one can predict what the next one would be ??....because losses and profits are not ditributed in an even way ??...
The loss and the gain that one makes has nothing to do with the 90% reliability of predicting if a bar is going to go up or down.
The loss and profit is the effect of the prediction and not the other way around.
Included I have added a Monte carlo simulation based on the weekly chart of the eur/usd. I have even placed the test on 80% probability instead of 90%.
You fill in the hitrate in the red cell (it reads 0.2-- you need to fill it in the other way around 0.8% = 0.2).
The weekly range is according the positons that one would take. That is the close minus the opening.
As you can see if you can predict a 1 single bar ONLY even with a 80% hitrate then you have yourself a holy grail that does not need any MM at all.
And it does not matter if some weekly losses are bigger then weekly profits or the way that they are ditributed.
Even if you push that "new sequence" button a thousand times it want make any difference the way that the losses or profits are distributed.
Even stronger, if you fill in "ONLY" 55% hitrate ( 0.45) you have a no losing system over the long end. That would, with the help of an solid MM, still make a holy grail.
So if you can come to me with a system that can predict the next bar with a 55% reliability and my SuPeR_MM we would make a great team![](https://c.mql5.com/forextsd/smiles/smile.png)
Friendly regards...iGoR
PS. The spreadsheet with the Monte Carlo simulation is based on a spread sheet that "Michel" has made (a member of our forum).
Included there is also the spreadsheet of the weekly ranges on the euro.