Machine learning in trading: theory, models, practice and algo-trading - page 2655
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
and vertex times in minutes or seconds (milliseconds is better)?
Seconds are stored, but sometimes just the zigzag vertex number is enough as a rough analogue of time.
Milliseconds - it is probably already HFT and the structure of the glass, which seems to be an insurmountable topic at the amateur level of approach.
Your visualisations are the coolest.
thank you
Seconds are stored, but sometimes just the number of the top of the zigzag is enough as a rough analogue of time.
Milliseconds - this is probably already HFT and the structure of the beaker, which seems to be an insurmountable topic at the amateur level of approach.
i.e. ticks are initially pumped up to find out the time of a vertex when forming its own vertex array?
For zigzags this is the way to do it.
It often happens that two vertices belong to one minute bar. Then the time will be uncertain. Which vertex is the first.
i.e. initially when forming its own array of vertices, ticks are pumped up to find out the time of a vertex?
For zigzags, this is the way to do it.
It often happens that two vertices belong to one minute bar. Then the time will be uncertain. Which vertex is the first.
Yes, when I have a chance, I do it by ticks.
But since I use a rather large zigzag (0.1% for EURUSD), I sometimes use OHLC minutes for speed - of course there are differences from ticks, but not too dramatic.
Cycle by history, in which a cycle by pattern length, etc. is nested - cycle by pattern type or by breakpoints
Alexey, what if you prepare intermediate data in advance, so that you don't have to loop when a pattern appears? That is, in an array of structures in a loop through history to keep constantly necessary data, which will allow you to get the answer at the right moment without a nested loop. As a result, a flying Expert Advisor, 2-3 seconds per run on an hour-long history of 20 years. Nah?
Alexey, what if we prepare intermediate data in advance, so that we don't have to loop when a pattern appears? That is, in an array of structures in a loop through history, keep constantly necessary data that will allow you to get an answer at the right moment without a nested loop. As a result, a flying Expert Advisor, 2-3 seconds per run on an hour-long history of 20 years. Nah?
So we are talking about searching and researching patterns (even more specifically using the R language). In a working Expert Advisor, of course, there are no cycles, everything is extremely simple and nothing superfluous at all.
A cycle by history, in which a cycle by pattern length, etc. is nested - a cycle by pattern type or by breakpoints, for example. If something seems to be there, then a more detailed study is in order.
Artifacts of the history of quotes are found very well)
Why nested cycles and not sequential ones?
Why nested loops and not sequential loops?
We traverse the entire history in a cycle. For each of its points, we look to see if pattern #X has been realised on a prehistory of length N, so there are two more levels of nested cycles - N and X).
And this is the bare minimum) Mining patterns by bruteforce is harsh and merciless, but simple and universal).
We traverse the entire history in the cycle. For each of its points, we look to see if pattern #X has been realised on a prehistory of length N, respectively, this is two more levels of nesting cycles - by N and X).
And this is the bare minimum) Mining patterns by bruteforce is harsh and merciless, but simple and universal).
Ah, well, so yes, nested by window size, well, and there you can also loop by types of patterns or something else.