Machine learning in trading: theory, models, practice and algo-trading - page 1498
![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
As for the gaps you want to split into multiple bars.
I would prefer not to trade at that time, but wait it out. I don't think so. 20-50 pt scalping will not work there because of spread widening. I've seen a 200 pt (5 mark) spread myself. With tp/sl 20-50 pt all deals will be closed by the SL, and not 20-50 pt, but by the worst spread price, i.e. 200 points worse than expected. It only depends on the spread, price jumps of 30-100 pt per tick will also make the strategy unusable.
A strategy with large stops or no stops at all will do.
If you define these time cycles, it turns out that this is the memory of the process, that time is the only parameter that distinguishes real BP from SB.
The cycles on M1 as well constantly change from 3 minutes to several hours. If gap candlesticks are divided into separate bars, the range of possible periods extends from 3-5 seconds to several hours. to several hours. In my opinion this would be even more unstable.
As for the gaps you want to split into multiple bars.
I would prefer not to trade at that time, but wait it out. Scalping at 20-50 pts won't work there due to spread widening. I've seen a 200 pt (5 mark) spread myself. With tp/sl 20-50 pt all deals will be closed by the SL, and not 20-50 pt, but by the worst spread price, i.e. 200 points worse than expected. This is only from the spread, price jumps of 30-100 pt per tick will also break the strategy.
Where is that quote from? I didn't write anything about gaps
nothing will work, tick bars give only a small advantage, i.e. cropped, by which the "memory" is still very difficult to pull out
And these volatility cycles are not always periodic either and sometimes there is a hell of a lot there, though clearly not SBWhere did you get that quote? I did not write anything about gaps
The other 95% of bars are unlikely to contain anything interesting inside them.
Switching to ticks implies exactly the deployment of gap bars. After all, they can be deployed to 10-20 separate bars.
It is unlikely that 95% of other bars contain anything interesting inside them.
In short, it's all chiromancy for sure, but Alexander can
In any case, without it no "patterns" can be extracted, no NS. And with this - very difficult and maybe impossiblenothing will work, tick bars give only a small advantage, i.e. cropped, by which the "memory" is still very difficult to pull out
And these volatility cycles are not always periodic either and sometimes there's a hell of a lot there, even though it's clearly not the SBIn short, it's all chiromancy for sure, but Alexander can
:))) I've found something in the time structure and I'm using it. Although I'm still afraid of my own knowledge :))) I loaded my TS in such a way, that there are almost no trades now. But I can fix it.
Briefly, my opinion to apply NS to BP - it is necessary to use a certain sliding window, measured in the number of thinned ticks. It should not be any 1000 or 2000 and not calculated, for example, from Chebyshev inequality, but it should be tied to a timeframe - a cycle. This cycle is a bit floating and it is IMHO a trading session. And then a day, a week etc. These are periods, i.e. a full cycle of price movement within the range of local minimum and maximum. Of course, there are semi-periods for finding, for example, a maximum. And cycles from above. In other words, the market in time is like a nesting doll :))).
If NS works only in these time zones, I think it will do the job.
Channel strategies tied to these cycles work. And my state proves it. So far :)))
:))) I have found something in the time structure and I am using it. Although I'm still afraid of my own knowledge :))) I loaded my TS in such a way, that there are almost no trades now. But I can fix it.
Briefly, my opinion to apply NS to BP - it is necessary to use a certain sliding window, measured in the number of thinned ticks. It should not be any 1000 or 2000 and not calculated, for example, from Chebyshev inequality, but it should be tied to a timeframe - a cycle. This cycle is a bit floating and it is IMHO a trading session. And then a day, a week etc. These are periods, i.e. a full cycle of price movement within the range of local minimum and maximum. Of course, there are semi-periods for finding, for example, a maximum. And cycles from above. In other words, the market in time is like a nesting doll :))).
If NS works only in these time zones, I think it will do the job.
Channel strategies tied to these cycles work. And my state proves it. So far :)))
I know, but it's too farfetched... I've got no idea what to attach them to, like a plantain. I'll also try a kind of channeler, I think I figured out how to do it.
There, even in the book, which I threw you from Quantum and a kind of manager - says that do not try to do everything yourself, it's life-threatening. This book is for the development team, and I can't promise anything.
But we are the smartest ones here.
Yes, but it's too damn complicated. The NS themselves eat up half of my brain, and I have to think what to attach them to, like a plantain. I'm going to try it as a channeler too, I think I've figured out how to do it.
Of course, it is faster to find cash in temporary channels.
But it's a pity about the NS - it's such a long way to go... Uh-oh... Yeah...
In temporal channels, of course, the cash is found faster.
But it's a pity about the NS - it's such a long way to go... Uh-oh... Well...
No, no, NS will stay, how could it be without it :)
I have an idea, but I can't figure out how to implement it...
1) We have a certain segment with market data (ODS)
2) We have certain market parameters, like width, height, speed, etc.
We have an indicator, let it be two moving averages
4) There is an ideal equity, built on (ORD), those "ideal deals" were made, on which "ideal equity" was built.
5) there are deals that are executed at the intersection of averages (classic)
The target is to find such trading that the built equities are maximally similar to the ideal equities from point 4), for example, to measure the correlation between them.
The algorithm consists in correcting average periods of each candle usingmarket parameters from Sec. 2) in order to get the trade closest to the ideal requity.
I apologize for not very clear statements, I always have a problem with it,
To put it simply:
On each candlestick, depending on the market parameters (point 2) we are looking for such periods of averages (points 3, 5) in order to get maximum close to the ideal equity in the entire trading area (point 4)
What are your ideas how this should be implemented?