Using Neural Networks in Trading. - page 17
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
Well, then you're not expressing it correctly. Predicting the price means saying that in a while the price will be 1.3567, for example. The direction is whether the price will go up or down at this point in time. All these predictions are based on the data you are using, it does not matter if it is a neural network or a standard TS.
I'm afraid that you are not stating your point correctly.
The man enters the trade and puts TP of 50 points, therefore he predicts the price, its behavior in the future relative to the entry point (entry point +50=1.3567). This is a forecast. Time only still it is necessary to add.
And here is the statement I predict the direction ? raises a lot of questions.
1. Direction, 1 tick up is a direction ? but two ticks ?
2. Let me complicate the question 3 ticks up 5 ticks down is what direction ?
3. even more complicated 100 ticks up and 100 ticks down. The first movement takes place within a minute, the second one - by the end of the year. The countdown is calculated from the entry point.
Right, it's not just a forecast of the direction, but the value of this direction and time during which the forecast is valid. (100 pips now for 15 min or 100 pips by the end of the year - feel the difference).
Mia.... the forum is certainly a good thing, but...... how different everyone's views and opinions are is a nightmare. The path that the world has long traveled is being taken here for the hundredth time. They have long since stepped on the same rake that everybody stepped on and learned how to avoid it. Well, as the saying goes, this is the forum. Let's go! Ninety boys! Ninety boys!
Mia.... the forum is certainly a good thing, but...... how different views and opinions everyone has is a nightmare. The path that the world has long traveled is being taken here for the hundredth time. They have long since stepped on the same rake that everybody stepped on and learned how to avoid it. Well, as the saying goes, this is the forum. Let's go! >> boys ninety! >> boys ninety!
Can you tell us more about the highlighted in your post or give us a link where we can read it.
When you cannot buy or sell and you have a large position to close, there is no liquidity and no price, because the uncertainty is high.
It is not the broker's fault that there is no liquidity at the moment.
The situation when there is almost no supply and demand arises from time to time and is quite critical.
On the market side (up to the dealer, price formation) the price is not continuous.
What does the continuity property give you? Are you going to derive some kind of formula?
1. Knowing this, I can precisely calculate the value of noise, and filter it out. I can show you a picture, but it's only in the evening.
2. Methods of predicting discrete and continuous values are different.
3. Gives me a clear understanding of what is the best DC. There is a criterion and a clear indicator that can be calculated and compared (if the speed of withdrawal is the same).
This is just a quick glance ...
Quant again, I'm talking about the price. About its physical nature. It (the price) is always there. No matter how or who quotes it. Quoting is a representation of price in a discrete form. Just think - well, a broker or a bank does not give us a quote, so what? Has America gone down the drain or Europe all at once? Has supply and demand disappeared? Doesn't anyone need dollars or euros?
no, Prival, you don't understand... we should not be interested in the physical nature or the real price of the market... or rather, we should, but only to the extent that it correlates with the price given to us by the BC... it's this price - the price at which the order is triggered - that we have to recover... in terms of DSP, do a digital-to-analog conversion... in terms of statistics - find the non-random component... because you can only predict the future on the basis of a pattern... that's the pattern we're identifying... you can set it analytically (formula), you can just fit a neural network or a simple expert (although as Neutron said, optimizer in MT is essentially a single layer perseptron)...
as I said before, this price sometimes coincides with an indicative quote and sometimes it does not... As a matter of fact, we should not see Bid or Ask on the screen, but a range where the price of order triggering will most probably fall... In normal brokerage companies, pending orders trigger precisely according to an indicative quote, market ones often with requotes...
In other words, the price has changed dramatically... If you don't know what kind of market conditions you are trying to reach, you may start a new one. so you have to analyse it...
I am afraid that you are not expressing your point correctly.
The man enters the trade and puts TP of 50 pips, therefore he predicts the price, its behaviour in the future relative to the entry point (entry point +50=1.3567). This is a forecast. Time only needs to be added.
The person predicts +50, and the price passed +49 - the forecast was successful?
Or the proverbial +50 +/-3% And statistics for xforex is not "pseudoscience"? ;)
'
And here to predict movement, i.e. when the price will go not less than 40 pips, yes sign - direction - much (IMHO :) ) easier.
But the statement I'm predicting the direction ?
Come up with the right "Teacher":
"-1 is when .... if the nearest future LowestLow=-100, with HighestHigh not more than +20 during this time...and all this is considered on an n-bar forecast horizon" etc.
This is where everyone's understanding of Flat/Trend is born... and in the context of this thread and predicting ( at least) exactly Flat or Trend
'
The same ZigZag (how many bytes have been killed :), discussing good teacher or bad) will tell you (at the learning stage) whether there has been a change of direction!!!! or not.
'
And +50 when opening a trade (yes even +128) is pips!!!! (IMHO, i.e. my understanding of the foreign term 'scalping' and the kitchen term 'pipsing').
:)
ZS. That Kagi/Renco is cooler I don't argue :), but certainly not forecasting "exactly +50" !
Can you tell me more about what you highlighted in your post, or give me a link where I can read it.
I haven't seen any specific, practical advice on this subject on the Internet, let alone free of charge. Of course, the original source is Shiryaev, but from reading this very clever book, my untrained brain curls up into a tube and refuses to understand so many letters and numbers. But maybe someone will succeed? After all, the main thing is not to read and understand - but most importantly to make the right conclusions and to apply in practice all that is written there. ))))
flag in your hand, I'm happy for you! :)
ZS. what do you approximate with, if it's not a secret?
trend curve derived from regression lines =)
trend curve obtained from regression lines =)
cool... let's not go into how you get a trend curve from regression lines... I can roughly imagine that (although frankly I have my doubts about the adequacy of such method, but okay)...
but even the graph shows that the variance of the random term (the noise component) is quite large in some areas... Anyway, I can hardly believe that it's even bigger on smaller timeframes... although maybe it's just a feature of the technology...
Anyway, I can't imagine that an ordinary linear regression would have a larger variance on small timeframes than on large ones...