Comparison of two quotation charts with non-linear distortions on the X-axis - page 2
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
This is a particular case of finding a master/delinquent relationship. It is solved through an appropriate transformation of the TSP. And then applying the usual linear methods.
P.S. The applicability of pattern theory should still be justified.
OK, about pattern theory. It is often the case that the world situation does not change for several days in a row. The Japanese are buying the Euro, the Europeans are buying the Pound, etc. The intraday chart has different forces at different times of the day. It is not continuous like a matfunction, but like a mosaic. For example, the Eur rises in the morning, falls sharply in the afternoon, recovers a little in the evening, then flies for the rest of the day. It would be possible to divide the day into sessions and analyze each piece independently, but nothing has been achieved. The trend starts are shifting - because operations are running 24 hours a day, plus news gets in the way. The search for patterns in wmifor did not help much either. Even the rougher method - analysis of the candlesticks repetition with the same time does not work. But visually the repeatability is there. So I thought...
You need to transform TZVR by moving from the discreteness of astronomical time to the other essence of time - price change.
Look again at the first point I pointed out above.
visually there are more "mind games" to come, have you specifically tried analyzing? counting bars, deviation values ....
I tried to analyze it by ZZ, all I saw was that trends have repeatability, but only the presence of trends, but neither time intervals nor ray lengths of ZZ have definite repeatability, while I investigate ZZ angles - I can clearly say that the rule really works on historical data: traders are more willing to sell than to buy, ZZ angles of lower tops are statistically sharper than those of higher tops
You need to transform TZVR by moving from the discreteness of astronomical time to the other essence of time - price change.
Look again at the first point I pointed out above.
Thanks for the link, took a look. Basically, there are several methods for non-linear transformations. I've experimented with synthetic bars and Renko. I don't think they will help much in this case. For example, a long tail on one of the bars will distort the non-linear graph beyond recognition, after which it cannot be compared to anything else.
I can say unequivocally that the rule really works on historical data: traders are more willing to sell than to buy, the ZZ angles of the lower tops are statistically sharper than the ZZ angles of the upper tops
That simply cannot be true. Apparently your model doesn't account for some factor. The slopes should be symmetrical.
how do you think the corners of ZZ represent a change in price?
Source data are ticks: Bid and Ask prices with corresponding volumes + time of their arrival. Nothing else. What does it have to do with OHLC filter with discreteness based on astronomical time, which was invented long ago, is unclear. It is also strange to make studies of TzVR on the basis of this filter (in particular, by ZZ angles).
I tried to analyze by ZZ, all I saw was that trends have repeatability, but only the presence of trends, but neither time intervals nor ray lengths of ZZ have unambiguous repeatability, while I examine the ZZ angles - I can clearly say that the rule really works on historical data: traders are more willing to sell than to buy, ZZ angles of lower peaks statistically sharper than those of higher peaks
logically - if someone bought, then someone sold, if there are trends, then the amount sold is not equal to the amount bought.....
here's the unloading of ZZ corners for ~10 years on H1, using the formula:segment y = kx+b ---> for ZZ k = (price1-price0)/(bar1-bar0)/Point
k for the downward corners of WP:
k for the corners of the backs upwards:
I can suggest this: enter a non-linear time for one of the graphs, e.g. in the form of a piecewise linear table function, dyn of the segments and their "ramp" parameters. Next, maximize the correlation coefficient of the two graphs using any available numerical method and selecting appropriate segment parameters. It is time-consuming, but it will work.
No need to invent anything.
Use dynamic time warping.
You don't have to make anything up.
Use dynamic time warping.
Thank you! I've read it, it looks promising.